1 2018-01-02 Andrew Waterman <andrew@sifive.com>
3 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
4 * config/riscv/riscv.md (clear_cache): Use it.
6 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
8 * web.c: Remove out-of-date comment.
10 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
12 * expr.c (fixup_args_size_notes): Check that any existing
13 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
14 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
15 (emit_single_push_insn): ...here.
17 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
20 (const_vector_encoded_nelts): New function.
21 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
22 (const_vector_int_elt, const_vector_elt): Declare.
23 * emit-rtl.c (const_vector_int_elt_1): New function.
24 (const_vector_elt): Likewise.
25 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
28 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30 * expr.c: Include rtx-vector-builder.h.
31 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
32 directly on the tree encoding.
33 (const_vector_from_tree): Likewise.
34 * optabs.c: Include rtx-vector-builder.h.
35 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
36 sequence of "u" values.
37 * vec-perm-indices.c: Include rtx-vector-builder.h.
38 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
39 directly on the vec_perm_indices encoding.
41 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
43 * doc/rtl.texi (const_vector): Describe new encoding scheme.
44 * Makefile.in (OBJS): Add rtx-vector-builder.o.
45 * rtx-vector-builder.h: New file.
46 * rtx-vector-builder.c: Likewise.
47 * rtl.h (rtx_def::u2): Add a const_vector field.
48 (CONST_VECTOR_NPATTERNS): New macro.
49 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
50 (CONST_VECTOR_DUPLICATE_P): Likewise.
51 (CONST_VECTOR_STEPPED_P): Likewise.
52 (CONST_VECTOR_ENCODED_ELT): Likewise.
53 (const_vec_duplicate_p): Check for a duplicated vector encoding.
54 (unwrap_const_vec_duplicate): Likewise.
55 (const_vec_series_p): Check for a non-duplicated vector encoding.
56 Say that the function only returns true for integer vectors.
57 * emit-rtl.c: Include rtx-vector-builder.h.
58 (gen_const_vec_duplicate_1): Delete.
59 (gen_const_vector): Call gen_const_vec_duplicate instead of
60 gen_const_vec_duplicate_1.
61 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
62 (gen_const_vec_duplicate): Use rtx_vector_builder.
63 (gen_const_vec_series): Likewise.
64 (gen_rtx_CONST_VECTOR): Likewise.
65 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
66 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
67 Build a new vector rather than modifying a CONST_VECTOR in-place.
68 (handle_special_swappables): Update call accordingly.
69 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
70 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
71 Build a new vector rather than modifying a CONST_VECTOR in-place.
72 (handle_special_swappables): Update call accordingly.
74 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
76 * simplify-rtx.c (simplify_const_binary_operation): Use
77 CONST_VECTOR_ELT instead of XVECEXP.
79 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
81 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
82 the selector elements to be different from the data elements
83 if the selector is a VECTOR_CST.
84 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
85 ssizetype for the selector.
87 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
89 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
90 before testing each element individually.
91 * tree-vect-generic.c (lower_vec_perm): Likewise.
93 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
95 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
96 * selftest-run-tests.c (selftest::run_tests): Call it.
97 * vector-builder.h (vector_builder::operator ==): New function.
98 (vector_builder::operator !=): Likewise.
99 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
100 (vec_perm_indices::all_from_input_p): New function.
101 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
102 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
103 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
104 instead of reading the VECTOR_CST directly. Detect whether both
105 vector inputs are the same before constructing the vec_perm_indices,
106 and update the number of inputs argument accordingly. Use the
107 utility functions added above. Only construct sel2 if we need to.
109 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
111 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
112 the broadcast of the low byte.
113 (expand_mult_highpart): Use an explicit encoding for the permutes.
114 * optabs-query.c (can_mult_highpart_p): Likewise.
115 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
116 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
117 (vectorizable_bswap): Likewise.
118 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
119 explicit encoding for the power-of-2 permutes.
120 (vect_permute_store_chain): Likewise.
121 (vect_grouped_load_supported): Likewise.
122 (vect_permute_load_chain): Likewise.
124 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
126 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
127 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
128 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
129 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
130 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
131 (vect_gen_perm_mask_any): Likewise.
133 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
135 * int-vector-builder.h: New file.
136 * vec-perm-indices.h: Include int-vector-builder.h.
137 (vec_perm_indices): Redefine as an int_vector_builder.
138 (auto_vec_perm_indices): Delete.
139 (vec_perm_builder): Redefine as a stand-alone class.
140 (vec_perm_indices::vec_perm_indices): New function.
141 (vec_perm_indices::clamp): Likewise.
142 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
143 (vec_perm_indices::new_vector): New function.
144 (vec_perm_indices::new_expanded_vector): Update for new
145 vec_perm_indices class.
146 (vec_perm_indices::rotate_inputs): New function.
147 (vec_perm_indices::all_in_range_p): Operate directly on the
148 encoded form, without computing elided elements.
149 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
150 encoding. Update for new vec_perm_indices class.
151 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
152 the given vec_perm_builder.
153 (expand_vec_perm_var): Update vec_perm_builder constructor.
154 (expand_mult_highpart): Use vec_perm_builder instead of
155 auto_vec_perm_indices.
156 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
157 vec_perm_indices instead of auto_vec_perm_indices. Use a single
158 or double series encoding as appropriate.
159 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
160 vec_perm_indices instead of auto_vec_perm_indices.
161 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
162 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
163 (vect_permute_store_chain): Likewise.
164 (vect_grouped_load_supported): Likewise.
165 (vect_permute_load_chain): Likewise.
166 (vect_shift_permute_load_chain): Likewise.
167 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
168 (vect_transform_slp_perm_load): Likewise.
169 (vect_schedule_slp_instance): Likewise.
170 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
171 (vectorizable_mask_load_store): Likewise.
172 (vectorizable_bswap): Likewise.
173 (vectorizable_store): Likewise.
174 (vectorizable_load): Likewise.
175 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
176 vec_perm_indices instead of auto_vec_perm_indices. Use
177 tree_to_vec_perm_builder to read the vector from a tree.
178 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
179 vec_perm_builder instead of a vec_perm_indices.
180 (have_whole_vector_shift): Use vec_perm_builder and
181 vec_perm_indices instead of auto_vec_perm_indices. Leave the
182 truncation to calc_vec_perm_mask_for_shift.
183 (vect_create_epilog_for_reduction): Likewise.
184 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
185 from auto_vec_perm_indices to vec_perm_indices.
186 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
187 instead of changing individual elements.
188 (aarch64_vectorize_vec_perm_const): Use new_vector to install
189 the vector in d.perm.
190 * config/arm/arm.c (expand_vec_perm_d::perm): Change
191 from auto_vec_perm_indices to vec_perm_indices.
192 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
193 instead of changing individual elements.
194 (arm_vectorize_vec_perm_const): Use new_vector to install
195 the vector in d.perm.
196 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
197 Update vec_perm_builder constructor.
198 (rs6000_expand_interleave): Likewise.
199 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
200 (rs6000_expand_interleave): Likewise.
202 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
204 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
205 to qimode could truncate the indices.
206 * optabs.c (expand_vec_perm_var): Likewise.
208 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
210 * Makefile.in (OBJS): Add vec-perm-indices.o.
211 * vec-perm-indices.h: New file.
212 * vec-perm-indices.c: Likewise.
213 * target.h (vec_perm_indices): Replace with a forward class
215 (auto_vec_perm_indices): Move to vec-perm-indices.h.
216 * optabs.h: Include vec-perm-indices.h.
217 (expand_vec_perm): Delete.
218 (selector_fits_mode_p, expand_vec_perm_var): Declare.
219 (expand_vec_perm_const): Declare.
220 * target.def (vec_perm_const_ok): Replace with...
221 (vec_perm_const): ...this new hook.
222 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
223 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
224 * doc/tm.texi: Regenerate.
225 * optabs.def (vec_perm_const): Delete.
226 * doc/md.texi (vec_perm_const): Likewise.
227 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
228 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
229 expand_vec_perm for constant permutation vectors. Assert that
230 the mode of variable permutation vectors is the integer equivalent
231 of the mode that is being permuted.
232 * optabs-query.h (selector_fits_mode_p): Declare.
233 * optabs-query.c: Include vec-perm-indices.h.
234 (selector_fits_mode_p): New function.
235 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
236 is defined, instead of checking whether the vec_perm_const_optab
237 exists. Use targetm.vectorize.vec_perm_const instead of
238 targetm.vectorize.vec_perm_const_ok. Check whether the indices
239 fit in the vector mode before using a variable permute.
240 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
241 vec_perm_indices instead of an rtx.
242 (expand_vec_perm): Replace with...
243 (expand_vec_perm_const): ...this new function. Take the selector
244 as a vec_perm_indices rather than an rtx. Also take the mode of
245 the selector. Update call to shift_amt_for_vec_perm_mask.
246 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
247 Use vec_perm_indices::new_expanded_vector to expand the original
248 selector into bytes. Check whether the indices fit in the vector
249 mode before using a variable permute.
250 (expand_vec_perm_var): Make global.
251 (expand_mult_highpart): Use expand_vec_perm_const.
252 * fold-const.c: Includes vec-perm-indices.h.
253 * tree-ssa-forwprop.c: Likewise.
254 * tree-vect-data-refs.c: Likewise.
255 * tree-vect-generic.c: Likewise.
256 * tree-vect-loop.c: Likewise.
257 * tree-vect-slp.c: Likewise.
258 * tree-vect-stmts.c: Likewise.
259 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
261 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
262 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
263 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
264 (aarch64_vectorize_vec_perm_const): ...this new function.
265 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
266 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
267 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
268 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
269 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
270 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
271 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
273 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
274 check for NEON modes.
275 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
276 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
277 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
278 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
280 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
281 the old VEC_PERM_CONST conditions.
282 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
283 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
284 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
285 (ia64_vectorize_vec_perm_const_ok): Merge into...
286 (ia64_vectorize_vec_perm_const): ...this new function.
287 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
288 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
289 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
290 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
291 * config/mips/mips.c (mips_expand_vec_perm_const)
292 (mips_vectorize_vec_perm_const_ok): Merge into...
293 (mips_vectorize_vec_perm_const): ...this new function.
294 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
295 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
296 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
297 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
298 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
299 (rs6000_expand_vec_perm_const): Delete.
300 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
302 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
303 (altivec_expand_vec_perm_const_le): Take each operand individually.
304 Operate on constant selectors rather than rtxes.
305 (altivec_expand_vec_perm_const): Likewise. Update call to
306 altivec_expand_vec_perm_const_le.
307 (rs6000_expand_vec_perm_const): Delete.
308 (rs6000_vectorize_vec_perm_const_ok): Delete.
309 (rs6000_vectorize_vec_perm_const): New function.
310 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
311 an element count and rtx array.
312 (rs6000_expand_extract_even): Update call accordingly.
313 (rs6000_expand_interleave): Likewise.
314 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
315 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
316 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
317 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
318 (rs6000_expand_vec_perm_const): Delete.
319 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
320 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
321 (altivec_expand_vec_perm_const_le): Take each operand individually.
322 Operate on constant selectors rather than rtxes.
323 (altivec_expand_vec_perm_const): Likewise. Update call to
324 altivec_expand_vec_perm_const_le.
325 (rs6000_expand_vec_perm_const): Delete.
326 (rs6000_vectorize_vec_perm_const_ok): Delete.
327 (rs6000_vectorize_vec_perm_const): New function. Remove stray
328 reference to the SPE evmerge intructions.
329 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
330 an element count and rtx array.
331 (rs6000_expand_extract_even): Update call accordingly.
332 (rs6000_expand_interleave): Likewise.
333 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
334 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
336 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
338 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
340 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
341 vector mode and that that mode matches the mode of the data
343 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
344 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
345 directly using expand_vec_perm_1 when forcing selectors into
347 (expand_vec_perm_var): New function, split out from expand_vec_perm.
349 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
351 * optabs-query.h (can_vec_perm_p): Delete.
352 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
353 * optabs-query.c (can_vec_perm_p): Split into...
354 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
355 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
356 particular selector is valid.
357 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
358 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
359 (vect_grouped_load_supported): Likewise.
360 (vect_shift_permute_load_chain): Likewise.
361 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
362 (vect_transform_slp_perm_load): Likewise.
363 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
364 (vectorizable_bswap): Likewise.
365 (vect_gen_perm_mask_checked): Likewise.
366 * fold-const.c (fold_ternary_loc): Likewise. Don't take
367 implementations of variable permutation vectors into account
368 when deciding which selector to use.
369 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
370 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
371 with a false third argument.
372 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
373 to test whether the constant selector is valid and can_vec_perm_var_p
374 to test whether a variable selector is valid.
376 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
378 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
379 * optabs-query.c (can_vec_perm_p): Likewise.
380 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
381 instead of vec_perm_indices.
382 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
383 (vect_gen_perm_mask_checked): Likewise,
384 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
385 (vect_gen_perm_mask_checked): Likewise,
387 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
389 * optabs-query.h (qimode_for_vec_perm): Declare.
390 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
391 (qimode_for_vec_perm): ...this new function.
392 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
394 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
396 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
397 does not have a conditional at the top.
399 2018-01-02 Richard Biener <rguenther@suse.de>
401 * ipa-inline.c (big_speedup_p): Fix expression.
403 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
406 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
409 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
413 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
414 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
415 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
416 cond_taken_branch_cost 3->4.
418 2018-01-01 Jakub Jelinek <jakub@redhat.com>
420 PR tree-optimization/83581
421 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
422 TODO_cleanup_cfg if any changes have been made.
425 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
426 convert_modes if target mode has the right side, but different mode
430 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
431 last argument when extracting from CONCAT. If either from_real or
432 from_imag is NULL, use expansion through memory. If result is not
433 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
434 the parts directly to inner mode, if even that fails, use expansion
438 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
439 check for bswap in mode rather than HImode and use that in expand_unop
442 2017-12-31 Jakub Jelinek <jakub@redhat.com>
445 * config/i386/i386.c (ix86_attribute_table): Remove excess
446 initializer for "shared" attribute.
448 2017-12-30 Tom de Vries <tom@codesourcery.com>
451 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
453 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
454 offload_funcs entries that no longer have a corresponding cgraph_node.
455 Mark the remaining ones as DECL_PRESERVE_P.
456 (output_lto): Call prune_offload_funcs.
458 2017-12-30 Jakub Jelinek <jakub@redhat.com>
460 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
461 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
462 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
463 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
464 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
465 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
466 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
467 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
468 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
469 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
470 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
471 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
472 vaesenc_<mode>, vpclmulqdq_<mode>,
473 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
475 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
477 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
478 _Float<N> and _Float<N>X variants for rounding built-in
480 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
481 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
482 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
483 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
484 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
485 * builtins.c (mathfn_built_in_2): Likewise.
486 * internal-fn.def (CEIL): Likewise.
488 (NEARBYINT): Likewise.
492 * convert.c (convert_to_integer_1): Likewise.
493 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
494 (integer_valued_real_call_p): Likewise.
495 * fold-const-call.c (fold_const_call_ss): Likewise.
496 * gencfn-macros.c (print_case_cfn): Change CFN and operator
497 printers to take a const char * suffix instead of a bool.
498 (print_define_operator_list): Likewise.
499 (fltall_suffixes): New list of suffixes, that include the
500 traditional suffixes as well as all of the _Float<N> and
502 (main): For _Float<N> and _Float<N>X functions, emit both
503 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
504 has the _Float<N> and _Float<N>X case names or operators. The
505 <name>_ALL variant has both the traditional and the
506 _Float<N>/_Float<N>X case names or operators.
507 * match.pd (COPYSIGN optimizations): Provide optimizations for
508 _Float<N> and _Float<N>X types where possible.
509 (MIN/MAX optimizations): Likewise.
510 (sqrt optimizations): Likewise.
511 (rounding optimizations): Likewise.
513 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
515 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
516 gen_int_mode rather than GEN_INT.
517 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
519 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
521 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
522 instead of CONSTANT_P.
523 (gen_vec_series): Likewise.
524 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
526 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
528 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
531 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
533 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
534 (aarch64_ld1x2<VDC:mode>): Likewise.
535 (aarch64_simd_ld1<mode>_x2): Likewise.
536 (aarch64_simd_ld1<mode>_x2): Likewise.
537 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
538 (vld1_s8_x2): Likewise.
539 (vld1_u16_x2): Likewise.
540 (vld1_s16_x2): Likewise.
541 (vld1_u32_x2): Likewise.
542 (vld1_s32_x2): Likewise.
543 (vld1_u64_x2): Likewise.
544 (vld1_s64_x2): Likewise.
545 (vld1_f16_x2): Likewise.
546 (vld1_f32_x2): Likewise.
547 (vld1_f64_x2): Likewise.
548 (vld1_p8_x2): Likewise.
549 (vld1_p16_x2): Likewise.
550 (vld1_p64_x2): Likewise.
551 (vld1q_u8_x2): Likewise.
552 (vld1q_s8_x2): Likewise.
553 (vld1q_u16_x2): Likewise.
554 (vld1q_s16_x2): Likewise.
555 (vld1q_u32_x2): Likewise.
556 (vld1q_s32_x2): Likewise.
557 (vld1q_u64_x2): Likewise.
558 (vld1q_s64_x2): Likewise.
559 (vld1q_f16_x2): Likewise.
560 (vld1q_f32_x2): Likewise.
561 (vld1q_f64_x2): Likewise.
562 (vld1q_p8_x2): Likewise.
563 (vld1q_p16_x2): Likewise.
564 (vld1q_p64_x2): Likewise.
566 2017-12-27 Martin Liska <mliska@suse.cz>
568 PR tree-optimization/83552
569 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
570 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
572 2017-12-27 Tom de Vries <tom@codesourcery.com>
574 * config/nvptx/nvptx.c (nvptx_option_override): Disable
575 -gstatement-frontiers.
577 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
579 PR rtl-optimization/83513
580 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
581 before priority comparison.
583 2017-12-25 Jakub Jelinek <jakub@redhat.com>
586 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
587 ix86_isa_flags2 to ix86_isa_flags.
588 * config/i386/i386-c.c (ix86_target_macros_internal): Test
589 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
590 isa_flags rather than isa_flags2.
591 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
592 and -mavx512bitalg from isa2_opts to isa_opts.
593 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
594 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
595 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
597 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
598 OPTION_MASK_ISA_AVX512F ored with another option similarly to
599 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
600 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
601 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
602 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
603 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
604 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
605 builtins from bdesc_args2 to bdesc_args section.
606 (__builtin_ia32_compressstoreuqi512_mask,
607 __builtin_ia32_compressstoreuhi512_mask,
608 __builtin_ia32_compressstoreuqi256_mask,
609 __builtin_ia32_expandloadqi512_mask,
610 __builtin_ia32_expandloadqi512_maskz,
611 __builtin_ia32_expandloadhi512_mask,
612 __builtin_ia32_expandloadhi512_maskz,
613 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
614 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
615 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
616 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
617 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
618 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
619 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
620 __builtin_ia32_vpshldv_v32hi_maskz,
621 __builtin_ia32_vpopcountb_v64qi_mask,
622 __builtin_ia32_vpopcountw_v32hi_mask,
623 __builtin_ia32_vpshufbitqmb512_mask,
624 __builtin_ia32_vpshufbitqmb256_mask): Add
625 " | OPTION_MASK_ISA_AVX512BW".
626 (__builtin_ia32_expandloadqi256_mask,
627 __builtin_ia32_expandloadqi256_maskz,
628 __builtin_ia32_vpopcountb_v32qi_mask): Add
629 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
630 (__builtin_ia32_expandloadhi256_mask,
631 __builtin_ia32_expandloadhi256_maskz,
632 __builtin_ia32_expandloadqi128_mask,
633 __builtin_ia32_expandloadqi128_maskz,
634 __builtin_ia32_expandloadhi128_mask,
635 __builtin_ia32_expandloadhi128_maskz,
636 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
637 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
638 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
639 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
640 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
641 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
642 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
643 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
644 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
645 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
646 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
647 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
648 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
649 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
650 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
651 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
652 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
653 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
654 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
655 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
656 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
657 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
658 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
659 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
660 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
661 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
662 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
663 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
664 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
665 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
666 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
667 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
668 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
669 __builtin_ia32_vpopcountw_v8hi_mask): Add
670 " | OPTION_MASK_ISA_AVX512VL".
671 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
672 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
673 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
674 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
675 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
676 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
677 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
678 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
679 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
680 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
681 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
682 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
683 _mm_mask_bitshuffle_epi64_mask): Likewise.
684 * common/config/i386/i386-common.c
685 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
686 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
687 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
688 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
689 OPTION_MASK_ISA_AVX512BITALG_UNSET.
690 (OPTION_MASK_ISA2_AVX512F_UNSET,
691 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
692 (ix86_handle_option): For -mno-general-regs-only, clear from
693 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
694 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
695 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
696 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
697 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
698 for moving from ix86_isa_flags2 to ix86_isa_flags.
700 2017-12-23 Jakub Jelinek <jakub@redhat.com>
703 * fold-const.c (struct contains_label_data): New type.
704 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
705 inside of a SWITCH_BODY seen during the walk.
706 (contains_label_p): Use walk_tree instead of
707 walk_tree_without_duplicates, prepare data for contains_label_1 and
710 2017-12-22 Jakub Jelinek <jakub@redhat.com>
713 * tree-iterator.c (alloc_stmt_list): Start with cleared
714 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
718 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
719 enable avx512f explicitly in #pragma GCC target.
720 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
721 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
722 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
723 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
724 __builtin_ia32_vpdpbusds_v8si_mask,
725 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
726 __builtin_ia32_vpdpbusds_v4si_mask,
727 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
728 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
729 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
730 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
731 __builtin_ia32_vpdpwssds_v8si_mask,
732 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
733 __builtin_ia32_vpdpwssds_v4si_mask,
734 __builtin_ia32_vpdpwssds_v4si_maskz): Use
735 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
736 just OPTION_MASK_ISA_AVX512VNNI.
738 2017-12-22 Martin Jambor <mjambor@suse.cz>
741 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
744 2017-12-22 Julia Koval <julia.koval@intel.com>
745 Sebastian Peryt <sebastian.peryt@intel.com>
747 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
748 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
749 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
750 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
751 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
752 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
753 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
754 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
755 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
756 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
757 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
758 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
759 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
760 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
761 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
762 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
763 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
764 _mm256_maskz_popcnt_epi64): New intrinsics.
765 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
766 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
767 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
768 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
769 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
770 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
771 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
772 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
773 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
774 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
775 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
776 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
777 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
778 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
779 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
780 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
781 __builtin_ia32_vpshufbitqmb256_mask,
782 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
783 * config/i386/i386-c.c (__AVX512BITALG__): New.
784 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
785 (ix86_valid_target_attribute_inner_p): Ditto.
786 (ix86_expand_args_builtin): Handle new types.
787 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
788 * config/i386/i386.opt: Add -mavx512bitalg.
789 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
790 avx512bitalgintrin.h.
791 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
792 (vpopcount<mode><mask_name>): Add more types.
793 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
794 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
796 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
798 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
799 Or in OPTION_MASK_ISA_AVX512F_SET.
800 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
801 OPTION_MASK_ISA_AVX512VNNI_UNSET.
802 (ix86_handle_option): Adjust for
803 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
804 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
806 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
807 isa_flag instead of isa_flag2.
808 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
809 isa_opts2 to isa_opts.
810 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
813 2017-12-22 Mike Stump <mikestump@comcast.net>
814 Eric Botcazou <ebotcazou@adacore.com>
816 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
818 2017-12-21 Jakub Jelinek <jakub@redhat.com>
821 * config/i386/i386.c (ix86_function_arg_boundary): Return
822 PARM_BOUNDARY for TYPE_EMPTY_P types.
825 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
826 if navail is >= dir.len.
828 2017-12-21 Steve Ellcey <sellcey@cavium.com>
830 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
833 2017-12-21 Jakub Jelinek <jakub@redhat.com>
835 PR rtl-optimization/80747
836 PR rtl-optimization/83512
837 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
838 succ edge from ENTRY, copy partition from e->dest to the newly
840 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
841 ENTRY, use BB_PARTITION of its successor block as current_partition.
842 Don't copy partition when splitting succ edge from ENTRY.
844 PR tree-optimization/83523
845 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
846 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
847 (convert_mult_to_fma): Likewise.
849 PR tree-optimization/83521
850 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
851 gimple_build_assign without code on result of
852 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
855 2017-12-21 Andrew Pinski <apinski@cavium.com>
856 Steve Ellcey <sellcey@cavium.com>
858 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
859 multi-arch for ilp32.
861 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
864 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
865 constraints to operand 2.
866 (*ashl<mode>3_mask_1): Ditto.
867 (*<shift_insn><mode>3_mask): Ditto.
868 (*<shift_insn><mode>3_mask_1): Ditto.
869 (*<rotate_insn><mode>3_mask): Ditto.
870 (*<rotate_insn><mode>3_mask_1): Ditto.
872 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
874 * reorg.c (make_return_insns): Reemit each insn with its own location.
876 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
878 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
879 generation for cases where splatting a value is not useful.
880 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
881 across a vec_duplicate and a paradoxical subreg forming a vector
882 mode to a vec_concat.
884 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
886 * combine.c (simplify_set): Do not transform subregs to zero_extends
887 if the destination is not a scalar int mode.
889 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
892 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
893 shared zero if the input has overflowed.
895 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
897 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
898 support for Cortex-A55 and Cortex-A75.
900 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
902 * common/config/arm/arm-common.c (compare_opt_names): Add function
903 comment. Use strcmp instead of manual loop.
905 2017-12-21 Martin Liska <mliska@suse.cz>
907 PR gcov-profile/83509
908 * gcov-dump.c (dump_gcov_file): Do not read info about
909 support_unexecuted_blocks for gcda files.
911 2017-12-21 Jakub Jelinek <jakub@redhat.com>
913 PR rtl-optimization/82973
914 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
915 (valid_for_const_vector_p): ... this.
916 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
917 (valid_for_const_vector_p): ... this. Adjust function comment.
918 (gen_vec_duplicate): Adjust caller.
919 * optabs.c (expand_vector_broadcast): Likewise.
920 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
921 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
923 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
924 instead of GET_CODE == CONST_FIXED.
925 (simplify_subreg): Use CONST_FIXED_P macro instead of
926 GET_CODE == CONST_FIXED.
929 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
930 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
931 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
932 (ix86_option_override_internal): Adjust for
933 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
934 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
935 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
936 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
937 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
938 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
939 def_builtin2 for CET builtins.
940 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
941 up formatting in IX86_BUILTIN_RDPID code.
942 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
943 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
944 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
945 ix86_isa_flags2 to ix86_isa_flags.
946 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
948 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
949 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
950 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
952 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
953 Or in OPTION_MASK_ISA_AVX512F_SET.
954 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
955 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
956 (ix86_handle_option): Adjust for
957 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
958 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
960 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
961 Alan Hayward <alan.hayward@arm.com>
962 David Sherwood <david.sherwood@arm.com>
964 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
965 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
966 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
967 Track polynomial offsets.
969 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
970 Alan Hayward <alan.hayward@arm.com>
971 David Sherwood <david.sherwood@arm.com>
973 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
974 Treat drb->init as a poly_int. Fail if its misalignment wrt
975 vector_alignment isn't known.
977 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
978 Alan Hayward <alan.hayward@arm.com>
979 David Sherwood <david.sherwood@arm.com>
981 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
983 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
984 (vect_loop_versioning): Take the loop versioning threshold as a
986 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
987 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
988 versioning_threshold.
989 (vect_analyze_loop_2): Compute the loop versioning threshold
990 whenever loop versioning is needed, and store it in the new
991 field rather than combining it with the cost model threshold.
992 (vect_transform_loop): Update call to vect_loop_versioning.
993 Try to combine the loop versioning and cost thresholds here.
995 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
996 Alan Hayward <alan.hayward@arm.com>
997 David Sherwood <david.sherwood@arm.com>
999 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
1000 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
1001 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
1002 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
1003 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
1004 (group_compare_offset): Update accordingly.
1005 (split_small_address_groups_p): Likewise.
1006 (record_use): Take addr_offset as a poly_uint64 rather than
1007 an unsigned HOST_WIDE_INT.
1008 (strip_offset): Return the offset as a poly_uint64 rather than
1009 an unsigned HOST_WIDE_INT.
1010 (record_group_use, split_address_groups): Track polynomial offsets.
1011 (add_iv_candidate_for_use): Likewise.
1012 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1013 than a HOST_WIDE_INT.
1014 (strip_offset_1): Return the offset as a poly_int64 rather than
1017 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1018 Alan Hayward <alan.hayward@arm.com>
1019 David Sherwood <david.sherwood@arm.com>
1021 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1022 rather than a HOST_WIDE_INT.
1023 * tree.c (get_binfo_at_offset): Likewise.
1025 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1026 Alan Hayward <alan.hayward@arm.com>
1027 David Sherwood <david.sherwood@arm.com>
1029 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1030 rather than a HOST_WIDE_INT.
1031 * tree-sra.c (build_ref_for_offset): Likewise.
1033 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1034 Alan Hayward <alan.hayward@arm.com>
1035 David Sherwood <david.sherwood@arm.com>
1037 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1039 * tree.c (mem_ref_offset): Likewise.
1040 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1041 * builtins.c (get_object_alignment_2): Likewise.
1042 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1043 * gimple-fold.c (get_base_constructor): Likewise.
1044 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1045 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1047 * ipa-polymorphic-call.c
1048 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1049 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1050 (get_ancestor_addr_info): Likewise.
1051 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1052 * match.pd: Likewise.
1053 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1054 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1055 * tree-eh.c (tree_could_trap_p): Likewise.
1056 * tree-object-size.c (addr_object_size): Likewise.
1057 * tree-ssa-address.c (copy_ref_info): Likewise.
1058 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1059 (indirect_refs_may_alias_p): Likewise.
1060 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1061 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1062 (non_rewritable_mem_ref_base): Likewise.
1063 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1064 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1065 * varasm.c (decode_addr_const): Likewise.
1067 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1068 Alan Hayward <alan.hayward@arm.com>
1069 David Sherwood <david.sherwood@arm.com>
1071 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1072 offsets for MEM_REFs.
1074 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1075 Alan Hayward <alan.hayward@arm.com>
1076 David Sherwood <david.sherwood@arm.com>
1078 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1079 as a poly_uint64 rather than an unsigned int.
1080 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
1082 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1083 Alan Hayward <alan.hayward@arm.com>
1084 David Sherwood <david.sherwood@arm.com>
1086 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1088 (decode_addr_const): Update accordingly.
1090 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1091 Alan Hayward <alan.hayward@arm.com>
1092 David Sherwood <david.sherwood@arm.com>
1094 * tree.h (bit_field_size, bit_field_offset): New functions.
1095 * hsa-gen.c (gen_hsa_addr): Use them.
1096 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1097 (simplify_vector_constructor): Likewise.
1098 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1099 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1100 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1101 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
1104 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1105 Alan Hayward <alan.hayward@arm.com>
1106 David Sherwood <david.sherwood@arm.com>
1108 * expr.h (emit_group_load, emit_group_load_into_temps)
1109 (emit_group_store): Take the size as a poly_int64 rather than an int.
1110 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1111 (emit_group_load_into_temp, emit_group_store): Likewise.
1113 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1114 Alan Hayward <alan.hayward@arm.com>
1115 David Sherwood <david.sherwood@arm.com>
1117 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1118 an unsigned int to a poly_uint64.
1119 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1120 sizes as poly_uint64s rather than unsigned ints.
1121 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1124 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1125 Alan Hayward <alan.hayward@arm.com>
1126 David Sherwood <david.sherwood@arm.com>
1128 * lra-constraints.c (emit_inc): Change inc_amount from an int
1131 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1132 Alan Hayward <alan.hayward@arm.com>
1133 David Sherwood <david.sherwood@arm.com>
1135 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1137 (add_stack_var, stack_var_cmp, partition_stack_vars)
1138 (dump_stack_var_partition): Update accordingly.
1139 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1140 than a HOST_WIDE_INT.
1141 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1142 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1143 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1144 rather than a HOST_WIDE_INT.
1146 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1147 Alan Hayward <alan.hayward@arm.com>
1148 David Sherwood <david.sherwood@arm.com>
1150 * target.def (return_pops_args): Treat both the input and output
1151 sizes as poly_int64s rather than HOST_WIDE_INTS.
1152 * targhooks.h (default_return_pops_args): Update accordingly.
1153 * targhooks.c (default_return_pops_args): Likewise.
1154 * doc/tm.texi: Regenerate.
1155 * emit-rtl.h (incoming_args): Change pops_args, size and
1156 outgoing_args_size from int to poly_int64_pod.
1157 * function.h (expr_status): Change x_pending_stack_adjust and
1158 x_stack_pointer_delta from int to poly_int64.
1159 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1160 (ARGS_SIZE_RTX): Update accordingly.
1161 * calls.c (highest_outgoing_arg_in_use): Change from int to
1163 (stack_usage_watermark, stored_args_watermark): New variables.
1164 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1165 (emit_call_1): Change the stack_size and rounded_stack_size
1166 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1168 (save_fixed_argument_area): Check stack_usage_watermark.
1169 (initialize_argument_information): Change old_pending_adj from
1170 a HOST_WIDE_INT * to a poly_int64_pod *.
1171 (compute_argument_block_size): Return the size as a poly_int64
1173 (finalize_must_preallocate): Track polynomial argument sizes.
1174 (compute_argument_addresses): Likewise.
1175 (internal_arg_pointer_based_exp): Track polynomial offsets.
1176 (mem_overlaps_already_clobbered_arg_p): Rename to...
1177 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1178 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1179 Check stored_args_used_watermark.
1180 (load_register_parameters): Update accordingly.
1181 (check_sibcall_argument_overlap_1): Likewise.
1182 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1183 args size as a poly_int64 rather than an int. Return a bool
1184 indicating whether the optimization was possible and return
1185 the new adjustment by reference.
1186 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1187 Update stored_args_watermark.
1188 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1189 (expand_call): Likewise. Maintain stack_usage_watermark and
1190 stored_args_watermark. Update calls to
1191 combine_pending_stack_adjustment_and_call.
1192 (emit_library_call_value_1): Handle polynomial argument sizes.
1193 Call stack_region_maybe_used_p and mark_stack_region_used.
1194 Maintain stack_usage_watermark.
1195 (store_one_arg): Likewise. Update call to
1196 mem_overlaps_already_clobbered_arg_p.
1197 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1199 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1200 * config/microblaze/microblaze.c (microblaze_function_prologue):
1202 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1203 TARGET_RETURN_POPS_ARGS interface.
1204 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1206 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1207 * config/i386/i386.c (ix86_return_pops_args): Update for new
1208 TARGET_RETURN_POPS_ARGS interface.
1209 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1210 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1211 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1212 TARGET_RETURN_POPS_ARGS interface.
1213 * config/vax/vax.c (vax_return_pops_args): Likewise.
1214 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1215 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1216 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1217 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1218 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1219 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1220 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1221 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1222 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1223 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1224 and x_stack_pointer_delta from int to poly_int64.
1225 * dojump.c (do_pending_stack_adjust): Update accordingly.
1226 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1227 stack_pointer_deltas.
1228 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1229 (pad_to_arg_alignment): Track polynomial offsets.
1230 (assign_parm_find_stack_rtl): Likewise.
1231 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1232 * toplev.c (output_stack_usage): Update reference to
1233 current_function_pushed_stack_size.
1235 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1236 Alan Hayward <alan.hayward@arm.com>
1237 David Sherwood <david.sherwood@arm.com>
1239 * function.c (in_arg_offset, var_offset, dynamic_offset)
1240 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1241 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1242 rather than a HOST_WIDE_INT.
1243 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1244 (instantiate_virtual_regs_in_insn): Likewise.
1246 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1247 Alan Hayward <alan.hayward@arm.com>
1248 David Sherwood <david.sherwood@arm.com>
1250 * rtl.h (get_args_size, add_args_size_note): New functions.
1251 (find_args_size_adjust): Return a poly_int64 rather than a
1253 (fixup_args_size_notes): Likewise. Make the same change to the
1254 end_args_size parameter.
1255 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1256 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1257 * calls.c (emit_call_1): Likewise.
1258 * explow.c (adjust_stack_1): Likewise.
1259 * cfgcleanup.c (old_insns_match_p): Update use of
1260 find_args_size_adjust.
1261 * combine.c (distribute_notes): Track polynomial arg sizes.
1262 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1263 end_true_args_size, beg_delay_args_size and end_delay_args_size
1264 from HOST_WIDE_INT to poly_int64.
1265 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1266 than a HOST_WIDE_INT.
1267 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1268 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1269 polynomial arg sizes.
1270 * emit-rtl.c (try_split): Use get_args_size.
1271 * recog.c (peep2_attempt): Likewise.
1272 * reload1.c (reload_as_needed): Likewise.
1273 * expr.c (find_args_size_adjust): Return the adjustment as a
1274 poly_int64 rather than a HOST_WIDE_INT.
1275 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1276 to a poly_int64 and change the return type in the same way.
1277 (emit_single_push_insn): Track polynomial arg sizes.
1279 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1280 Alan Hayward <alan.hayward@arm.com>
1281 David Sherwood <david.sherwood@arm.com>
1283 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1284 from HOST_WIDE_INT to poly_int64.
1285 * expr.c (push_block, emit_push_insn): Likewise.
1287 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1288 Alan Hayward <alan.hayward@arm.com>
1289 David Sherwood <david.sherwood@arm.com>
1291 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1293 (get_frame_size): Return the size as a poly_int64 rather than a
1295 (frame_offset_overflow): Take the offset as a poly_int64 rather
1296 than a HOST_WIDE_INT.
1297 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1298 (assign_stack_temp): Likewise for the size.
1299 * function.c (get_frame_size): Return a poly_int64 rather than
1301 (frame_offset_overflow): Take the offset as a poly_int64 rather
1302 than a HOST_WIDE_INT.
1303 (try_fit_stack_local): Take the start, length and size as poly_int64s
1304 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1305 rather than a HOST_WIDE_INT.
1306 (add_frame_space): Take the start and end as poly_int64s rather than
1308 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1309 (assign_stack_temp): Likewise for the size.
1310 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1312 (find_temp_slot_from_address): Handle polynomial offsets.
1313 (combine_temp_slots): Likewise.
1314 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1316 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1317 poly_int64 rather than a HOST_WIDE_INT.
1318 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1319 than a HOST_WIDE_INT.
1320 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1321 polynomial frame offsets.
1322 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1323 as a poly_int64 rather than an int.
1324 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1325 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1326 * config/v850/v850.c (compute_frame_size): Likewise.
1327 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1328 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1329 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1330 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1331 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1332 rather than a HOST_WIDE_INT.
1333 * explow.c (get_dynamic_stack_base): Likewise.
1334 * final.c (final_start_function): Use the constant lower bound
1335 of the frame size for -Wframe-larger-than.
1336 * ira.c (do_reload): Adjust for new get_frame_size return type.
1337 * lra.c (lra): Likewise.
1338 * reload1.c (reload): Likewise.
1339 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1340 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1341 * rtlanal.c (get_initial_register_offset): Return the offset as
1342 a poly_int64 rather than a HOST_WIDE_INT.
1344 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1345 Alan Hayward <alan.hayward@arm.com>
1346 David Sherwood <david.sherwood@arm.com>
1348 * reload1.c (elim_table): Change initial_offset, offset and
1349 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1350 (offsets_at): Change the target array's element type from
1351 HOST_WIDE_INT to poly_int64_pod.
1352 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1353 (elimination_costs_in_insn, update_eliminable_offsets)
1354 (verify_initial_elim_offsets, set_offsets_for_label)
1355 (init_eliminable_invariants): Update after above changes.
1357 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1358 Alan Hayward <alan.hayward@arm.com>
1359 David Sherwood <david.sherwood@arm.com>
1361 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1362 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1363 (decomposition): Change start and end from HOST_WIDE_INT
1365 (decompose, immune_p): Update accordingly.
1366 (find_inc_amount): Return a poly_int64 rather than an int.
1367 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1370 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1371 Alan Hayward <alan.hayward@arm.com>
1372 David Sherwood <david.sherwood@arm.com>
1374 * tree.h (get_inner_reference): Return the bitsize and bitpos
1375 as poly_int64_pods rather than HOST_WIDE_INT.
1376 * fold-const.h (ptr_difference_const): Return the pointer difference
1377 as a poly_int64_pod rather than a HOST_WIDE_INT.
1378 * expr.c (get_inner_reference): Return the bitsize and bitpos
1379 as poly_int64_pods rather than HOST_WIDE_INT.
1380 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1382 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1383 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1384 (optimize_bit_field_compare): Update call to get_inner_reference.
1385 (decode_field_reference): Likewise.
1386 (fold_unary_loc): Track polynomial offsets and sizes.
1387 (split_address_to_core_and_offset): Return the bitpos as a
1388 poly_int64_pod rather than a HOST_WIDE_INT.
1389 (ptr_difference_const): Likewise for the pointer difference.
1390 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1391 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1392 * dbxout.c (dbxout_expand_expr): Likewise.
1393 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1394 (loc_list_from_tree_1, fortran_common): Likewise.
1395 * gimple-laddress.c (pass_laddress::execute): Likewise.
1396 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1397 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1398 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1399 * tree-affine.c (tree_to_aff_combination): Likewise.
1400 (get_inner_reference_aff): Likewise.
1401 * tree-data-ref.c (split_constant_offset_1): Likewise.
1402 (dr_analyze_innermost): Likewise.
1403 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1404 * tree-sra.c (ipa_sra_check_caller): Likewise.
1405 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1406 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1407 (instrument_bool_enum_load, instrument_object_size): Likewise.
1408 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1409 to get_inner_reference.
1410 * hsa-gen.c (gen_hsa_addr): Likewise.
1411 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1412 * tsan.c (instrument_expr): Likewise.
1413 * match.pd: Update call to ptr_difference_const.
1415 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1416 Alan Hayward <alan.hayward@arm.com>
1417 David Sherwood <david.sherwood@arm.com>
1419 * fold-const.c (fold_comparison): Track sizes and offsets as
1420 poly_int64s rather than HOST_WIDE_INTs when folding address
1423 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1424 Alan Hayward <alan.hayward@arm.com>
1425 David Sherwood <david.sherwood@arm.com>
1427 * expr.h (get_bit_range): Return the bitstart and bitend as
1428 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1429 as a poly_int64 rather than a HOST_WIDE_INT.
1430 * expr.c (get_bit_range): Likewise.
1431 (expand_assignment): Update call accordingly.
1432 * fold-const.c (optimize_bit_field_compare): Likewise.
1434 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1435 Alan Hayward <alan.hayward@arm.com>
1436 David Sherwood <david.sherwood@arm.com>
1438 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1439 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1440 as an optional final parameter.
1441 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1442 (aarch64_legitimate_address_p): Likewise.
1443 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1444 instead of an rtx code.
1445 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1446 (aarch64_legitimate_address_hook_p): Likewise.
1447 (aarch64_print_ldpstp_address): Likewise.
1448 (aarch64_print_operand_address): Likewise.
1449 (aarch64_address_cost): Likewise.
1450 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1451 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1452 (aarch64_mem_pair_lanes_operand): Likewise.
1454 2017-12-20 Richard Biener <rguenther@suse.de>
1456 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1457 update_stmt_if_modified.
1459 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1461 PR tree-optimization/83491
1462 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1463 before walking uses. Improve coding style and comments.
1465 2017-12-20 Tom de Vries <tom@codesourcery.com>
1467 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1469 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1472 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1474 (pass_ipa_free_fn_summary): Add small_p private data member,
1475 initialize to false in the ctor.
1476 (pass_ipa_free_fn_summary::clone,
1477 pass_ipa_free_fn_summary::set_pass_param,
1478 pass_ipa_free_fn_summary::gate): New methods.
1479 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1480 | TODO_dump_symtab if small_p.
1481 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1482 entry and add another instance of the pass with false parm after
1484 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1485 ipa_free_fn_summary here.
1487 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1489 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1491 2017-12-20 Martin Sebor <msebor@redhat.com>
1494 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1495 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1496 (expand_builtin_strncmp): Same.
1498 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1501 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1502 allowed debug stmts before labels.
1503 (expand_gimple_basic_block): Likewise.
1504 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1505 * gimple-iterator.h (gsi_after_labels): Likewise.
1506 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1507 rename reused variable, and simplify using gsi_move_before.
1508 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1509 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1510 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1511 (gimple_verify_flow_info, gimple_block_label): Likewise.
1512 (make_blocks): Move debug markers after adjacent labels.
1513 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1514 allowed debug insns outside blocks.
1515 * df-scan.c (df_insn_delete): Likewise.
1516 * lra-constraints.c (update_ebb_live_info): Likewise.
1517 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1518 (vt_initialize, delete_vta_debug_insns): Likewise.
1519 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1521 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1522 Alan Hayward <alan.hayward@arm.com>
1523 David Sherwood <david.sherwood@arm.com>
1525 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1526 (round_up_to_byte_boundary): Likewise.
1527 * expr.h (get_bit_range): Add temporary shim.
1528 * gimple-ssa-store-merging.c (store_operand_info): Change the
1529 bitsize, bitpos, bitregion_start and bitregion_end fields from
1530 unsigned HOST_WIDE_INT to poly_uint64.
1531 (merged_store_group): Likewise load_align_base.
1532 (compatible_load_p, compatible_load_p): Update accordingly.
1533 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1534 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1535 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1536 bitregion_start and bitregion_end as poly_uint64s rather than
1537 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1538 (handled_load): Take the bitsize, bitpos,
1539 bitregion_start and bitregion_end as poly_uint64s rather than
1540 unsigned HOST_WIDE_INTs.
1541 (pass_store_merging::process_store): Update call to
1542 mem_valid_for_store_merging.
1544 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1545 Alan Hayward <alan.hayward@arm.com>
1546 David Sherwood <david.sherwood@arm.com>
1548 * builtins.c (get_object_alignment_2): Track polynomial offsets
1549 and sizes. Update the alignment handling.
1551 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1552 Alan Hayward <alan.hayward@arm.com>
1553 David Sherwood <david.sherwood@arm.com>
1555 * tree.h (get_inner_reference): Add a version that returns the
1556 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1557 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1558 the case in which bitpos is not associated with the first byte.
1560 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1561 Alan Hayward <alan.hayward@arm.com>
1562 David Sherwood <david.sherwood@arm.com>
1564 * tree-affine.h (get_inner_reference_aff): Return the size as a
1566 * tree-affine.c (get_inner_reference_aff): Likewise.
1567 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1568 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1570 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1571 Alan Hayward <alan.hayward@arm.com>
1572 David Sherwood <david.sherwood@arm.com>
1574 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1575 HOST_WIDE_INT rather than a poly_int64.
1577 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1578 Alan Hayward <alan.hayward@arm.com>
1579 David Sherwood <david.sherwood@arm.com>
1581 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1582 HOST_WIDE_INT to poly_int64_pod.
1583 (perform_symbolic_merge): Update accordingly.
1584 (bswap_replace): Likewise.
1586 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1587 Alan Hayward <alan.hayward@arm.com>
1588 David Sherwood <david.sherwood@arm.com>
1590 * tree-affine.h (aff_tree::offset): Change from widest_int
1592 (wide_int_ext_for_comb): Delete.
1593 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1594 constants as poly_widest_int rather than widest_int.
1595 (aff_combination_constant_multiple_p): Return the multiplier
1596 as a poly_widest_int.
1597 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1599 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1600 version static and add an overload for poly_widest_int.
1601 (aff_combination_const, aff_combination_add_cst)
1602 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1603 the constants as poly_widest_int rather than widest_int.
1604 (tree_to_aff_combination): Generalize INTEGER_CST case to
1606 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1607 (aff_combination_add_product, aff_combination_mult): Handle
1609 (aff_combination_constant_multiple_p): Return the multiplier
1610 as a poly_widest_int.
1611 * tree-predcom.c (determine_offset): Return the offset as a
1613 (split_data_refs_to_components, suitable_component_p): Update
1615 (valid_initializer_p): Update call to
1616 aff_combination_constant_multiple_p.
1617 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1618 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1619 as a poly_int64 rather than a HOST_WIDE_INT.
1620 (get_address_cost): Handle polynomial offsets.
1621 (iv_elimination_compare_lt): Likewise.
1622 (rewrite_use_nonlinear_expr): Likewise.
1624 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1625 Alan Hayward <alan.hayward@arm.com>
1626 David Sherwood <david.sherwood@arm.com>
1628 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1629 as a poly_int64_pod rather than a HOST_WIDE_INT.
1630 (get_addr_base_and_unit_offset): Likewise.
1631 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1632 (get_addr_base_and_unit_offset): Likewise.
1633 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1634 to poly_int64 in example.
1635 * fold-const.c (fold_binary_loc): Update call to
1636 get_addr_base_and_unit_offset.
1637 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1638 (maybe_canonicalize_mem_ref_addr): Likewise.
1639 (gimple_fold_stmt_to_constant_1): Likewise.
1640 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1642 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1643 * match.pd: Likewise.
1644 * omp-low.c (lower_omp_target): Likewise.
1645 * tree-sra.c (build_ref_for_offset): Likewise.
1646 (build_debug_ref_for_model): Likewise.
1647 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1648 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1649 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1650 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1651 (constant_pointer_difference): Likewise.
1652 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1653 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1654 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1655 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1656 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1657 (set_ssa_val_to): Likewise.
1658 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1659 (maybe_diag_stxncpy_trunc): Likewise.
1660 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1661 * tree.c (build_simple_mem_ref_loc): Likewise.
1662 (array_at_struct_end_p): Likewise.
1664 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1665 Alan Hayward <alan.hayward@arm.com>
1666 David Sherwood <david.sherwood@arm.com>
1668 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1669 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1670 (get_ref_base_and_extent_hwi): Declare.
1671 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1672 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1673 (get_ref_base_and_extent_hwi): New function.
1674 * cfgexpand.c (expand_debug_expr): Update call to
1675 get_ref_base_and_extent.
1676 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1677 * gimple-fold.c (get_base_constructor): Return the offset as a
1678 poly_int64_pod rather than a HOST_WIDE_INT.
1679 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1680 * ipa-polymorphic-call.c
1681 (ipa_polymorphic_call_context::set_by_invariant)
1682 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1683 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1684 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1685 rather than get_ref_base_and_extent.
1686 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1687 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1688 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1690 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1691 call to get_ref_base_and_extent.
1692 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1693 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1694 (stmt_kills_ref_p): Likewise.
1695 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1696 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1698 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1699 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1700 when calling native_encode_expr.
1701 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1702 call to get_ref_base_and_extent.
1703 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1704 get_ref_base_and_extent.
1705 * var-tracking.c (track_expr_p): Likewise.
1707 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1708 Alan Hayward <alan.hayward@arm.com>
1709 David Sherwood <david.sherwood@arm.com>
1711 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1712 HOST_WIDE_INT to poly_int64_pod.
1713 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1714 polynomail parameter offsets.
1716 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1717 Alan Hayward <alan.hayward@arm.com>
1718 David Sherwood <david.sherwood@arm.com>
1720 * gengtype.c (main): Handle poly_int64_pod.
1721 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1722 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1723 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1724 from HOST_WIDE_INT to poly_int64_pod.
1725 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1726 (copy_cfa): New function.
1727 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1728 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1729 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1730 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1731 (initial_return_save): Treat offsets as poly_ints.
1732 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1733 offset is nonconstant.
1734 (reg_save): Take the offset as a poly_int64. Fall back to
1735 DW_CFA_expression for nonconstant offsets.
1736 (queue_reg_save): Take the offset as a poly_int64.
1737 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1739 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1740 Alan Hayward <alan.hayward@arm.com>
1741 David Sherwood <david.sherwood@arm.com>
1743 * rtl.h (operand_subword, operand_subword_force): Take the offset
1744 as a poly_uint64 an unsigned int.
1745 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1747 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1748 Alan Hayward <alan.hayward@arm.com>
1749 David Sherwood <david.sherwood@arm.com>
1751 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1752 'p' format code. Use INT_LIST rather than SUBREG as the example of
1753 a code with an XINT and an XEXP. Remove the implication that
1754 accessing an rtx field using XINT is expected to work.
1755 * rtl.def (SUBREG): Change format from "ei" to "ep".
1756 * rtl.h (rtunion::rt_subreg): New field.
1757 (XCSUBREG): New macro.
1758 (SUBREG_BYTE): Use it.
1759 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1760 Update constructor accordingly.
1761 (subreg_shape::operator ==): Update accordingly.
1762 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1763 than an unsigned int.
1764 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1765 a poly_uint64 rather than an unsigned int.
1766 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1767 than an unsigned int.
1768 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1769 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1770 an unsigned int. Take the sizes as poly_uint64s.
1771 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1772 an unsigned int. Take the shift as a poly_uint64 rather than
1774 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1775 as a poly_uint64 rather than an unsigned int.
1776 (simplify_subreg_regno): Likewise.
1777 (byte_lowpart_offset): Return the memory offset as a poly_int64
1779 (subreg_memory_offset): Likewise. Take the subreg offset as a
1780 poly_uint64 rather than an unsigned int.
1781 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1782 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1783 poly_uint64 rather than an unsigned int.
1784 * rtl.c (rtx_format): Describe 'p' in comment.
1785 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1786 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1787 offset as a poly_uint64 rather than an unsigned int.
1788 (byte_lowpart_offset): Return the memory offset as a poly_int64
1790 (subreg_memory_offset): Likewise. Take the subreg offset as a
1791 poly_uint64 rather than an unsigned int.
1792 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1793 mode sizes as poly_uint64s rather than unsigned ints. Return a
1794 poly_uint64 rather than an unsigned int.
1795 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1796 (copy_insn_1): Handle 'p'.
1797 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1798 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
1799 an unsigned int. Return the shift in the same way.
1800 (subreg_lsb): Return the shift as a poly_uint64 rather than an
1802 (subreg_size_offset_from_lsb): Take the sizes and shift as
1803 poly_uint64s rather than unsigned ints. Return the offset as
1805 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
1806 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
1808 * rtlhash.c (add_rtx): Handle 'p'.
1809 * genemit.c (gen_exp): Likewise.
1810 * gengenrtl.c (type_from_format, gendef): Likewise.
1811 * gensupport.c (subst_pattern_match, get_alternatives_number)
1812 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
1813 (subst_dup): Likewise.
1814 * gengtype.c (adjust_field_rtx_def): Likewise.
1815 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
1816 (match_pattern_2): Likewise.
1817 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
1818 (rtx_test::subreg_field): New function.
1819 (operator ==, safe_to_hoist_p, transition_parameter_type)
1820 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1821 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1822 * genpeep.c (match_rtx): Likewise.
1823 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1824 (rtx_writer::print_rtx_operand): Handle 'p'.
1825 (print_value): Handle SUBREG.
1826 * read-rtl.c (apply_int_iterator): Likewise.
1827 (rtx_reader::read_rtx_operand): Handle 'p'.
1828 * alias.c (rtx_equal_for_memref_p): Likewise.
1829 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1830 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1832 * calls.c (expand_call): Likewise.
1833 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1834 (make_extraction, gen_lowpart_for_combine): Likewise.
1835 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1837 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1838 rather than an unsigned int. Treat subreg offsets as poly_ints.
1839 (exp_equiv_p): Handle 'p'.
1840 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1841 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1842 * dse.c (find_shift_sequence): Likewise.
1843 * dwarf2out.c (rtl_for_decl_location): Likewise.
1844 * expmed.c (extract_low_bits): Likewise.
1845 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1846 (expand_expr_real_2): Likewise.
1847 * final.c (alter_subreg): Likewise.
1848 (leaf_renumber_regs_insn): Handle 'p'.
1849 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1850 Treat subreg offsets as poly_ints.
1851 * fwprop.c (forward_propagate_and_simplify): Likewise.
1852 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
1853 * ira.c (get_subreg_tracking_sizes): Likewise.
1854 * ira-conflicts.c (go_through_subreg): Likewise.
1855 * ira-lives.c (process_single_reg_class_operands): Likewise.
1856 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1857 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1858 as a poly_uint64 rather than an unsigned int.
1859 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1860 subreg offsets as poly_ints.
1861 * lra-constraints.c (operands_match_p): Handle 'p'.
1862 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1863 * lra-spills.c (assign_mem_slot): Likewise.
1864 * postreload.c (move2add_valid_value_p): Likewise.
1865 * recog.c (general_operand, indirect_operand): Likewise.
1866 * regcprop.c (copy_value, maybe_mode_change): Likewise.
1867 (copyprop_hardreg_forward_1): Likewise.
1868 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
1869 (record_subregs_of_mode): Likewise.
1870 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
1871 * reload.c (operands_match_p): Handle 'p'.
1872 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
1873 * reload1.c (alter_reg, choose_reload_regs): Likewise.
1874 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1875 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1876 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1877 (simplify_const_poly_int_tests<N>::run): Likewise.
1878 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1879 a poly_uint64 rather than an unsigned int.
1880 * valtrack.c (debug_lowpart_subreg): Likewise.
1881 * var-tracking.c (var_lowpart): Likewise.
1882 (loc_cmp): Handle 'p'.
1884 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1885 Alan Hayward <alan.hayward@arm.com>
1886 David Sherwood <david.sherwood@arm.com>
1888 * ira.c (get_subreg_tracking_sizes): New function.
1889 (init_live_subregs): Take an integer size rather than a register.
1890 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1891 to init_live_subregs.
1893 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1894 Alan Hayward <alan.hayward@arm.com>
1895 David Sherwood <david.sherwood@arm.com>
1897 * expr.c (store_constructor_field): Change bitsize from a
1898 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1899 HOST_WIDE_INT to a poly_int64.
1900 (store_constructor): Change size from a HOST_WIDE_INT to
1902 (store_field): Likewise bitsize and bitpos.
1904 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1905 Alan Hayward <alan.hayward@arm.com>
1906 David Sherwood <david.sherwood@arm.com>
1908 * expmed.h (store_bit_field): Change bitregion_start and
1909 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1910 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1911 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1912 (store_fixed_bit_field, store_split_bit_field): Likewise.
1913 * expr.c (store_constructor_field, store_field): Likewise.
1914 (optimize_bitfield_assignment_op): Likewise. Make the same change
1915 to bitsize and bitpos.
1916 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1917 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1918 same change in the constructor arguments.
1919 (get_best_mode): Change bitregion_start and bitregion_end from
1920 unsigned HOST_WIDE_INT to poly_uint64.
1921 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1922 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1924 (bit_field_mode_iterator::next_mode): Update for new types
1925 of m_bitregion_start and m_bitregion_end.
1926 (get_best_mode): Change bitregion_start and bitregion_end from
1927 unsigned HOST_WIDE_INT to poly_uint64.
1929 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1930 Alan Hayward <alan.hayward@arm.com>
1931 David Sherwood <david.sherwood@arm.com>
1933 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1934 accepts poly_uint64 offsets.
1935 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1936 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1937 * expmed.c (lowpart_bit_field_p): Likewise.
1938 (extract_bit_field_as_subreg): New function, split out from...
1939 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1940 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1941 extractions, check that BITSIZE matches the size of the extracted
1942 value and that BITNUM is an exact multiple of that size.
1943 If all else fails, try forcing the value into memory if
1944 BITNUM is variable, and adjusting the address so that the
1945 offset is constant. Split the part that can only handle constant
1946 bitsize and bitnum out into...
1947 (extract_integral_bit_field): ...this new function.
1948 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1949 rather than unsigned HOST_WIDE_INTs.
1951 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1952 Alan Hayward <alan.hayward@arm.com>
1953 David Sherwood <david.sherwood@arm.com>
1955 * expmed.h (store_bit_field): Take bitsize and bitnum as
1956 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1957 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1958 that returns the byte size.
1959 (store_bit_field_1): Take bitsize and bitnum as
1960 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1961 to simple_mem_bitfield_p. Split the part that can only handle
1962 constant bitsize and bitnum out into...
1963 (store_integral_bit_field): ...this new function.
1964 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1965 than unsigned HOST_WIDE_INTs.
1966 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1968 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1969 Alan Hayward <alan.hayward@arm.com>
1970 David Sherwood <david.sherwood@arm.com>
1972 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1973 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1975 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1976 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1977 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1978 offset as a poly_int64 rather than an int.
1979 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1980 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1981 * lra-constraints.c (equiv_address_substitution): Track offsets
1983 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1984 (curr_insn_transform): Handle the new form of sp_offset.
1985 * lra-eliminations.c (lra_elim_table): Change previous_offset
1986 and offset from HOST_WIDE_INT to poly_int64.
1987 (print_elim_table, update_reg_eliminate): Update accordingly.
1988 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1989 (get_elimination): Update accordingly.
1990 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1991 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1992 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1993 poly_int64 offsets generally.
1994 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1995 (mark_not_eliminable, init_elimination): Update accordingly.
1996 (remove_reg_equal_offset_note): Return a bool and pass the new
1997 offset back by pointer as a poly_int64.
1998 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1999 rather than a HOST_WIDE_INT.
2000 (do_remat): Track offsets poly_int64s.
2001 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
2003 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2004 Alan Hayward <alan.hayward@arm.com>
2005 David Sherwood <david.sherwood@arm.com>
2007 * rtl.h (mem_attrs): Add a default constructor. Change size and
2008 offset from HOST_WIDE_INT to poly_int64.
2009 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2010 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2011 (widen_memory_access): Take the sizes and offsets as poly_int64s
2012 rather than HOST_WIDE_INTs.
2013 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2014 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2016 (adjust_offset_for_component_ref): Change the offset from a
2017 HOST_WIDE_INT to a poly_int64.
2018 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2019 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2020 * dce.c (find_call_stack_args): Likewise.
2021 * dse.c (record_store): Likewise.
2022 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2023 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2024 * read-rtl-function.c (test_loading_mem): Likewise.
2025 * rtlanal.c (may_trap_p_1): Likewise.
2026 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2027 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2028 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2029 (mem_attrs::mem_attrs): New function.
2030 (set_mem_attributes_minus_bitpos): Change bitpos from a
2031 HOST_WIDE_INT to poly_int64.
2032 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2033 (clear_mem_offset, clear_mem_size, change_address)
2034 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2035 initialize mem_attrs.
2036 (set_mem_offset, set_mem_size, adjust_address_1)
2037 (adjust_automodify_address_1, offset_address, widen_memory_access):
2038 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2040 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2041 Alan Hayward <alan.hayward@arm.com>
2042 David Sherwood <david.sherwood@arm.com>
2044 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2045 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2046 rather than 0 to represent an unknown size. Assert that the size
2047 is known when the mode isn't BLKmode.
2048 (may_trap_p_1): Use -1 for unknown sizes.
2049 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2051 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2052 Alan Hayward <alan.hayward@arm.com>
2053 David Sherwood <david.sherwood@arm.com>
2055 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2056 to poly_int64. Update commentary for positions_needed.large.
2057 (read_info_type): Change offset and width from HOST_WIDE_INT
2059 (set_usage_bits): Likewise.
2060 (canon_address): Return the offset as a poly_int64 rather than
2061 a HOST_WIDE_INT. Use strip_offset_and_add.
2062 (set_all_positions_unneeded, any_positions_needed_p): Use
2063 positions_needed.large to track stores with non-constant widths.
2064 (all_positions_needed_p): Likewise. Take the offset and width
2065 as poly_int64s rather than ints. Assert that rhs is nonnull.
2066 (record_store): Cope with non-constant offsets and widths.
2067 Nullify the rhs of an earlier store if we can't tell which bytes
2069 (find_shift_sequence): Take the access_size and shift as poly_int64s
2071 (get_stored_val): Take the read_offset and read_width as poly_int64s
2072 rather than HOST_WIDE_INTs.
2073 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2074 non-constant offsets and widths.
2076 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2077 Alan Hayward <alan.hayward@arm.com>
2078 David Sherwood <david.sherwood@arm.com>
2080 * inchash.h (inchash::hash::add_poly_int): New function.
2081 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2082 Use poly_int64 rather than HOST_WIDE_INT.
2083 (ao_ref::max_size_known_p): New function.
2084 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2085 rather than HOST_WIDE_INT.
2086 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2087 to temporaries until its interface is adjusted to match.
2088 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2089 (aliasing_component_refs_p, decl_refs_may_alias_p)
2090 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2091 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2092 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2094 * alias.c (ao_ref_from_mem): Likewise.
2095 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2096 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2097 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2098 (maybe_trim_complex_store, maybe_trim_constructor_store)
2099 (live_bytes_read, dse_classify_store): Likewise.
2100 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2101 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2102 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2103 (vn_reference_lookup_3): Likewise.
2104 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
2106 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2107 Alan Hayward <alan.hayward@arm.com>
2108 David Sherwood <david.sherwood@arm.com>
2110 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2111 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2112 instead of ranges_overlap_p.
2114 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2115 Alan Hayward <alan.hayward@arm.com>
2116 David Sherwood <david.sherwood@arm.com>
2118 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2119 sizes as poly_int64s rather than HOST_WIDE_INTs.
2121 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2122 Alan Hayward <alan.hayward@arm.com>
2123 David Sherwood <david.sherwood@arm.com>
2125 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2126 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2127 * gimple-fold.c (fold_ctor_reference): Likewise.
2129 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2130 Alan Hayward <alan.hayward@arm.com>
2131 David Sherwood <david.sherwood@arm.com>
2133 * target.def (dwarf_poly_indeterminate_value): New hook.
2134 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2135 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2136 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2137 * doc/tm.texi: Regenerate.
2138 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2139 offset as a poly_int64.
2140 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2141 offset as a poly_int64.
2142 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2143 Take the offset as a poly_int64.
2144 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2145 (frame_pointer_fb_offset): Change to a poly_int64.
2146 (int_loc_descriptor): Take the offset as a poly_int64. Use
2147 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2148 (based_loc_descr): Take the offset as a poly_int64.
2149 Use strip_offset_and_add to handle (plus X (const)).
2150 Use new_reg_loc_descr instead of an open-coded version of the
2151 previous implementation.
2152 (mem_loc_descriptor): Handle CONST_POLY_INT.
2153 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2154 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2156 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2157 Alan Hayward <alan.hayward@arm.com>
2158 David Sherwood <david.sherwood@arm.com>
2160 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2161 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2162 * inchash.h (inchash::hash::add_poly_hwi): New function.
2163 * gengtype.c (main): Register poly_int64.
2164 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2165 offset as a poly_int.
2166 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2167 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2168 offset as a poly_int64.
2169 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2170 * print-rtl.c (print_poly_int): New function.
2171 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2173 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2175 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2176 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2177 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2178 Enforce track_offset_p for parts of a PARALLEL.
2179 (vt_add_function_parameter): Use const_offset for the final
2180 offset to track. Use get_tracked_reg_offset for the parts
2183 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2184 Alan Hayward <alan.hayward@arm.com>
2185 David Sherwood <david.sherwood@arm.com>
2187 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2188 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2189 * doc/tm.texi: Regenerate.
2190 * hooks.h (hook_bool_uint_uint_true): Delete.
2191 (hook_bool_puint64_puint64_true): Declare.
2192 * hooks.c (hook_bool_uint_uint_true): Delete.
2193 (hook_bool_puint64_puint64_true): New function.
2194 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2195 instead of unsigned ints.
2196 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2197 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2199 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2200 Alan Hayward <alan.hayward@arm.com>
2201 David Sherwood <david.sherwood@arm.com>
2203 * optabs.h (expand_operand): Add an int_value field.
2204 (create_expand_operand): Add an int_value parameter and use it
2205 to initialize the new expand_operand field.
2206 (create_integer_operand): Replace with a declaration of a function
2207 that accepts poly_int64s. Move the implementation to...
2208 * optabs.c (create_integer_operand): ...here.
2209 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2210 the mode preserves the value of int_value, instead of calling
2211 const_int_operand on the rtx. Use gen_int_mode to generate
2214 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2215 Alan Hayward <alan.hayward@arm.com>
2216 David Sherwood <david.sherwood@arm.com>
2218 * dumpfile.h (dump_dec): Declare.
2219 * dumpfile.c (dump_dec): New function.
2220 * pretty-print.h (pp_wide_integer): Turn into a function and
2221 declare a poly_int version.
2222 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2224 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2225 Alan Hayward <alan.hayward@arm.com>
2226 David Sherwood <david.sherwood@arm.com>
2228 * doc/generic.texi (POLY_INT_CST): Document.
2229 * tree.def (POLY_INT_CST): New tree code.
2230 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2231 * tree-core.h (tree_poly_int_cst): New struct.
2232 (tree_node): Add a poly_int_cst field.
2233 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2234 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2235 instead of a wide_int_ref.
2236 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2238 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2239 instead of an unsigned HOST_WIDE_INT.
2240 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2241 (ptrdiff_tree_p): Declare.
2242 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2243 extern inline implementations if the target doesn't use POLY_INT_CST.
2244 (poly_int_tree_p): New function.
2245 (wi::unextended_tree): New class.
2246 (wi::int_traits <unextended_tree>): New override.
2247 (wi::extended_tree): Add a default constructor.
2248 (wi::extended_tree::get_tree): New function.
2249 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2250 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2251 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2252 (wi::tree_to_poly_wide_ref): New typedefs.
2253 (wi::ints_for): Provide overloads for extended_tree and
2255 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2256 (wi::to_wide): New functions.
2257 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2258 * tree.c (poly_int_cst_hasher): New struct.
2259 (poly_int_cst_hash_table): New variable.
2260 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2261 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2263 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2264 (init_ttree): Initialize poly_int_cst_hash_table.
2265 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2266 a poly_int64 instead of a HOST_WIDE_INT.
2267 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2268 instead of an unsigned HOST_WIDE_INT.
2269 (wide_int_to_tree): Rename to...
2270 (wide_int_to_tree_1): ...this.
2271 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2272 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2273 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2274 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2275 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2276 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2278 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2279 (streamer_read_tree_body): Likewise.
2280 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2281 (streamer_write_tree_body): Likewise.
2282 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2283 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2284 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2285 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2286 * gimple-expr.h (is_gimple_constant): Likewise.
2287 * gimplify.c (maybe_with_size_expr): Likewise.
2288 * print-tree.c (print_node): Likewise.
2289 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2290 * tree-pretty-print.c (dump_generic_node): Likewise.
2291 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2292 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2293 * tree-vrp.c (compare_values_warnv): Likewise.
2294 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2295 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2296 (force_expr_to_var_cost): Likewise.
2297 * tree-ssa-loop.c (for_each_index): Likewise.
2298 * fold-const.h (build_invariant_address, size_int_kind): Take a
2299 poly_int64 instead of a HOST_WIDE_INT.
2300 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2301 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2303 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2304 (int_const_binop_2): New function, split out from...
2305 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2306 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2307 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2308 * gimple-ssa-strength-reduction.c (slsr_process_add)
2309 (slsr_process_mul): Check for INTEGER_CSTs before using them
2311 * stor-layout.c (bits_from_bytes): New function.
2312 (bit_from_pos): Use it.
2313 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2314 by BITS_PER_UNIT to get the TYPE_SIZE.
2315 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2316 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2318 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2319 Alan Hayward <alan.hayward@arm.com>
2320 David Sherwood <david.sherwood@arm.com>
2322 * doc/rtl.texi (const_poly_int): Document. Also document the
2323 rtl sharing behavior.
2324 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2325 * rtl.h (const_poly_int_def): New struct.
2326 (rtx_def::u): Add a cpi field.
2327 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2328 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2329 (wi::rtx_to_poly_wide_ref): New typedef
2330 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2331 (poly_int_rtx_p): New functions.
2332 (trunc_int_for_mode): Declare a poly_int64 version.
2333 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2334 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2336 (strip_offset): Declare.
2337 (strip_offset_and_add): New function.
2338 * rtl.def (CONST_POLY_INT): New rtx code.
2339 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2340 (shared_const_p): Use poly_int_rtx_p.
2341 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2343 (gen_int_shift_amount): Likewise.
2344 * emit-rtl.c (const_poly_int_hasher): New class.
2345 (const_poly_int_htab): New variable.
2346 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2347 (const_poly_int_hasher::hash): New function.
2348 (const_poly_int_hasher::equal): Likewise.
2349 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2350 (immed_wide_int_const): Rename to...
2351 (immed_wide_int_const_1): ...this and make static.
2352 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2353 instead of a wide_int_ref.
2354 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2355 (gen_lowpart_common): Handle CONST_POLY_INT.
2356 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2357 * cselib.c (cselib_hash_rtx): Likewise.
2358 * dwarf2out.c (const_ok_for_output_1): Likewise.
2359 * expr.c (convert_modes): Likewise.
2360 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2361 * rtlhash.c (add_rtx): Likewise.
2362 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2363 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2364 Handle existing CONST_POLY_INT rtxes.
2365 * expmed.h (expand_shift): Take a poly_int64 instead of a
2367 * expmed.c (expand_shift): Likewise.
2368 * rtlanal.c (strip_offset): New function.
2369 (commutative_operand_precedence): Give CONST_POLY_INT the same
2370 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2372 * rtl-tests.c (const_poly_int_tests): New struct.
2373 (rtl_tests_c_tests): Use it.
2374 * simplify-rtx.c (simplify_const_unary_operation): Handle
2376 (simplify_const_binary_operation): Likewise.
2377 (simplify_binary_operation_1): Fold additions of symbolic constants
2378 and CONST_POLY_INTs.
2379 (simplify_subreg): Handle extensions and truncations of
2381 (simplify_const_poly_int_tests): New struct.
2382 (simplify_rtx_c_tests): Use it.
2383 * wide-int.h (storage_ref): Add default constructor.
2384 (wide_int_ref_storage): Likewise.
2385 (trailing_wide_ints): Use GTY((user)).
2386 (trailing_wide_ints::operator[]): Add a const version.
2387 (trailing_wide_ints::get_precision): New function.
2388 (trailing_wide_ints::extra_size): Likewise.
2390 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2391 Alan Hayward <alan.hayward@arm.com>
2392 David Sherwood <david.sherwood@arm.com>
2394 * emit-rtl.h (gen_int_shift_amount): Declare.
2395 * emit-rtl.c (gen_int_shift_amount): New function.
2396 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2398 * calls.c (shift_return_value): Likewise.
2399 * cse.c (fold_rtx): Likewise.
2400 * dse.c (find_shift_sequence): Likewise.
2401 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2402 (expand_shift, expand_smod_pow2): Likewise.
2403 * lower-subreg.c (shift_cost): Likewise.
2404 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2405 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2406 (expand_vec_perm_var): Likewise.
2407 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2408 (simplify_binary_operation_1): Likewise.
2409 * combine.c (try_combine, find_split_point, force_int_to_mode)
2410 (simplify_shift_const_1, simplify_shift_const): Likewise.
2411 (change_zero_ext): Likewise. Use simplify_gen_binary.
2413 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2415 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2417 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2419 * doc/invoke.texi (ARM Options): Document accepted extension options
2420 for -march=armv8.3-a.
2422 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2425 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2426 or --with-float=softfp, set the default CPU to arm10e.
2428 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2430 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2431 * config/visium/predicates.md (const_shift_operand): Likewise.
2432 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2433 (visium_legitimize_reload_address): Likewise.
2435 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2438 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2440 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2443 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2445 2017-12-20 Julia Koval <julia.koval@intel.com>
2447 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2448 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2449 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2450 * config.gcc: Include vpclmulqdqintrin.h.
2451 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2452 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2454 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2455 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2456 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2457 * config/i386/i386.c (isa2_opts): Add -mcx16.
2458 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2459 (ix86_option_override_internal): Move mcx16 to flags2.
2460 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2461 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2462 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2463 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2464 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2465 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2466 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2467 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2468 * doc/invoke.texi: Add -mvpclmulqdq.
2470 2017-12-20 Tom de Vries <tom@codesourcery.com>
2473 * config/i386/i386.c (ix86_static_chain): Move
2474 DECL_STATIC_CHAIN test ...
2475 * calls.c (rtx_for_static_chain): ... here. New function.
2476 * calls.h (rtx_for_static_chain): Declare.
2477 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2478 instead of targetm.calls.static_chain.
2479 * df-scan.c (df_get_entry_block_def_set): Same.
2481 2017-12-19 Tom de Vries <tom@codesourcery.com>
2483 PR tree-optimization/83493
2484 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2487 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2489 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2490 inform with hardcoded english plural handling.
2492 2017-12-18 Jeff Law <law@redhat.com>
2494 PR tree-optimization/83477
2495 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2496 a non-virtual PHI, always push a new range.
2498 2017-12-19 Martin Sebor <msebor@redhat.com>
2501 * builtins.c (compute_objsize): Handle non-constant offsets.
2503 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2505 PR tree-optimization/83444
2506 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2507 character load case, if get_stridx on MEM_REF's operand doesn't
2508 look usable, retry with get_addr_stridx.
2510 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2513 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2514 (variable_tracking_main_1): Keep markers even when VTA fails.
2517 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2518 even if there are markers before it.
2519 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2521 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2523 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2524 typical order conditions.
2525 * sel-sched.c (extract_new_fences_from): Likewise.
2526 * config/visium/constraints.md (J, K, L): Likewise.
2527 * config/visium/predicates.md (const_shift_operand): Likewise.
2528 * config/visium/visium.c (visium_legitimize_address,
2529 visium_legitimize_reload_address): Likewise.
2530 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2531 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2532 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2533 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2534 SET_DUMP_DETAIL): Likewise.
2535 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2536 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2537 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2538 avr_set_core_architecture, avr_set_current_function,
2539 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2540 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2541 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2542 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2543 avr_map_decompose, avr_fold_builtin): Likewise.
2544 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2545 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2546 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2547 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2548 * config/m32c/m32c.c (m32c_conditional_register_usage,
2549 m32c_address_cost): Likewise.
2550 * config/m32c/predicates.md (shiftcount_operand,
2551 longshiftcount_operand): Likewise.
2552 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2553 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2554 can_use_cdx_ldstw): Likewise.
2555 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2556 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2558 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2559 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2560 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2561 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2562 * config/vax/vax.c (adjacent_operands_p): Likewise.
2563 * config/ft32/constraints.md (L, b, KA): Likewise.
2564 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2566 * cfgexpand.c (expand_stack_alignment): Likewise.
2567 * gcse.c (insert_expr_in_table): Likewise.
2568 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2569 * cgraphunit.c (cgraph_node::expand): Likewise.
2570 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2571 * emit-rtl.c (add_insn): Likewise.
2572 * input.c (dump_location_info): Likewise.
2573 * passes.c (NEXT_PASS): Likewise.
2574 * read-rtl-function.c (parse_note_insn_name,
2575 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2577 * sched-rgn.c (sched_rgn_init): Likewise.
2578 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2579 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2580 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2582 * explow.c (eliminate_constant_term): Likewise.
2583 * final.c (leaf_renumber_regs_insn): Likewise.
2584 * cfgrtl.c (print_rtl_with_bb): Likewise.
2585 * genhooks.c (emit_init_macros): Likewise.
2586 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2587 * tree-data-ref.c (conflict_fn): Likewise.
2588 * selftest.c (assert_streq): Likewise.
2589 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2590 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2591 fold_binary_loc, multiple_of_p): Likewise.
2592 * reload.c (push_reload, find_equiv_reg): Likewise.
2593 * et-forest.c (et_nca, et_below): Likewise.
2594 * dbxout.c (dbxout_symbol_location): Likewise.
2595 * reorg.c (relax_delay_slots): Likewise.
2596 * dojump.c (do_compare_rtx_and_jump): Likewise.
2597 * gengtype-parse.c (type): Likewise.
2598 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2599 simplify_const_relational_operation): Likewise.
2600 * reload1.c (do_output_reload): Likewise.
2601 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2602 * gengtype.c (type_for_name): Likewise.
2603 * gimple-ssa-sprintf.c (format_directive): Likewise.
2605 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2608 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2609 accessing it. Adjust comment.
2611 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2614 * predict.c (zero_one_minusone): New function.
2615 (apply_return_prediction): Avoid return prediction for functions
2616 returning only -1, 0 and 1 values, unless they only return -1 and 0
2619 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2621 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2622 fall-through warning.
2623 (arc_compute_frame_size): Remove unused variables.
2624 (arc_print_operand): Fix fprintif format.
2625 (arc_can_follow_jump): Clean fall-through warning.
2627 2017-12-19 Marek Polacek <polacek@redhat.com>
2630 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2633 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2635 * config/arc/arc.c (overriderregs): New variable.
2636 (arc_override_options): Track fixed/call saved/call options.
2637 (arc_conditional_register_usage): Check against overrideregs
2638 variable whenever we change register properties.
2640 2017-12-19 Nathan Sidwell <nathan@acm.org>
2642 * opts.c (finish_options): Don't prefix dump_base_name if it
2643 already contains directories.
2645 2017-12-19 Martin Liska <mliska@suse.cz>
2647 PR rtl-optimization/82675
2648 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2649 more element in sbitmap.
2651 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2653 * gimplify.c (gimplify_expr): Use error_operand_p.
2655 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2658 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2661 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2663 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2664 the jump is to a label.
2666 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2668 PR tree-optimization/83444
2669 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2674 * ipa-inline.c (flatten_remove_node_hook): New function.
2675 (ipa_inline): Keep only nodes with flatten attribute at the end of
2676 the array in the order from ipa_reverse_postorder, only walk that
2677 portion of array for flattening, if there is more than one such
2678 node, temporarily register a removal hook and ignore removed nodes.
2680 PR tree-optimization/80631
2681 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2682 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2683 IFN_REDUC_MAX or IFN_REDUC_MIN.
2685 2017-12-18 Jeff Law <law@redhat.com>
2687 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2688 record symbolic equivalences from backedges in the CFG.
2691 2017-11-19 Jeff Law <law@redhat.com>
2693 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2694 of degenerates resulting from ignoring an edge.
2696 2017-12-18 Martin Sebor <msebor@redhat.com>
2699 PR tree-optimization/78450
2700 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2701 (handle_builtin_strlen): Call it.
2703 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2705 PR rtl-optimization/83424
2706 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2708 2017-12-18 Marek Polacek <polacek@redhat.com>
2711 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2712 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2715 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2717 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2718 (umaddsidi4, umaddsidi_split): Likewise.
2720 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2722 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2725 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2728 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2730 2017-12-18 Richard Biener <rguenther@suse.de>
2732 PR tree-optimization/81877
2733 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2734 (outermost_indep_loop): Adjust.
2735 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2736 (can_sm_ref_p): Adjust.
2738 2017-12-18 Richard Biener <rguenther@suse.de>
2741 * tree.c (array_at_struct_end_p): Return true if the underlying
2742 object has space for at least one element in excess of what
2743 the array domain specifies.
2745 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2747 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2748 example to avoid overfull hbox.
2749 * doc/invoke.texi (Option Summary): Add missing @gol.
2750 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2753 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2754 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2756 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2757 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2758 (AArch64 Options): Document -mverbose-cost-dump.
2759 (ARM Options): Likewise, plus -mflip-thumb.
2761 2017-12-17 Martin Sebor <msebor@redhat.com>
2764 * gimple-ssa-warn-restrict.c
2765 (builtin_memref::offset_out_of_bounds): Correct the handling of
2768 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2770 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2772 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2774 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2776 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2778 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2781 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2783 * doc/invoke.texi: Fix some typos.
2785 2017-12-16 Martin Sebor <msebor@redhat.com>
2787 PR tree-optimization/78918
2788 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2789 * builtins.c (check_sizes): Rename...
2790 (check_access): ...to this. Rename function arguments for clarity.
2791 (check_memop_sizes): Adjust names.
2792 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2793 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2794 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2795 (check_strncat_sizes, expand_builtin_strncat): Same.
2796 (expand_builtin_strncpy, expand_builtin_memset): Same.
2797 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2798 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
2799 (maybe_emit_sprintf_chk_warning): Same.
2800 (expand_builtin_strcpy): Adjust.
2801 (expand_builtin_stpcpy): Same.
2802 (expand_builtin_with_bounds): Detect out-of-bounds accesses
2803 in pointer-checking forms of memcpy, memmove, and mempcpy.
2804 (gcall_to_tree_minimal, max_object_size): Define new functions.
2805 * builtins.h (max_object_size): Declare.
2806 * calls.c (alloc_max_size): Call max_object_size instead of
2807 hardcoding ssizetype limit.
2808 (get_size_range): Handle new argument.
2809 * calls.h (get_size_range): Add a new argument.
2810 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
2811 * doc/invoke.texi (-Wrestrict): Adjust, add example.
2812 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
2814 (gimple_fold_builtin_memory_chk): Same.
2815 (gimple_fold_builtin_stxcpy_chk): New function.
2816 * gimple-ssa-warn-restrict.c: New source.
2817 * gimple-ssa-warn-restrict.h: New header.
2818 * gimple.c (gimple_build_call_from_tree): Propagate location.
2819 * passes.def (pass_warn_restrict): Add new pass.
2820 * tree-pass.h (make_pass_warn_restrict): Declare.
2821 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2823 (handle_builtin_strcat): Same.
2824 (strlen_optimize_stmt): Rename...
2825 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2826 stpncpy, strncpy, and their checking forms.
2828 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2830 PR rtl-optimization/82849
2831 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2832 and get_max_loop_iterations_int.
2834 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2835 Alan Hayward <alan.hayward@arm.com>
2836 David Sherwood <david.sherwood@arm.com>
2838 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2839 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2840 size as a poly_uint64.
2841 (mode_for_vector, mode_for_int_vector): Take the number of vector
2842 elements as a poly_uint64.
2843 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2844 size as a poly_uint64.
2845 (mode_for_vector, mode_for_int_vector): Take the number of vector
2846 elements as a poly_uint64.
2848 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2849 Alan Hayward <alan.hayward@arm.com>
2850 David Sherwood <david.sherwood@arm.com>
2852 * machmode.h (MACRO_MODE): New macro.
2853 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2854 * targhooks.c (default_libcall_value, default_secondary_reload)
2855 (default_memory_move_cost, default_register_move_cost)
2856 (default_class_max_nregs): Likewise.
2858 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2859 Alan Hayward <alan.hayward@arm.com>
2860 David Sherwood <david.sherwood@arm.com>
2862 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2863 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
2865 * genautomata.c (main): Likewise.
2866 * genconditions.c (write_header): Likewise.
2867 * genemit.c (main): Likewise.
2868 * genextract.c (print_header): Likewise.
2869 * genopinit.c (main): Likewise.
2870 * genoutput.c (output_prologue): Likewise.
2871 * genpeep.c (main): Likewise.
2872 * genpreds.c (write_insn_preds_c): Likewise.
2873 * genrecog.c (writer_header): Likewise.
2874 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
2875 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
2876 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2877 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2878 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2879 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2880 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2881 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2882 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2883 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2884 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2885 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2886 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2887 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2888 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2889 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2890 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2891 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2892 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2893 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2894 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2895 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2896 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2897 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2898 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2899 * config/darwin.c (IN_TARGET_CODE): Likewise.
2900 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2901 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2902 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2903 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2904 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2905 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2906 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2907 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2908 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2909 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2910 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2911 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2912 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2913 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2914 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2915 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2916 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2917 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2918 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2919 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2920 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2921 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2922 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2923 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2924 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2925 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2926 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2927 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2928 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2929 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2930 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2931 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2932 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2933 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2934 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2935 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2936 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2937 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2938 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2939 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2940 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2941 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2942 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2943 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2944 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2945 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2946 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2947 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2948 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2949 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2950 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2951 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2952 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2953 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2954 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2955 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2956 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2957 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2958 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2959 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2960 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2961 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2962 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2963 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2964 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2965 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2966 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2967 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2968 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2969 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2970 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2971 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2972 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2973 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2974 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2975 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2976 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2977 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2978 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2979 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2980 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2981 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2982 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2983 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2984 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2985 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2986 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2987 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2988 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2989 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2990 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2991 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2992 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2993 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2994 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2995 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2996 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2997 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2998 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2999 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
3000 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
3001 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
3002 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
3003 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
3004 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
3005 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
3007 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3008 Alan Hayward <alan.hayward@arm.com>
3009 David Sherwood <david.sherwood@arm.com>
3011 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3014 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3015 Alan Hayward <alan.hayward@arm.com>
3016 David Sherwood <david.sherwood@arm.com>
3018 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3019 * doc/md.texi (vec_series@var{m}): Document.
3020 * tree.def (VEC_SERIES_EXPR): New tree code.
3021 * tree.h (build_vec_series): Declare.
3022 * tree.c (build_vec_series): New function.
3023 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3024 * tree-pretty-print.c (dump_generic_node): Likewise.
3025 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3026 * tree-inline.c (estimate_operator_cost): Likewise.
3027 * expr.c (expand_expr_real_2): Likewise.
3028 * optabs-tree.c (optab_for_tree_code): Likewise.
3029 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3030 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3031 * expmed.c (make_tree): Handle VEC_SERIES.
3032 * optabs.def (vec_series_optab): New optab.
3033 * optabs.h (expand_vec_series_expr): Declare.
3034 * optabs.c (expand_vec_series_expr): New function.
3035 * tree-vect-generic.c (expand_vector_operations_1): Check that
3036 the operands also have vector type.
3038 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3039 Alan Hayward <alan.hawyard@arm.com>
3040 David Sherwood <david.sherwood@arm.com>
3042 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3043 (VEC_COND_EXPR): Add missing @tindex.
3044 * doc/md.texi (vec_duplicate@var{m}): Document.
3045 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3046 * tree.c (build_vector_from_val): Add stubbed-out handling of
3047 variable-length vectors, using VEC_DUPLICATE_EXPR.
3048 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3049 * cfgexpand.c (expand_debug_expr): Likewise.
3050 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3051 * tree-inline.c (estimate_operator_cost): Likewise.
3052 * tree-pretty-print.c (dump_generic_node): Likewise.
3053 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3054 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3055 (test_vec_duplicate_folding): New function.
3056 (fold_const_c_tests): Call it.
3057 * optabs.def (vec_duplicate_optab): New optab.
3058 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3059 * optabs.h (expand_vector_broadcast): Declare.
3060 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3061 vec_duplicate_optab.
3062 * expr.c (store_constructor): Try using vec_duplicate_optab for
3064 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3066 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3069 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3070 div/mod latencies a bit.
3072 2017-12-15 Jeff Law <law@redhat.com>
3074 PR tree-optimization/36550
3075 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3076 (mark_threaded_blocks): Rewrite code to avoid block copying when
3077 optimizing for size. Don't pessimize blocks which will be
3078 copied, but all the statements will be dead.
3080 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3082 PR tree-optimization/81165
3083 * tree-ssa-threadupdate.c (uses_in_bb): New.
3084 (estimate_threading_killed_stmts): New.
3085 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3086 * tree-ssa-threadedge.c
3087 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3090 2017-12-15 Jeff Law <law@redhat.com>
3092 PR tree-optimization/83410
3093 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3094 threads when parallelizing loops.
3096 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3098 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3100 * config/alpha/alpha.c (vms_attribute_table): Swap
3101 affects_type_identity and handler fields, adjust comments.
3102 * config/mips/mips.c (mips_attribute_table): Likewise.
3103 * config/visium/visium.c (visium_attribute_table): Likewise.
3104 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3105 * config/microblaze/microblaze.c (microblaze_attribute_table):
3107 * config/spu/spu.c (spu_attribute_table): Likewise.
3108 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3109 * config/arc/arc.c (arc_attribute_table): Likewise.
3110 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3111 * config/v850/v850.c (v850_handle_interrupt_attribute,
3112 v850_handle_data_area_attribute): Formatting fixes.
3113 (v850_attribute_table): Swap affects_type_identity and handler
3114 fields, adjust comments.
3115 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3116 * config/arm/arm.c (arm_attribute_table): Likewise.
3117 * config/avr/avr.c (avr_attribute_table): Likewise.
3118 * config/s390/s390.c (s390_attribute_table): Likewise.
3119 * config/sh/sh.c (sh_attribute_table): Likewise.
3120 * config/i386/i386.c (ix86_handle_cconv_attribute,
3121 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3122 (ix86_attribute_table): Swap affects_type_identity and handler
3123 fields, adjust comments.
3124 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3125 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3126 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3127 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3128 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3129 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3130 * config/rx/rx.c (rx_attribute_table): Likewise.
3131 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3132 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3133 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3134 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3135 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3136 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3137 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3138 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3139 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3140 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3141 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3143 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3145 (register_scoped_attributes, decl_attributes): Formatting fixes.
3147 PR tree-optimization/83269
3148 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3149 subtraction in arg0's type if type is signed and arg0 is unsigned.
3153 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3154 convert? on @0 instead of convert. Check type of @1, not @0.
3155 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3156 convert? on @0 instead of convert. Check type of @1, not @0.
3157 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3158 only optimize if either both @1 and @2 types are narrower
3159 precision, or both are wider or equal precision, and in the former
3160 case only if both have undefined overflow.
3162 2017-12-15 Richard Biener <rguenther@suse.de>
3165 * internal-fn.def (IFN_NOP): Add.
3166 * internal-fn.c (expand_NOP): Do nothing.
3167 * lto-streamer-in.c (input_function): Instead of removing
3168 sanitizer calls replace them with IFN_NOP calls.
3170 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3171 Alan Hayward <alan.hayward@arm.com>
3172 David Sherwood <david.sherwood@arm.com>
3174 * dse.c (store_info, read_info_type): Replace begin and end with
3176 (print_range): New function.
3177 (set_all_positions_unneeded, any_positions_needed_p)
3178 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3180 (record_store): Likewise. Optimize the case in which all positions
3182 (get_stored_val): Replace read_begin and read_end with read_offset
3184 (replace_read): Update call accordingly.
3186 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3188 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3189 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3190 (loop_cand::loop_cand): Initialize above members.
3191 (loop_cand::supported_operations): Delete.
3192 (loop_cand::can_interchange_p): Inline above function.
3193 (loop_cand::classify_simple_reduction): Record number of constant
3194 initialized simple reductions.
3195 (should_interchange_loops): New parameters. Check stmt cost of loops
3197 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3198 Update call to should_interchange_loops.
3199 (should_interchange_loop_nest): Update call to
3200 should_interchange_loops.
3202 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3205 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3206 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3207 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3209 2017-12-15 Julia Koval <julia.koval@intel.com>
3211 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3212 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3214 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3215 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3216 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3218 2017-12-15 Julia Koval <julia.koval@intel.com>
3220 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3221 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3222 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3223 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3224 _mm_aesenc_epi128): New intrinsics.
3226 2017-12-15 Julia Koval <julia.koval@intel.com>
3228 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3229 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3231 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3232 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3233 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3235 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3237 * gimple-ssa-strength-reduction.c (analyze_increments):
3238 Distinguish replacement costs for constant strides from those for
3241 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3243 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3245 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3247 * doc/invoke.texi: Document -Wcast-function-type.
3248 * recog.h (stored_funcptr): Change signature.
3249 * tree-dump.c (dump_node): Avoid warning.
3250 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3252 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3255 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3258 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3260 PR tree-optimization/83312
3261 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3262 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3263 "val" param, and to cope with arbitrary basic blocks.
3264 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3265 handle NULL_TREE for "val", dropping "bb" param.
3266 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3267 drop "bb" param. Handle NULL_TREE for "val".
3268 (find_case_label_for_value): Make "switch_stmt" param const.
3269 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3271 (vrp_prop::check_all_array_refs): Reimplement as...
3272 (check_array_bounds_dom_walker::before_dom_children): ...this new
3273 vfunc. Replace linear search through BB block list, excluding
3274 those with non-executable in-edges via dominator walk.
3276 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3278 * config/arm/arm.opt (mverbose-cost-dump): New option.
3279 * config/arm/arm.c (arm_rtx_costs): Use it.
3281 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3284 * reload1.c (emit_input_reload_insns): Skip debug markers.
3286 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3288 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3292 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3293 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3294 * config/sh/sh.c: Skip debug insns besides notes.
3295 * config/sh/sh.md: Likewise.
3296 * config/sh/sh_treg_combine.cc: Likewise.
3297 * config/sh/sync.md: Likewise.
3299 2017-12-14 Tom de Vries <tom@codesourcery.com>
3301 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3302 Add item for weak_undefined.
3304 2017-12-14 Richard Biener <rguenther@suse.de>
3306 PR tree-optimization/67842
3307 * sese.h (bb_in_region): Remove #if 0'ed code.
3309 2017-12-14 Richard Biener <rguenther@suse.de>
3311 PR tree-optimization/83326
3312 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3313 may_be_zero parameter and handle it by not marking the first
3314 peeled copy as not exiting the loop.
3315 (try_peel_loop): Likewise.
3316 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3317 to handle the case of constant or zero iterations and perform
3318 loop header copying on-the-fly.
3320 2017-12-14 Richard Biener <rguenther@suse.de>
3322 PR tree-optimization/83418
3323 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3324 Instead of asserting we don't get unfolded comparisons deal with them.
3326 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3329 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3330 basic blocks. Assert debug bind insns don't appear outside of bbs,
3331 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3334 PR tree-optimization/83198
3335 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3336 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3337 value if arg is a REAL_CST with incompatible type.
3339 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3340 Bin Cheng <bin.cheng@arm.com>
3343 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3345 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3348 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3350 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3351 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3353 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3354 Alan Hayward <alan.hayward@arm.com>
3355 David Sherwood <david.sherwood@arm.com>
3357 * poly-int.h: New file.
3358 * poly-int-types.h: Likewise.
3359 * coretypes.h: Include them.
3360 (POLY_INT_CONVERSION): Define.
3361 * target.def (estimated_poly_value): New hook.
3362 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3363 * doc/tm.texi: Regenerate.
3364 * doc/poly-int.texi: New file.
3365 * doc/gccint.texi: Include it.
3366 * doc/rtl.texi: Describe restrictions on subreg modes.
3367 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3368 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3369 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3370 * targhooks.h (default_estimated_poly_value): Declare.
3371 * targhooks.c (default_estimated_poly_value): New function.
3372 * target.h (estimated_poly_value): Likewise.
3373 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3374 (wi::unary_traits): Delete.
3375 (wi::binary_traits::signed_shift_result_type): Define for
3376 offset_int << HOST_WIDE_INT, etc.
3377 (generic_wide_int::operator <<=): Define for all types and use
3378 wi::lshift instead of <<.
3379 (wi::hwi_with_prec): Add a default constructor.
3380 (wi::ints_for): New class.
3381 (operator <<): Define for all wide-int types.
3382 (operator /): New function.
3383 (operator %): Likewise.
3384 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3385 (ASSERT_MAYBE_NE_AT): New macros.
3387 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3388 Dominik Vogt <vogt@linux.vnet.ibm.com>
3391 * emit-rtl.c (init_emit): Remove ??? comment.
3392 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3393 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3394 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3395 alignment of 3 virtual registers to BITS_PER_WORD.
3397 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3399 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3401 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3402 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3404 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3407 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3409 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3411 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3412 with the exception of debug begin stmt markers appear before
3416 * final.c (rest_of_handle_final): Call variable_tracking_main only
3417 if !flag_var_tracking.
3419 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3420 Jakub Jelinek <jakub@redhat.com>
3424 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3425 labels debug stmts that can only appear after labels.
3427 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3429 PR rtl-optimization/82398
3430 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3431 EXPR_USEFULNESS in priority comparison.
3433 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3435 PR rtl-optimization/83393
3436 * combine.c (move_deaths): If reg_stat points to a too new insn in
3437 last_death, do not use it: find the proper insn instead.
3439 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3442 * config/rs6000/rs6000-string.c (expand_block_compare,
3443 expand_strn_compare): Fix set_mem_size() calls.
3445 2017-12-12 Jeff Law <law@redhat.com>
3447 PR tree-optimization/83298
3448 PR tree-optimization/83362
3449 PR tree-optimization/83383
3450 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3451 push_value_range a public interface. Add new argument to
3452 record_ranges_from_stmt.
3453 * gimple-ssa-evrp-analyze.c
3454 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3455 Update comments. Handle recording temporary equivalences.
3456 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3457 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3458 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3459 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3460 gimple-ssa-evrp-analyze.h.
3461 (record_temporary_equivalences_from_phis): Add new argument. When
3462 the PHI arg is an SSA_NAME, set the result's range to the range
3464 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3465 from statements too.
3466 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3467 Pass it down to children as needed.
3468 (thread_outgoing_edges): Likewise.
3469 (thread_across_edge): Likewise. Push/pop range state as needed.
3470 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3472 2017-12-12 Julia Koval <julia.koval@intel.com>
3474 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3475 (PTA_CANNONLAKE): Remove PTA_CLWB.
3477 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3480 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3481 (vcondv2div2df): Likewise.
3482 (vconduv2dfv2di): Likewise.
3484 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3486 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3487 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3488 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3489 ATTR_MATHFN_FPROUNDING_ERRNO.
3491 2017-12-12 Richard Biener <rguenther@suse.de>
3493 PR tree-optimization/83385
3494 * graphite-scop-detection.c (get_order, order): Remove.
3495 (bb_to_rpo): New global.
3497 (build_scops): Sort pbbs in RPO order.
3499 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3501 * combine.c (simplify_set): Do not transform subregs to zero_extends
3502 if the destination mode is a vector mode.
3504 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3506 PR tree-optimization/83359
3507 * tree-cfg.h (fold_loop_internal_call): Declare.
3508 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3509 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3510 (find_loop_dist_alias): New function.
3511 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3512 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3515 PR tree-optimization/80631
3516 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3517 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3518 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3519 hardcoding zero as the value if COND_EXPR is never true. For
3520 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3521 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3522 hardcoding MAX_EXPR as the reduction operation.
3523 (is_nonwrapping_integer_induction): Allow negative step.
3524 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3525 vect_create_epilog_for_reduction, if no value is suitable, don't
3526 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3528 2017-12-12 Richard Biener <rguenther@suse.de>
3530 PR tree-optimization/81889
3531 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3532 range info from the non-wrapping IV instead of just the range
3535 2017-12-12 Julia Koval <julia.koval@intel.com>
3537 * config.gcc: Add vaesintrin.h.
3538 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3540 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3541 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3543 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3544 * config/i386/immintrin.h: Include vaesintrin.h.
3545 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3546 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3547 _mm_aesdec_epi128): New intrinsics.
3549 2017-12-12 Julia Koval <julia.koval@intel.com>
3551 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3552 OPTION_MASK_ISA_VAES_UNSET): New.
3553 (ix86_handle_option): Handle -mvaes.
3554 * config/i386/cpuid.h: Define bit_VAES.
3555 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3556 * config/i386/i386-c.c (__VAES__): New.
3557 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3558 (ix86_valid_target_attribute_inner_p): Ditto.
3559 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3560 * config/i386/i386.opt: Add -mvaes.
3561 * doc/invoke.texi: Ditto.
3563 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3565 * debug.h (gcc_debug_hooks): Add inline_entry.
3566 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3567 * debug.c (do_nothing_debug_hooks): Likewise.
3568 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3569 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3570 (dwarf2_lineno_debug_hooks): Likewise.
3572 * common.opt (gstatement-frontiers): New, setting
3573 debug_nonbind_markers_p.
3574 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3575 * toplev.c (process_options): Autodetect value for debug statement
3577 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3578 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3580 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3581 markers. Integrate source bind into debug stmt expand loop.
3582 (pass_expand::execute): Check debug marker limit. Avoid deep
3583 TER and expand debug locations for debug bind insns only.
3584 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3586 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3587 * final.c (reemit_insn_block_notes): Take current block from
3588 nonbind markers. Declare note where it's first set.
3589 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3590 begin stmt markers if enabled.
3591 (notice_source_line): Handle nonbind markers. Fail if their
3592 location is unknown or that of builtins.
3593 (rest_of_handle_final): Convert begin stmt markers to notes if
3594 var-tracking didn't run.
3595 (rest_of_clean_state): Skip begin stmt markers.
3596 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3598 * function.c (allocate_struct_function): Set begin_stmt_markers.
3599 * function.h (struct function): Add debug_marker_count counter
3600 and debug_nonbind_markers flag.
3601 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3602 * gimple-low.c (lower_function_body): Adjust
3603 debug_nonbind_markers.
3604 (lower_stmt): Drop or skip gimple debug stmts.
3605 (lower_try_catch): Skip debug stmts.
3606 * gimple.c (gimple_build_debug_begin_stmt): New.
3607 (gimple_copy): Increment debug_marker_count if copying one.
3608 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3609 * gimplify.c (rexpr_location): New.
3610 (rexpr_has_location): New.
3611 (warn_switch_unreachable_r): Handle gimple debug stmts.
3612 (shortcut_cond_r): Call expr_location.
3614 (find_goto_label): New.
3615 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3617 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3619 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3620 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3621 (LANG_HOOKS_INITIALIZER): ... this.
3622 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3623 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3625 (update_ebb_live_info): Skip debug insn markers.
3626 * lra.c (debug_insn_static_data): Rename to...
3627 (debug_bind_static_data): ... this.
3628 (debug_marker_static_data): New.
3629 (lra_set_insn_recog_data): Select one of the above depending
3631 (lra_update_isn_regno_info): Don't assume debug insns have
3633 (push_insns): Skip debug insns.
3634 * lto-streamer-in.c (input_function): Drop debug stmts
3635 depending on active options. Adjust debug_nonbind_markers.
3636 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3637 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3638 begin stmt marker notes.
3639 (print_insn): Likewise.
3640 * recog.c (extract_insn): Recognize rtl for debug markers.
3641 * rtl.def (DEBUG_MARKER): New.
3642 * tree-inline.c: Include params.h.
3643 (remap_gimple_stmt): Handle nonbind markers.
3644 (maybe_move_debug_stmts_to_successors): Likewise.
3645 (copy_debug_stmt): Likewise.
3646 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3647 markers regardless of no side effects.
3648 (tsi_link_before): Don't update container's side effects when adding
3649 a begin stmt marker.
3650 (tsi_link_after): Likewise.
3651 (expr_first): Skip begin stmt markers.
3652 (expr_last): Likewise.
3653 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3654 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3655 Disregard nonbind markers.
3656 * tree.c (make_node_stat): Don't set side effects for begin stmt
3658 (build1_stat): Likewise.
3659 * tree.def (DEBUG_BEGIN_STMT): New.
3660 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3661 * var-tracking.c (delete_debug_insns): Renamed to...
3662 (delete_vta_debug_insns): ... this.
3663 (reemit_marker_as_note): New.
3664 (vt_initialize): Reemit markers.
3665 (delete_vta_debug_insns): Likewise.
3666 (vt_debug_insns_local): Reemit or delete markers.
3667 (variable_tracking_main_1): Likewise.
3668 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3669 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3670 (gimple_debug_nonbind_marker_p): New.
3671 (gimple_build_debug_bind): Adjust.
3672 (gimple_build_debug_begin_stmt): New.
3673 * doc/invoke.texi (max-debug-marker-count): New param.
3674 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3675 (debug_parameter_ref, debug_marker): New.
3676 (NOTE_INSN_BEGIN_STMT): New.
3677 (DEBUG_INSN): Describe begin stmt markers.
3679 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3680 without debug insns, we wouldn't, but clean up debug insns
3681 after a control flow insn nevertheless.
3683 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3684 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3685 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3686 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3687 (rtl_tidy_fallthru_edge): Likewise.
3688 (rtl_verify_fallthru): Likewise.
3689 (rtl_verify_bb_layout): Likewise.
3690 (skip_insns_after_block): Likewise.
3691 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3692 * dwarf2out.c: Include print-rtl.h.
3693 (dwarf2out_next_real_insn): New.
3694 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3695 Dump debug binds in asm comments.
3696 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3697 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3698 callers to use gsi_start_nondebug_bb instead.
3699 (gsi_after_labels): Skip gimple debug stmts.
3700 (gsi_start_nondebug): New.
3701 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3702 (proper_loop_form_for_interchange): Adjust.
3703 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3704 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3705 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3706 (collect_fallthrough_labels): Likewise.
3707 (should_warn_for_implicit_fallthrough): Likewise.
3708 (warn_implicit_fallthrough_r): Likewise.
3709 (expand_FALLTHROUGH_r): Likewise.
3710 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3711 (graphite_copy_stmts_from_block): Skip nonbind markers.
3712 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3713 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3714 * jump.c (clean_barriers): Skip debug insns.
3715 * omp-expand.c (expand_parallel_call): Skip debug insns.
3716 (expand_task_call): Likewise.
3717 (remove_exit_barrier): Likewise.
3718 (expand_omp_taskreg): Likewise.
3719 (expand_omp_for_init_counts): Likewise.
3720 (expand_omp_for_generic): Likewise.
3721 (expand_omp_for_static_nochunk): Likewise.
3722 (expand_omp_for_static_chunk): Likewise.
3723 (expand_omp_simd): Likewise.
3724 (expand_omp_taskloop_for_outer): Likewise.
3725 (expand_omp_taskloop_for_inner): Likewise.
3726 (expand_oacc_for): Likewise.
3727 (expand_omp_sections): Likewise.
3728 (expand_omp_single): Likewise.
3729 (expand_omp_synch): Likewise.
3730 (expand_omp_atomic_load): Likewise.
3731 (expand_omp_atomic_store): Likewise.
3732 (expand_omp_atomic_fetch_op): Likewise.
3733 (expand_omp_atomic_pipeline): Likewise.
3734 (expand_omp_atomic_mutex): Likewise.
3735 (expand_omp_target): Likewise.
3736 (grid_expand_omp_for_loop): Likewise.
3737 (grid_expand_target_grid_body): Likewise.
3738 (build_omp_regions_1): Likewise.
3739 * omp-low.c (check_combined_parallel): Skip debug stmts.
3740 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3741 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3742 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3744 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3745 (make_edges): Likewise.
3746 (cleanup_dead_labels): Likewise.
3747 (gimple_can_merge_blocks_p): Likewise.
3748 (stmt_starts_bb_p): Likewise.
3749 (gimple_block_label): Likewise.
3750 (gimple_redirect_edge_and_branch): Likewise.
3751 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3753 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3755 * tree-pretty-print (print_declaration): Omit initializer in slim
3757 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3759 (eliminate_unnecessary_stmts): Stabilize block removal order.
3760 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3761 * var-tracking.c (get_first_insn): New.
3762 (vt_emit_notes): Call it.
3763 (vt_initialize): Walk any insns before the first BB.
3764 (delete_debug_insns): Likewise.
3766 * gimple.h (enum gimple_debug_subcode): Add
3767 GIMPLE_DEBUG_BEGIN_STMT.
3768 (gimple_debug_begin_stmt_p): New.
3769 (gimple_debug_nonbind_marker_p): New.
3770 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3771 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3772 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3773 * insn-notes.def (BEGIN_STMT): New.
3774 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3775 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3776 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3777 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3778 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3779 (INSN_DEBUG_MARKER_KIND): New.
3780 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3781 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3782 (INSN_VAR_LOCATION_PTR): New.
3783 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3785 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3786 deps only if debug bind insns are enabled.
3787 (pass_expand::execute): Avoid deep TER and expand
3788 debug locations for debug bind insns only.
3789 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3790 debug stmts special handling down to debug bind stmts.
3791 * combine.c (try_combine): Narrow debug insns special handling
3792 down to debug bind insns.
3793 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3794 Narrow debug insns preexisting special handling down to debug
3796 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3797 handling down to debug bind insns.
3798 * function.c (instantiate_virtual_regs): Skip debug markers,
3799 adjust handling of debug binds.
3800 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
3801 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
3802 * haifa-sched.c (schedule_insn): Narrow special handling of debug
3803 insns to debug bind insns.
3804 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
3805 special handling of debug stmts to debug bind stmts.
3806 * ipa-split.c (split_function): Likewise.
3807 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
3808 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
3810 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
3811 * regrename.c (build_def_use): Likewise.
3812 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
3813 (pass_cprop_hardreg): Narrow special casing of debug insns to
3815 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
3816 * reload1.c (reload): Likewise.
3817 * sese.c (sese_insert_phis_for_liveouts): Narrow special
3818 casing of debug stmts to debug bind stmts.
3819 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
3820 * ssa-iterators.h (num_imm_uses): Likewise.
3821 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
3822 debug stmts to debug bind stmts.
3823 * tree-inline.c (tree_function_versioning): Narrow special casing
3824 of debug stmts to debug bind stmts.
3825 * tree-loop-distribution.c (generate_loops_for_partition):
3826 Narrow special casing of debug stmts to debug bind stmts.
3827 * tree-sra.c (analyze_access_subtree): Narrow special casing
3828 of debug stmts to debug bind stmts.
3829 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3830 stmts to debug bind stmts.
3831 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3832 casing of debug stmts to debug bind stmts.
3833 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3834 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3835 casing of debug stmts to debug bind stmts.
3836 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3838 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3839 debug stmts to debug bind stmts.
3840 (gimple_replace_ssa_lhs): Likewise.
3841 (insert_debug_temp_for_var_def): Likewise.
3842 (insert_debug_temps_for_defs): Likewise.
3843 (reset_debug_uses): Likewise.
3844 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3845 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3846 (adjust_debug_stmts): Likewise.
3847 (adjust_phi_and_debug_stmts): Likewise.
3848 (vect_do_peeling): Likewise.
3849 * tree-vect-loop.c (vect_transform_loop): Likewise.
3850 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3851 * var-tracking.c (adjust_mems): Narrow special casing of debug
3852 insns to debug bind insns.
3853 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
3854 (compute_bb_dataflow, vt_find_locations): Likewise.
3855 (vt_expand_loc, emit_notes_for_changes): Likewise.
3856 (vt_init_cfa_base): Likewise.
3857 (vt_emit_notes): Likewise.
3858 (vt_initialize): Likewise.
3859 (vt_finalize): Likewise.
3861 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3862 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3863 (next_nonnote_nondebug_insn_bb): New.
3864 (prev_nonnote_nondebug_insn_bb): New.
3865 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
3866 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
3867 (prev_nonnote_nondebug_insn_bb): Declare.
3868 (next_nonnote_nondebug_insn_bb): Declare.
3869 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
3870 * cfgrtl.c (get_last_bb_insn): Likewise.
3871 * lra.c (push_insns): Likewise.
3873 2017-12-11 David Malcolm <dmalcolm@redhat.com>
3876 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
3878 * selftest.c (selftest::test_runner::test_runner): New ctor.
3879 (selftest::test_runner::~test_runner): New dtor.
3880 * selftest.h (class selftest::test_runner): New class.
3882 2017-12-11 Carl Love <cel@us.ibm.com>
3884 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3885 vec_extract_fp32_from_shortl]): Add #defines.
3886 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3887 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3888 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3889 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3890 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3891 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3892 * doc/extend.texi: Add documentation for the added builtins.
3894 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3896 PR rtl-optimization/80693
3897 PR rtl-optimization/81019
3898 PR rtl-optimization/81020
3899 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3900 are not mentioned in i3. Place the REG_UNUSED note on i2,
3901 possibly modified to REG_DEAD, if it did not originate in i3.
3903 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3905 * recog.c (store_data_bypass_p_1): New function.
3906 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3907 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3909 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3911 PR rtl-optimization/83361
3912 * ifcvt.c (if_convert): Call fixup_partitions.
3914 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3916 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3917 early folding of splat_u{8,16,32}.
3919 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3921 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3922 output_operand_lossage first argument with capital letter.
3923 (aarch64_override_options): Don't start error and sorry first argument
3924 with capital letter.
3926 2017-12-11 Andi Kleen <ak@linux.intel.com>
3928 PR gcov-profile/83355
3929 * auto-profile.c (string_table::get_index_by_decl): Don't
3930 recurse when abstract origin points to itself.
3932 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3934 PR tree-optimization/83320
3935 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3936 (prune_datarefs_not_in_loop): Ditto.
3938 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3940 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3942 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3944 PR tree-optimization/83337
3945 * gimple-loop-interchange.cc (compute_access_stride): Handle
3946 bitfield DRs properly.
3948 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3950 PR tree-optimization/83338
3951 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3952 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3953 vector of unsigned integers to vector of signed integers.
3955 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3957 PR rtl-optimization/83317
3958 * lra-constraints.c (process_address_1): Add insn code check.
3960 2017-12-08 Michael Matz <matz@suse.de>
3962 Fix PR tree-optimization/83323
3963 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3964 head-controlled loops and loop BBs.
3965 * common.opt (funroll-and-jam): Remove, instead ...
3966 (floop-unroll-and-jam): ... reuse this option.
3967 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3968 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3969 (-floop-unroll-and-jam): ... this option.
3971 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3973 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3974 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3975 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3977 2017-12-08 Richard Biener <rguenther@suse.de>
3980 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3981 handle accesses outside of zero-sized vars.
3983 2017-12-08 Martin Jambor <mjambor@suse.cz>
3985 PR tree-optimization/83141
3986 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3987 test for MEM_REFs implicitely changing types with padding. Remove
3989 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3991 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3993 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3995 * config/avr/avr.c (avr_attribute_table): Likewise.
3996 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3997 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3998 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3999 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
4000 of struct attribute_spec.
4001 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
4003 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
4006 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
4008 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4011 * profile-count.c (profile_count::from_gcov_type): Move from
4012 profile-count.h; handle overflow.
4013 * profile-count.h (profile_count::from_gcov_type): Move offline.
4015 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4017 PR rtl-optimization/83304
4018 * combine.c (move_deaths): If we do not know where a register died,
4021 2017-12-08 Richard Biener <rguenther@suse.de>
4023 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4024 Provide -fopt-info-loop feedback when we interchange in a nest.
4026 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4028 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4029 for armv6 ARM CPU IDs.
4031 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4033 * common/config/arm/arm-common.c: Include <algorithm>.
4034 (INCLUDE_VECTOR): Define.
4035 (compare_opt_names): New function.
4036 (arm_rewrite_selected_arch): Only strip out extensions that can be
4037 expressed through -mfpu. Sort the remaining extensions
4040 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4042 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4043 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4044 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4045 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4047 2017-06-08 Tristan Gingold <gindold@adacore.com>
4050 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4051 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4053 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4055 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4056 of struct attribute_spec.
4058 2017-12-08 Julia Koval <julia.koval@intel.com>
4060 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4061 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4062 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4063 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4064 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4065 _mm_maskz_dpwssds_epi32): Ditto.
4067 2017-12-08 Richard Biener <rguenther@suse.de>
4069 PR tree-optimization/81303
4070 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4071 conditions try to create a comparison vector type matching
4072 the data vector type.
4073 (vectorizable_condition): Adjust.
4074 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4075 Leave invariant conditions alone in case we can vectorize those.
4077 2017-12-08 Julia Koval <julia.koval@intel.com>
4079 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4080 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4081 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4082 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4083 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4085 2017-12-08 Julia Koval <julia.koval@intel.com>
4087 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4088 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4089 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4090 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4091 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4092 _mm_maskz_dpbusds_epi32): New intrinsics.
4094 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4096 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4098 (ld<bh>io_signed): Likewise.
4099 (st<bhw>io): Likewise.
4100 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4101 SMALL_INT12 constant integer operand.
4103 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4105 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4107 (nios2_large_constant_allowed): ...this. Adjust uses.
4108 (nios2_plus_symbolic_constant_p): Rename to...
4109 (nios2_plus_large_constant_p): ...this. Adjust uses.
4110 (nios2_legitimate_address_p): Correct CONST_INT handling.
4111 (nios2_symbolic_memory_operand_p): Rename to...
4112 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4113 (nios2_large_constant_p): Check for large constant integers too.
4114 (nios2_split_large_constant): Handle constant integers.
4115 (nios2_split_symbolic_memory_operand): Rename to...
4116 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4117 (nios2_legitimize_constant_address): Handle constant integers.
4118 (r0rel_constant_p): Handle small constant integers.
4119 (nios2_print_operand_address): Handle r0-relative integer addresses.
4120 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4121 * config/nios2/nios2.md: Adjust for renamed functions.
4123 2017-12-07 Andrew Waterman <andrew@sifive.com>
4125 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4126 (TARGET_HAVE_SRODATA_SECTION): New define.
4127 (riscv_select_section): New function.
4129 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4130 Alexander Monakov <amonakov@ispras.ru>
4131 Jakub Jelinek <jakub@redhat.com>
4134 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4136 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4138 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4139 a bool success value. Don't call output_operand_lossage here.
4140 (aarch64_print_ldpstp_address): Return a bool success value.
4141 (aarch64_print_operand_address): Call output_addr_const if
4142 aarch64_print_address_internal fails.
4143 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4144 'y'; call output_operand_lossage instead. Call output_operand_lossage
4145 if aarch64_print_ldpstp_address fails.
4147 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4149 * tree-vector-builder.h
4150 (tree_vector_builder::binary_encoded_nelts): Declare.
4151 * tree-vector-builder.c
4152 (tree_vector_builder::binary_encoded_nelts): New function.
4153 * fold-const.c (negate_expr_p): Likewise.
4154 (operand_equal_p, fold_checksum_tree): Likewise.
4155 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4156 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4157 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4158 (uniform_vector_p): Likewise.
4159 * varasm.c (const_hash_1, compare_constant): Likewise.
4160 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4161 (valid_lattice_transition): Operate directly on the VECTOR_CST
4163 * ipa-icf.c: Include tree-vector-builder.h.
4164 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4165 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4167 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4169 * tree.c (build_vector): Delete.
4170 * tree.h (build_vector): Make static and move into the self-testing
4173 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4175 * vector-builder.h (vector_builder::derived): New const overload.
4176 (vector_builder::elt): New function.
4177 * tree-vector-builder.h (tree_vector_builder::type): New function.
4178 (tree_vector_builder::apply_step): Declare.
4179 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4181 * gimple-fold.h (tree_vector_builder): Declare.
4182 (gimple_build_vector): Take a tree_vector_builder instead of a
4183 type and vector of elements.
4184 * gimple-fold.c (gimple_build_vector): Likewise.
4185 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4187 (get_initial_defs_for_reduction): Likewise.
4188 (vectorizable_induction): Likewise.
4190 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4192 * tree-vector-builder.h
4193 (tree_vector_builder::new_binary_operation): Declare.
4194 * tree-vector-builder.c
4195 (tree_vector_builder::new_binary_operation): New function.
4196 * fold-const.c (fold_relational_const): Use it.
4197 (const_binop): Likewise. Check that both input vectors have
4198 the same number of elements, thus excluding things like WIDEN_SUM.
4199 Check whether it is possible to operate directly on the encodings
4202 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4204 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4205 new_unary_operation, operating only on the encoded elements.
4206 (const_unop): Likewise.
4207 (exact_inverse): Likewise.
4208 (distributes_over_addition_p): New function.
4209 (const_binop): Use tree_vector_builder and new_unary_operation
4210 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4211 on the encoded elements unless the encoding is strided and the
4212 operation does not distribute over addition.
4213 (fold_convert_const): Use tree_vector_builder and
4214 new_unary_operation. Operate only on the encoded elements
4215 for truncating integer conversions, or for non-stepped encodings.
4217 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4219 * config/sparc/sparc.c: Include tree-vector-builder.h.
4220 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4221 * expmed.c: Include tree-vector-builder.h.
4222 (make_tree): Use tree_vector_builder instead of build_vector.
4223 * fold-const.c: Include tree-vector-builder.h.
4224 (const_binop): Use tree_vector_builder instead of build_vector.
4225 (const_unop): Likewise.
4226 (native_interpret_vector): Likewise.
4227 (fold_vec_perm): Likewise.
4228 (fold_ternary_loc): Likewise.
4229 * gimple-fold.c: Include tree-vector-builder.h.
4230 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4232 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4233 (simplify_vector_constructor): Use tree_vector_builder instead
4235 * tree-vect-generic.c: Include tree-vector-builder.h.
4236 (add_rshift): Use tree_vector_builder instead of build_vector.
4237 (expand_vector_divmod): Likewise.
4238 (optimize_vector_constructor): Likewise.
4239 * tree-vect-loop.c: Include tree-vector-builder.h.
4240 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4241 of build_vector. Explicitly use a stepped encoding for
4243 * tree-vect-slp.c: Include tree-vector-builder.h.
4244 (vect_get_constant_vectors): Use tree_vector_builder instead
4246 (vect_transform_slp_perm_load): Likewise.
4247 (vect_schedule_slp_instance): Likewise.
4248 * tree-vect-stmts.c: Include tree-vector-builder.h.
4249 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4250 (vect_gen_perm_mask_any): Likewise.
4251 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4252 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4254 (build_vector_from_val): Likewise. Explicitly use a duplicate
4257 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4259 * doc/generic.texi (VECTOR_CST): Describe new representation of
4261 * vector-builder.h: New file.
4262 * tree-vector-builder.h: Likewise.
4263 * tree-vector-builder.c: Likewise.
4264 * Makefile.in (OBJS): Add tree-vector-builder.o.
4265 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4266 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4267 (tree_vector): Change the number of elements to
4268 vector_cst_encoded_nelts.
4269 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4270 (VECTOR_CST_ELTS): Delete.
4271 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4272 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4273 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4274 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4275 (VECTOR_CST_ENCODED_ELT): Likewise.
4276 (vector_cst_encoded_nelts): New function.
4277 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4278 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4279 (vector_cst_int_elt, vector_cst_elt): Declare.
4280 * tree.c: Include tree-vector-builder.h.
4281 (tree_code_size): Abort if passed VECTOR_CST.
4282 (tree_size): Update for new VECTOR_CST layout.
4283 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4284 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4285 (build_vector): Use tree_vector_builder.
4286 (vector_cst_int_elt, vector_cst_elt): New functions.
4287 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4288 encoded elements and then create the vector in the canonical form.
4289 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4290 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4291 (tree_c_tests): Call test_vector_cst_patterns.
4292 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4294 (hash_tree): Likewise.
4295 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4296 (streamer_write_tree_header): Likewise.
4297 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4298 (streamer_alloc_tree): Likewise. Update call to make_vector.
4299 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4301 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4303 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4304 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4305 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4307 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4308 Richard Biener <rguenther@suse.de>
4310 PR tree-optimization/81303
4311 * Makefile.in (gimple-loop-interchange.o): New object file.
4312 * common.opt (floop-interchange): Reuse the option from graphite.
4313 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4314 -floop-interchange and mention it for -O3.
4315 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4316 * gimple-loop-interchange.cc: New file.
4317 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4318 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4319 * passes.def (pass_linterchange): New pass.
4320 * timevar.def (TV_LINTERCHANGE): New time var.
4321 * tree-pass.h (make_pass_linterchange): New declaration.
4322 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4323 interchange. Record IV before/after increment in new parameters.
4324 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4325 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4327 (check_reduction_path): ...New function here.
4328 * tree-vectorizer.h (check_reduction_path): New declaration.
4330 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4333 PR rtl-optimization/80818
4334 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4335 always early clobbered.
4336 * lra-lives.c (process_bb_lives): Check input hard regs for early
4337 clobbered non-operand hard reg.
4339 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4342 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4343 types_compatible_p, just that TYPE_MODE is the same.
4345 2017-12-07 Martin Sebor <msebor@redhat.com>
4348 * attribs.c (empty_attribute_table): Initialize new member of
4349 struct attribute_spec.
4350 (decl_attributes): Add argument. Handle mutually exclusive
4351 combinations of attributes.
4352 (selftests::test_attribute_exclusions): New function.
4353 (selftests::attribute_c_tests): Ditto.
4354 * attribs.h (decl_attributes): Add default argument.
4355 * selftest.h (attribute_c_tests): Declare.
4356 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4357 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4359 * doc/extend.texi (Common Function Attributes): Update const and pure.
4360 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4361 of struct attribute_spec.
4362 * config/arc/arc.c (arc_attribute_table): Same.
4363 * config/arm/arm.c (arm_attribute_table): Same.
4364 * config/avr/avr.c ( avr_attribute_table): Same.
4365 * config/bfin/bfin.c (bfin_attribute_table): Same.
4366 * config/cr16/cr16.c (cr16_attribute_table): Same.
4367 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4368 * config/h8300/h8300.c (h8300_attribute_table): Same.
4369 * config/i386/i386.c (ix86_attribute_table): Same.
4370 * config/ia64/ia64.c (ia64_attribute_table): Same.
4371 * config/m32c/m32c.c (m32c_attribute_table): Same.
4372 * config/m32r/m32r.c (m32r_attribute_table): Same.
4373 * config/m68k/m68k.c (m68k_attribute_table): Same.
4374 * config/mcore/mcore.c (mcore_attribute_table): Same.
4375 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4376 * config/mips/mips.c (mips_attribute_table): Same.
4377 * config/msp430/msp430.c (msp430_attribute_table): Same.
4378 * config/nds32/nds32.c (nds32_attribute_table): Same.
4379 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4380 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4381 * config/rl78/rl78.c (rl78__attribute_table): Same.
4382 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4383 * onfig/rx/rx.c (rx_attribute_table): Same.
4384 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4385 * config/sh/sh.c (sh_attribute_table): Same.
4386 * config/sparc/sparc.c (sparc_attribute_table): Same.
4387 * config/spu/spu.c (spu_attribute_table): Same.
4388 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4389 * config/v850/v850.c (v850_attribute_table): Same.
4390 * config/visium/visium.c (visium_attribute_table): Same.
4392 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4395 * config/arm/arm.c (INCLUDE_STRING): Define.
4396 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4397 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4400 2017-12-07 Michael Matz <matz@suse.de>
4402 Add unroll and jam pass
4404 * gimple-loop-jam.c: New file.
4405 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4406 * common.opt (funroll-and-jam): New option.
4407 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4408 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4409 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4410 * passes.def: Add pass_loop_jam.
4411 * timevar.def (TV_LOOP_JAM): Add.
4412 * tree-pass.h (make_pass_loop_jam): Declare.
4413 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4414 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4415 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4416 to flow_loop_tree_node_add.
4417 (duplicate_subloops, copy_loops_to): Append to sibling list.
4418 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4419 * doc/invoke.texi (-funroll-and-jam): Document new option.
4420 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4422 2017-12-07 Richard Biener <rguenther@suse.de>
4424 PR tree-optimization/83296
4425 PR tree-optimization/67769
4426 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4427 flow sensitive info in an unrelated BB.
4428 (value_replacement): Use reset_flow_sensitive_info.
4429 (minmax_replacement): Reset flow sensitive info on the def
4430 we move. Do not reset flow sensitive info in the whole BB
4431 we move the stmt to.
4432 (abs_replacement): Likewise.
4434 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4437 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4438 rs6000_cpu to the given -mcpu=, or to the default processor.
4440 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4442 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4443 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4444 instead of rs6000_cpu_attr.
4445 (rs6000_adjust_cost): Ditto.
4446 (is_microcoded_insn): Ditto.
4447 (rs6000_adjust_priority): Ditto.
4448 (rs6000_issue_rate): Ditto.
4449 (rs6000_use_sched_lookahead): Ditto.
4450 (rs6000_use_sched_lookahead_guard): Ditto.
4451 (rs6000_sched_reorder): Ditto.
4452 (force_new_group): Ditto.
4453 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4454 (group_ending_nop): Ditto.
4456 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4458 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4459 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4460 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4461 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4463 (rs6000_reassociation_width): Ditto.
4464 (rs6000_emit_epilogue): Ditto.
4465 (rs6000_adjust_cost): Ditto.
4466 (is_microcoded_insn): Ditto.
4467 (is_cracked_insn): Ditto.
4468 (rs6000_adjust_priority): Ditto.
4469 (rs6000_sched_reorder): Ditto.
4470 (rs6000_sched_reorder2): Ditto.
4471 (insn_must_be_first_in_group): Ditto.
4472 (insn_must_be_last_in_group): Ditto.
4473 (rs6000_register_move_cost): Ditto.
4474 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4477 2017-12-07 Julia Koval <julia.koval@intel.com>
4479 * config.gcc: Add vaesintrin.h.
4480 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4482 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4483 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4485 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4486 * config/i386/immintrin.h: Include vaesintrin.h.
4487 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4488 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4489 _mm_aesdec_epi128): New intrinsics.
4491 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4493 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4494 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4495 c-family/c-spellcheck.cc.
4496 (best_macro_match::best_macro_match): Likewise.
4498 (struct edit_distance_traits<cpp_hashnode *>): Move to
4499 c-family/c-spellcheck.h.
4500 (class best_macro_match): Likewise.
4502 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4504 PR tree-optimization/83293
4505 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4506 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4507 might insert into empty bb.
4510 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4511 simplify for plus with :c added, and pointer_plus without that.
4512 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4513 with undefined overflow and the conversion is not widening,
4514 perform negation in utype and only convert to type afterwards.
4515 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4516 simplify for plus with :c added, and pointer_plus without that.
4517 If type is integral with undefined overflow and the conversion is
4518 not widening, perform minus in utype and only convert to type
4519 afterwards. Move the last pointer_diff_expr simplify into the
4522 2017-12-06 Martin Sebor <msebor@redhat.com>
4524 PR tree-optimization/82646
4525 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4526 strncpy, not maxlen.
4528 2017-12-06 Martin Sebor <msebor@redhat.com>
4530 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4533 PR tree-optimization/83075
4534 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4535 strncat/strncpy don't change length of source string.
4537 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4540 2017-11-29 Martin Aberg <maberg@gaisler.com>
4542 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4543 to prevent b2bst errata sequence.
4544 (sqrtdf2_fix): Likewise.
4546 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4548 PR tree-optimization/81945
4549 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4550 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4551 to dest_cfun has orig_loop_num set, either remap it to the new
4552 loop number if the loop got moved too, or clear it.
4554 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4556 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4558 (thunderx2t99_nothing): New insn reservation.
4559 (thunderx2t99_mrs): New insn reservation.
4560 (thunderx2t99_multiple): New insn reservation.
4561 (thunderx2t99_alu_basi): Add bfx to reservation.
4562 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4564 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4567 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4569 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4571 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4572 tree-ssa-pre.c::remove_dead_inserted_code.
4573 * tree-ssa-dce.h: New file.
4574 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4575 (remove_dead_inserted_code): Move and rename to function
4576 tree-ssa-dce.c::simple_dce_from_worklist.
4577 (pass_pre::execute): Update use.
4579 2017-12-05 Richard Biener <rguenther@suse.de>
4581 PR tree-optimization/83277
4582 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4583 to code-gen liveout vars.
4585 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4587 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4588 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4589 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4592 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4594 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4596 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4597 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4600 2017-12-05 Richard Biener <rguenther@suse.de>
4602 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4603 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4604 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4605 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4607 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4609 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4611 2017-12-05 Martin Liska <mliska@suse.cz>
4612 Jakub Jelinek <jakub@redhat.com>
4614 * doc/invoke.texi: Document the options.
4615 * flag-types.h (enum sanitize_code): Add
4616 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4617 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4618 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4619 * opts.c: Define new sanitizer options.
4620 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4621 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4623 2017-12-05 Julia Koval <julia.koval@intel.com>
4625 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4626 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4627 (ix86_handle_option): Handle -mavx512vnni.
4628 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4629 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4630 * config/i386/i386-c (__AVX512VNNI__): New.
4631 * config/i386/i386.c (ix86_target_string): Handle new option.
4632 (ix86_valid_target_attribute_inner_p): Handle new option.
4633 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4634 * config/i386/i386.opt (mavx512vnni): New option.
4636 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4639 * athlon.md: Disable for generic.
4640 * haswell.md: Enable for generic.
4641 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4642 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4644 (ix86_adjust_cost): Move generic to haswell path.
4646 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4648 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4649 instead of MEM_P in a couple more places. Fix formatting issues.
4651 2017-12-04 Jim Wilson <jimw@sifive.com>
4653 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4654 instead of GP_REG_LAST-1.
4655 (riscv_adjust_libcall_cfi_prologue): Likewise.
4656 (riscv_adjust_libcall_cri_epilogue): Likewise.
4657 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4660 2017-12-04 Luis Machado <luis.machado@linaro.org>
4662 * ipa-pure-const.c (check_decl): Add missing newline.
4663 (state_from_flags): Likewise.
4665 2017-12-04 Jeff Law <law@redhat.com>
4667 PR tree-optimizatin/78496
4668 * gimple-ssa-evrp-analyze.h
4669 (evrp_range_analyzer::get_vr_values): Simplify.
4670 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4671 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4672 and gimple-ssa-evrp-analyze.h.
4673 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4674 (simplify_stmt_for_jump_threading): Copy a blob of code from
4675 tree-vrp.c to use ranges to simplify statements.
4676 (dom_opt_dom_walker::before_dom_children): Call
4677 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4678 (dom_opt_dom_walker::after_dom_children): Similarly for
4679 evrp_range_analyzer::leave.
4680 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4683 * gimple-ssa-evrp-analyze.c
4684 (evrp_range_analyzer::extract_range_from_stmt): Always use
4685 vr_values::update_value_range so preexisting range info is
4686 medged with new range info, even if the new range is VR_VARYING.
4688 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4690 * combine.c: Adjust comment.
4691 (use_crosses_set_p): Delete.
4692 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4693 (try_combine): Ditto.
4695 2017-12-04 Richard Biener <rguenther@suse.de>
4697 PR tree-optimization/83255
4698 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4699 Re-add zero-iteration check.
4701 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4703 PR rtl-optimization/83245
4704 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4705 hard registers as earlyclobber, also if not in an asm.
4707 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4711 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4714 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4715 rs6000_cpu based on cpu_index, not tune_index.
4717 2017-12-04 Richard Biener <rguenther@suse.de>
4719 PR tree-optimization/83238
4720 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4721 code match comment, rejecting invalid SESE regions.
4723 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4725 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4726 require base operand is a REG_POINTER prior to reload on targets
4727 with non-equivalent space registers.
4729 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4731 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4732 (update_specialized_profile): Fix updating of counts.
4733 (perhaps_add_new_callers): Likewise.
4735 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4738 * x86-tune.def: Remove obsolette FIXMEs.
4739 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4740 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4741 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4743 (X86_TUNE_PAD_RETURNS): Disable for generic.
4745 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4747 PR tree-optimization/83170
4748 PR tree-optimization/83241
4749 * gimple-ssa-store-merging.c
4750 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4751 gimple_vuse (ins_stmt) in case it has changed.
4752 (imm_store_chain_info::output_merged_store): Likewise.
4754 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4758 * tree-cfg.c (pass_warn_function_return::execute): Handle
4759 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4760 with BUILTINS_LOCATION.
4764 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4765 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4766 instead of DECL_MODE.
4768 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4769 last argument to rtx pointer.
4770 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4771 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4772 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4773 depending on the chosen ISAs.
4774 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4775 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4776 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4778 * config/i386/sse.md (mov<mode>_internal): Likewise.
4779 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4781 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4783 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4785 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4787 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4788 parameters from prototype.
4789 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4790 parameters. Don't print a comment.
4791 (emit_fusion_gpr_load): Adjust.
4792 (emit_fusion_load_store): Adjust.
4793 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4794 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4795 comment on the second line.
4797 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4800 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4801 rs6000_cpu based on cpu_index, not tune_index.
4803 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4805 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
4806 which_alternative instead of which_alternative + 1.
4807 (output_asm_insn): Print an extra tab if the template is short.
4809 2017-12-01 Jim Wilson <jimw@sifive.com>
4811 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
4813 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
4814 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
4816 * doc/tm.texi: Regenerate.
4818 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4821 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
4822 whether we can allocate pseudos before trying to fix an address.
4823 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
4824 memory address is indexed or indirect.
4825 (floatuns_<mode>si2_hw2): Likewise.
4827 2017-12-01 Jason Merrill <jason@redhat.com>
4829 * Makefile.in (TAGS): Add c-family/*.cc.
4831 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4833 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4834 (call_value_insn): Likewise.
4835 (sibcall_insn): Likewise.
4836 (sibcall_value_insn): Likewise.
4837 (movsi_aarch64): Likewise.
4838 (movdi_aarch64): Likewise.
4839 (add_losym_): Likewise.
4840 (ldr_got_small_): Likewise.
4841 (ldr_got_small_sidi): Likewise.
4842 (ldr_got_small_28k_): Likewise.
4843 (ldr_got_small_28k_sidi): Likewise.
4844 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4845 Move output_addr_const to symbolic case. Add error check.
4847 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4849 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4850 * config/s390/s390.md: Disable MVC merging peephole if it would
4851 disable operand forwarding.
4852 (new peephole2): Split MVCs if it would turn them into up to 2
4855 2017-12-01 Richard Biener <rguenther@suse.de>
4857 PR tree-optimization/83232
4858 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
4859 detection of same access. Instead of breaking the group here
4860 do not consider the duplicate. Add comment explaining real fix.
4862 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4864 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
4866 2017-12-01 Sudakshina Das <sudi.das@arm.com>
4868 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
4870 2017-12-01 Jakub Jelinek <jakub@redhat.com>
4872 * function.h (struct function): Remove cilk_frame_decl,
4873 is_cilk_function and calls_cilk_spawn fields.
4874 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
4876 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
4877 cilk_elemental field.
4878 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
4879 * target.def: Adjust comment.
4880 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
4881 Don't test cilk_elemental.
4883 PR tree-optimization/83233
4884 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4885 bswap_stat name for the struct.
4888 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4889 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4890 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4891 SWITCH_BREAK_LABEL_P set on the label.
4892 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4893 added for default case if it was missing and not all cases covered.
4894 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4895 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4897 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4898 to avoid -Wimplicit-fallthrough warning.
4899 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4900 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4902 PR tree-optimization/83221
4903 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4905 (init_reassoc): Formatting fix.
4908 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4909 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4913 * tree-cfg.c: Include asan.h.
4914 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4916 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4918 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4920 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4922 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4923 vpcmpeqd instruction.
4925 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4927 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4928 registers type for 128bit mode.
4930 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4932 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4934 * gengtype-state.c (read_state_pair): Likewise.
4935 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4936 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4937 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4938 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4939 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4940 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4941 * builtins.c (fold_builtin_3): Likewise.
4942 * graphite-scop-detection.c
4943 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4944 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4947 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4950 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4951 parameter. Return true if ignore_decl_rtl_set_p is true and other
4952 conditions are satisfied.
4953 * asan.h (asan_protect_global): Add new parameter.
4954 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4955 to asan_protect_global calls.
4957 2017-11-30 Jim Wilson <jimw@sifive.com>
4959 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4960 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4961 -mstrict-align, add info on default value. Delete redundant lines for
4962 -mabi. Add missing -mexplicit-relocs docs.
4964 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4966 * config/arc/arc.md (trap): New pattern.
4968 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4970 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4971 instruction to end into a delay slot.
4972 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4973 can be placed into a delay slot against reg_note.
4975 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4977 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4978 labels number of usages.
4980 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4982 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4984 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4986 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4989 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4990 multiplication by power of 2 constant into two shifts + comparison.
4992 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4995 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4997 2017-11-30 Richard Biener <rguenther@suse.de>
4999 PR tree-optimization/83202
5000 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
5001 allow_peel argument and guard peeling.
5002 (canonicalize_loop_induction_variables): Likewise.
5003 (canonicalize_induction_variables): Pass false.
5004 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
5005 peeling from cunrolli.
5007 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5009 * combine.c (try_combine): Print a message to dump file whenever
5010 I0, I1, or I2 cannot be combined into I3.
5012 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5014 PR rtl-optimization/83156
5015 PR rtl-optimization/82621
5016 * combine.c (try_combine): Don't split an I2 if one of the dests is
5017 set again before I3. Allow unused dests.
5019 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5021 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5023 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5025 PR rtl-optimization/80818
5026 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5027 recursively. Use insn code for clobber.
5028 (lra_set_insn_recog_data): Pass the new arg to
5029 collect_non_operand_hard_regs.
5030 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5032 (lra_update_insn_regno_info): Pass insn to
5033 add_regs_to_insn_regno_info.
5035 2017-11-29 Jim Wilson <jimw@sifive.com>
5036 Andrew Waterman <andrew@sifive.com>
5038 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5039 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5040 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5041 use SINGLE_SHIFT_COST.
5042 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5043 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5045 2017-11-29 Julia Koval <julia.koval@intel.com>
5047 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5048 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5049 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5050 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5051 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5052 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5053 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5054 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5055 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5056 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5057 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5058 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5059 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5060 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5061 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5062 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5063 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5064 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5065 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5066 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5067 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5068 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5069 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5070 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5071 __builtin_ia32_vpshldv_v2di_mask,
5072 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5073 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5074 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5076 2017-11-29 Julia Koval <julia.koval@intel.com>
5078 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5079 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5080 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5081 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5082 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5083 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5084 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5085 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5086 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5087 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5088 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5089 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5090 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5091 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5092 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5093 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5094 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5095 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5096 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5097 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5098 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5100 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5102 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5103 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5106 2017-11-29 Martin Aberg <maberg@gaisler.com>
5108 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5109 to prevent b2bst errata sequence.
5110 (sqrtdf2_fix): Likewise.
5112 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5114 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5115 (div_sqrt_insn_p): New function.
5116 (sparc_do_work_around_errata): Insert NOP instructions to
5117 prevent sequences that could trigger the TN-0013 errata for
5118 certain LEON3 processors.
5119 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5120 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5121 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5122 (in_branch_delay): Prevent div and sqrt in delay slot if
5124 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5126 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5128 * config/sparc/sparc.c (atomic_insn_p): New function.
5129 (sparc_do_work_around_errata): Insert NOP instructions to
5130 prevent sequences that could trigger the TN-0010 errata for
5132 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5133 instruction referable in atomic_insns_p.
5135 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5137 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5138 (atomic_compare_and_swap_leon3_1): Likewise.
5141 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5143 * config/sparc/sparc.c (fpop_insn_p): New function.
5144 (sparc_do_work_around_errata): Insert NOP instructions to
5145 prevent sequences that could trigger the TN-0012 errata for
5147 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5148 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5149 (in_branch_annul_delay): Prevent floating-point instructions
5150 in delay slot of annulled integer branch.
5152 2017-11-29 Richard Biener <rguenther@suse.de>
5154 PR tree-optimization/83202
5155 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5157 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5158 nodes vectorized to the same stmts multiple times.
5159 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5160 (vect_analyze_slp_instance): Adjust.
5161 (scalar_stmts_to_slp_tree_map_t): New typedef.
5162 (vect_schedule_slp_instance): Add a map recording the SLP node
5163 representing the vectorized stmts for a set of scalar stmts.
5164 Avoid code-generating redundancies.
5165 (vect_schedule_slp): Allocate map and pass it down.
5167 2017-11-29 Nathan Sidwell <nathan@acm.org>
5170 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5173 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5175 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5176 cases for printing LDP/STP memory addresses.
5177 (aarch64_print_address_internal): Renamed from
5178 aarch64_print_operand_address, added parameter, add Pmode check.
5179 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5180 (aarch64_print_operand_address): Indirect to
5181 aarch64_print_address_internal.
5182 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5183 'y' operand output specifier.
5185 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5188 * tree.c (build_simple_mem_ref_loc): Handle
5189 get_addr_base_and_unit_offset returning a MEM_REF.
5192 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5195 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5198 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5200 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5202 (layout::move_to_column): Likewise.
5204 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5207 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5208 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5210 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5211 Martin Jambor <mjambor@suse.cz>
5214 * tree.h (expr_type_first_operand_type_p): Declare
5215 * tree.c (expr_type_first_operand_type_p): New function.
5216 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5217 (ipa_value_from_jfunc): Adjust declaration.
5218 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5219 Use it as result type for arithmetics, unless it is NULL in which case
5220 be more conservative.
5221 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5222 ipa_get_jf_pass_through_result.
5223 (propagate_vals_across_pass_through): Likewise.
5224 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5225 is to propagate_vals_across_pass_through.
5226 (propagate_constants_across_call): Pass PARM_TYPE to
5227 propagate_scalar_across_jump_function.
5228 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5229 ipa_value_from_jfunc.
5230 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5231 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5232 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5233 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5234 target_type, pass it to ipa_value_from_jfunc.
5235 (update_indirect_edges_after_inlining): Pass parameter type to
5236 try_make_edge_direct_simple_call.
5238 2017-11-28 Jeff Law <law@redhat.com>
5240 * gimple-ssa-evrp-analyze.c
5241 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5242 refine ranges if scev_initialized_p returns true.
5243 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5245 2017-11-28 Julia Koval <julia.koval@intel.com>
5247 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5248 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5249 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5250 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5251 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5252 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5253 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5254 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5255 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5256 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5257 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5258 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5259 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5260 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5261 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5262 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5263 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5264 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5265 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5266 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5267 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5268 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5270 2017-11-28 Julia Koval <julia.koval@intel.com>
5272 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5273 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5274 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5275 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5276 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5277 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5278 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5279 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5280 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5281 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5282 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5283 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5284 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5285 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5286 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5287 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5288 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5289 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5290 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5291 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5292 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5293 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5294 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5295 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5296 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5297 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5298 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5299 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5300 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5301 __builtin_ia32_vpshld_v2di_mask): New builtins.
5302 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5303 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5305 2017-11-28 Richard Biener <rguenther@suse.de>
5307 PR tree-optimization/80776
5308 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5310 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5312 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5313 If the incoming edge is an effective fallthru because the other
5314 edge only reaches a __builtin_unreachable () then record ranges
5315 derived from the controlling condition in SSA info.
5316 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5317 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5319 2017-11-28 Olivier Hainque <hainque@adacore.com>
5321 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5322 on mingw build hosts.
5324 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5327 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5329 2017-11-28 Richard Biener <rguenther@suse.de>
5332 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5333 copies generated from memcpy use a character array as reference
5336 2017-11-28 Julia Koval <julia.koval@intel.com>
5337 Sebastian Peryt <sebastian.peryt@intel.com>
5339 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5340 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5341 cilk-common.o, cilk.h, cilk-common.c): Remove.
5343 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5344 * builtins.c (is_builtin_name): Remove cilkplus condition.
5345 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5346 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5347 cilk-builtins.def, cilkplus.def): Remove.
5348 * cif-code.def (CILK_SPAWN): Remove.
5349 * cilk-builtins.def: Delete.
5350 * cilk-common.c: Ditto.
5352 * cilkplus.def: Ditto.
5353 * config/darwin.h (fcilkplus): Delete.
5354 * cppbuiltin.c: Ditto.
5355 * doc/extend.texi: Remove cilkplus doc.
5356 * doc/generic.texi: Ditto.
5357 * doc/invoke.texi: Ditto.
5358 * doc/passes.texi: Ditto.
5359 * gcc.c (fcilkplus): Remove.
5360 * gengtype.c (cilk.h): Remove.
5361 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5363 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5365 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5367 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5368 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5369 cilkplus conditions.
5370 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5371 inline_read_section): Ditto.
5372 * ipa-inline-analysis.c (cilk.h): Remove.
5373 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5374 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5375 append_linker_options): Remove condition for fcilkplus.
5376 * lto/lto-lang.c (cilk.h): Remove.
5377 (lto_init): Remove condition for fcilkplus.
5378 * omp-expand.c (expand_cilk_for_call): Delete.
5379 (expand_omp_taskreg, expand_omp_for_static_chunk,
5380 expand_omp_for): Remove cilkplus
5382 (expand_cilk_for): Delete.
5383 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5384 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5385 execute_lower_omp, diagnose_sb_0): Ditto.
5386 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5387 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5388 * tree-nested.c: Ditto.
5389 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5390 (dump_generic_node): Ditto.
5391 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5392 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5394 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5396 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5398 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5399 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5400 for vector mode and !TARGET_SIMD.
5402 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5404 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5406 * tree.h (SWITCH_LABELS): Remove.
5407 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5408 assert SWITCH_BODY is non-NULL.
5409 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5411 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5413 PR tree-optimization/80788
5414 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5415 has TREE_OVERFLOW set, call drop_tree_overflow.
5417 2017-11-28 Richard Biener <rguenther@suse.de>
5419 PR tree-optimization/83158
5420 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5422 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5425 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5426 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5428 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5430 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5431 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5433 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5436 * varasm.c (compare_constant): Take the mode of the constants into
5437 account when comparing floating point constants.
5439 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5441 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5442 from explicit instantiation of debug_helper.
5443 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5445 2017-11-27 Richard Biener <rguenther@suse.de>
5447 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5450 2017-11-27 Richard Biener <rguenther@suse.de>
5452 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5453 (type_hash_canon): Also clear int_cst_hash_table entry for
5455 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5457 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5459 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5460 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5461 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5463 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5465 * hash-map.h (gt_cleare_cache): Avoid UB.
5467 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5469 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5470 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5471 (decide_unroll_runtime_iterations): Likewise.
5472 (decide_unroll_stupid): Likewise.
5474 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5477 * config/i386/i386.md: Add a loop with incssp.
5479 2017-11-27 Martin Jambor <mjambor@suse.cz>
5481 PR tree-optimization/81248
5482 * tree-sra.c (splice_param_accesses): Remove size check.
5483 (decide_one_param_reduction): Fix size check.
5484 * gimple-pretty-print.c (dump_profile): Silence warning.
5485 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5487 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5490 * dbxout.c (lastlineno): New variable.
5491 (dbx_debug_hooks): Use dbxout_switch_text_section as
5492 switch_text_section debug hook.
5493 (dbxout_function_end): Switch to current_function_section
5494 rather than function_section. If crtl->has_bb_partition,
5495 output just one N_FUN, depending on in_cold_section_p.
5496 (dbxout_source_line): Remember last lineno in lastlineno.
5497 (dbxout_switch_text_section): New function.
5498 (dbxout_function_decl): Adjust dbxout_block caller.
5499 (dbx_block_with_cold_children): New function.
5500 (dbxout_block): Return true if any LBRAC/RBRAC have been
5501 emitted. Use dbx_block_with_cold_children at depth == 0
5502 in second partition. Add PARENT_BLOCKNUM argument, pass
5503 it optionally adjusted to children. Output LBRAC/RBRAC
5504 around recursive call only if the block is in the current
5505 partition, if not and anything was output, emit empty
5507 * final.c (final_scan_insn): Compute cold_function_name
5508 before calling switch_text_section debug hook. Call
5509 that hook even if dwarf2out_do_frame if not emitting
5513 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5514 TREE_READONLY decls.
5516 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5518 PR rtl-optimization/82488
5519 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5521 2017-11-26 Julia Koval <julia.koval@intel.com>
5523 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5525 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5528 2017-11-26 Julia Koval <julia.koval@intel.com>
5530 * config/i386/driver-i386.c (host_detect_local_cpu):
5531 Detect skylake-avx512.
5533 2017-11-26 Julia Koval <julia.koval@intel.com>
5535 * config.gcc: Add -march=cannonlake.
5536 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5537 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5538 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5539 (PTA_CANNONLAKE): New.
5540 (processor_target_table): Add cannonlake.
5541 (ix86_option_override_internal): Ditto.
5542 (fold_builtin_cpu): Ditto.
5543 (get_builtin_code_for_version): Handle cannonlake.
5544 (M_INTEL_COREI7_CANNONLAKE): New.
5545 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5546 * doc/invoke.texi: Add -march=cannonlake.
5548 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5550 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5551 (try_init_one_plugin): Alternative implementation for MinGW.
5552 * Makefile.in (plugin_implib): New.
5553 (gengtype-lex.c): Fix broken AIX workaround.
5554 * configure: Regenerate.
5555 * doc/plugins.texi: Document support for MinGW.
5557 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5559 PR rtl-optimization/81553
5560 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5561 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5562 is a shift where C1 has different mode than the whole shift, use C1's
5563 mode for MULT rather than the shift's mode.
5566 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5567 builtins not enabled in the currently selected ISA.
5569 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5571 PR tree-optimization/71026
5572 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5573 (insert_reciprocals): Change to insert reciprocals before a division
5574 by a square and to insert the square of a reciprocal.
5575 (execute_cse_reciprocals_1): Change to consider division by a square.
5576 (register_division_in): Add importance parameter.
5578 2017-11-24 Richard Biener <rguenther@suse.de>
5580 PR tree-optimization/82402
5581 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5582 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5584 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5586 * match.pd (0-ptr): New transformation.
5588 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5591 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5592 when skipping checking.
5594 2017-11-24 Richard Biener <rguenther@suse.de>
5596 PR tree-optimization/83128
5597 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5598 (vn_reference_lookup_3): Likewise.
5600 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5603 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5604 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5607 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5608 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5609 check for the latter with gimple_call_builtin_p. Do not handle
5610 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5612 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5614 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5617 2017-11-23 Julia Koval <julia.koval@intel.com>
5619 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5620 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5621 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5622 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5623 _mm512_maskz_expandloadu_epi16): New intrinsics.
5624 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5625 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5626 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5627 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5628 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5629 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5630 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5631 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5632 _mm256_maskz_expandloadu_epi8): New intrinsics.
5633 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5634 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5635 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5636 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5637 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5638 * config/i386/sse.md (VI248_VLBW): New iterator.
5639 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5641 2017-11-23 Julia Koval <julia.koval@intel.com>
5643 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5644 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5645 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5646 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5647 _mm512_mask_compressstoreu_epi16): New.
5648 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5649 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5650 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5651 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5652 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5653 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5654 _mm256_mask_compressstoreu_epi8): New.
5655 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5656 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5657 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5658 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5659 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5660 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5661 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5662 __builtin_ia32_compresshi128_mask,
5663 __builtin_ia32_compressstoreuqi512_mask,
5664 __builtin_ia32_compressstoreuhi512_mask,
5665 __builtin_ia32_compressstoreuqi256_mask,
5666 __builtin_ia32_compressstoreuqi128_mask,
5667 __builtin_ia32_compressstoreuhi256_mask,
5668 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5669 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5671 (ix86_expand_special_args_builtin): Handle new types.
5672 (s4fma_expand): Handle new builtin array.
5673 * config/i386/immintrin.h: Include new headers.
5674 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5675 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5677 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5679 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5680 fixes. Declare temp and g variables at the top in order to avoid
5681 {} in most of the cases.
5683 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5685 * match.pd (ptr-0): New transformation.
5687 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5689 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5690 (struct addr_mode_cost_table): New.
5691 (struct tune_params): Add field addr_mode_costs.
5692 * config/arm/arm.c (generic_addr_mode_costs): New.
5693 (arm_slowmul_tune): Initialise addr_mode_costs field.
5694 (arm_fastmul_tune): Likewise.
5695 (arm_strongarm_tune): Likewise.
5696 (arm_xscale_tune): Likewise.
5697 (arm_9e_tune): Likewise.
5698 (arm_marvell_pj4_tune): Likewise.
5699 (arm_v6t2_tune): Likewise.
5700 (arm_cortex_tune): Likewise.
5701 (arm_cortex_a8_tune): Likewise.
5702 (arm_cortex_a7_tune): Likewise.
5703 (arm_cortex_a15_tune): Likewise.
5704 (arm_cortex_a35_tune): Likewise.
5705 (arm_cortex_a53_tune): Likewise.
5706 (arm_cortex_a57_tune): Likewise.
5707 (arm_exynosm1_tune): Likewise.
5708 (arm_xgene1_tune): Likewise.
5709 (arm_cortex_a5_tune): Likewise.
5710 (arm_cortex_a9_tune): Likewise.
5711 (arm_cortex_a12_tune): Likewise.
5712 (arm_cortex_a73_tune): Likewise.
5713 (arm_v7m_tune): Likewise.
5714 (arm_cortex_m7_tune): Likewise.
5715 (arm_v6m_tune): Likewise.
5716 (arm_fa726te_tune): Likewise.
5717 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5719 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5721 * config/arm/arm.c (arm_mem_costs): New function.
5722 (arm_rtx_costs_internal): Use arm_mem_costs.
5724 2017-11-23 Mark Wielaard <mark@klomp.org>
5726 * dwarf2out.c (init_sections_and_labels): Use generation to create
5727 unique ranges_section_label and ranges_base_label. Return generation.
5728 (output_rnglists): Add generation argument. Use generation to create
5729 unique ranges labels.
5730 (dwarf2out_finish): Get generation from init_sections_and_labels
5731 and pass generation to output_rnglists.
5733 2017-11-23 Mike Stump <mikestump@comcast.net>
5734 Eric Botcazou <ebotcazou@adacore.com>
5736 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5737 * cfgloop.h (struct loop): Add unroll field.
5738 * function.h (struct function): Add has_unroll bitfield.
5739 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5740 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5741 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5742 (pass_rtl_unroll_loops::gate): Likewise.
5743 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5744 for which loop->unroll==1.
5745 (decide_unroll_constant_iterations): Use note for consistency and
5746 take loop->unroll into account. Return early if loop->unroll is set.
5747 Fix thinko in existing test.
5748 (decide_unroll_runtime_iterations): Use note for consistency and
5749 take loop->unroll into account.
5750 (decide_unroll_stupid): Likewise.
5751 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5752 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5753 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5755 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5756 (print_loop): Print loop->unroll if set.
5757 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5758 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5759 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5761 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5762 loop->unroll is set and smaller than the trip count. Otherwise bypass
5763 entirely the heuristics if loop->unroll is set. Remove dead note.
5764 Fix off-by-one bug in other note.
5765 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5766 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5768 (tree_unroll_loops_completely): Make static.
5769 (pass_complete_unroll::execute): Use correct type for variable.
5770 (pass_complete_unrolli::execute): Fix formatting.
5771 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5773 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5775 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5776 enable when TARGET_PREFER_AVX128 is set.
5778 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5780 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5782 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5783 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5785 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5787 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5788 frequencies. Fix estimation of aggregate parameters.
5790 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5792 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5795 2017-11-23 Tom de Vries <tom@codesourcery.com>
5797 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
5799 2017-11-23 Tom de Vries <tom@codesourcery.com>
5801 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
5804 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
5807 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
5808 sibcall_value_pcrel_fdpic): Use local variable instead of
5810 (calli_tbr_rel): Add missing operand 2.
5811 (call_valuei_tbr_rel): Add missing operand 3.
5813 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5816 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
5817 bitpos/bitsize covering the whole destination, use store_expr only if
5818 the complex mode is the same. Otherwise, use expand_normal and if
5819 it returns CONCAT, subreg each part separately instead of trying to
5820 subreg the whole result.
5822 2017-11-23 Richard Biener <rguenther@suse.de>
5824 PR tree-optimization/23094
5825 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
5826 come here from walking over backedges in the first iteration.
5827 (vn_reference_lookup_3): Skip clobbers that store the same value.
5829 2017-11-23 Richard Biener <rguenther@suse.de>
5831 PR tree-optimization/81403
5832 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5833 a block we need a leader relative to.
5834 (phi_translate_1): For nary processing require a leader from
5835 get_representative_for given we run expression simplification
5836 using match-and-simplify. Remove previous fix.
5838 2017-11-22 Jeff Law <law@redhat.com>
5840 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5841 Use new method allocate_value_range rather than accessing the
5842 vrp_value_range_pool data member directly.
5843 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5844 to use extract_range_from_stmt method to avoid need for
5845 extract_range_from_assignment method.
5846 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5847 method rather than setting values_propgated data member directly.
5848 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5849 and values propagated data members and extract_range_from_assignment
5850 method. Reorder private data members to conform to standards.
5851 Add new methods set_lattice_propagation_complete and
5852 allocate_value_range.
5854 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
5856 PR rtl-optimization/83030
5857 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
5858 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
5859 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5860 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5861 CROSSING_JUMP_P flag is already set before setting it.
5862 * cfgrtl.c (fixup_partition_crossing): Likewise.
5863 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
5866 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5868 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
5869 Handle the case where both arguments are using gen_const_vec_series.
5871 2017-11-22 David Malcolm <dmalcolm@redhat.com>
5874 * pretty-print.c (pp_format): Move quoting implementation to
5875 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
5876 to pass address of "quote" local.
5877 (pp_begin_quote): New function.
5878 (pp_end_quote): New function.
5879 * pretty-print.h (printer_fn): Convert penultimate param from bool
5881 (pp_begin_quote): New decl.
5882 (pp_end_quote): New decl.
5883 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5884 param from bool to bool *.
5885 * tree-diagnostic.h (default_tree_printer): Likewise.
5887 2017-11-22 Jeff Law <law@redhat.com>
5889 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5890 Initialize vr_values.
5891 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5892 vr_values via vr_values class instance rather than delegators.
5893 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5894 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5895 (evrp_range_analyzer::push_value_range): Likewise.
5896 (evrp_range_analyzer::pop_value_range): Likewise.
5897 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5898 most delegators. Those remaining are exposed as public interfaces.
5899 Make vr_values a pointer and private.
5900 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5902 (evrp_range_analyzer::get_vr_value): New method.
5903 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5904 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5905 (class evrp_dom_walker): Attach evrp_folder class, initialize
5906 it in the ctor. Remove temporary delegators.
5907 (evrp_dom_walker::before_dom_children): Call methods in attached
5908 evrp_range_analyzer class via class instance pointer. Use
5909 free value_range_constant_singleton to remove need for
5910 op_with_constant_singleton_value delegator method. Do not
5911 create a vrp_prop class instance for every call! Narrow
5912 scope of a couple variables.
5913 (evrp_dom_walker::cleanup): Call methods in attached
5914 evrp_range_analyzer class via class instance pointer.
5915 * vr-values.h (class vr_values): Privatize many methods and
5918 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5920 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5921 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5922 * expr.c (expand_expr_real_2): Likewise.
5923 * fold-const.c (const_unop): Likewise.
5924 * optabs-tree.c (optab_for_tree_code): Likewise.
5925 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5926 * tree-inline.c (estimate_operator_cost): Likewise.
5927 * tree-pretty-print.c (dump_generic_node): Likewise.
5928 (op_code_prio): Likewise.
5929 (op_symbol_code): Likewise.
5930 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5931 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5932 * internal-fn.c (direct_internal_fn_optab): New function.
5933 (direct_internal_fn_array, direct_internal_fn_supported_p
5934 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5935 * fold-const-call.c (fold_const_reduction): New function.
5936 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5938 * tree-vect-loop.c: Include internal-fn.h.
5939 (reduction_code_for_scalar_code): Rename to...
5940 (reduction_fn_for_scalar_code): ...this and return an internal
5942 (vect_model_reduction_cost): Take an internal_fn rather than
5944 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5946 (vectorizable_reduction): Use internal functions rather than tree
5947 codes for the reduction operation. Update calls to the functions
5949 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5950 Use calls to internal functions rather than REDUC tree codes.
5951 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5953 2017-11-22 Olivier Hainque <hainque@adacore.com>
5955 * config/vxworks.c (vxworks_override_options): Pick default
5956 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5957 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5958 DWARF_GNAT_ENCODINGS_DEFAULT.
5959 * config/vxworksae.h: Likewise.
5961 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5963 PR tree-optimization/83104
5964 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5967 2017-11-22 Marek Polacek <polacek@redhat.com>
5968 H.J. Lu <hongjiu.lu@intel.com>
5969 Jason Merrill <jason@redhat.com>
5974 * calls.c (initialize_argument_information): Call
5975 warn_parameter_passing_abi target hook.
5976 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5977 argument onto stack.
5978 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5979 * common.opt: Update -fabi-version description.
5980 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5981 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5983 (ix86_is_empty_record): New function.
5984 (ix86_warn_parameter_passing_abi): New function.
5985 (TARGET_EMPTY_RECORD_P): Redefine.
5986 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5987 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5988 * doc/tm.texi: Regenerated.
5989 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5990 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5991 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5992 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5993 instead of int_size_in_bytes.
5994 * expr.c (copy_blkmode_to_reg): Likewise.
5995 * function.c (aggregate_value_p): Return 0 for empty types.
5996 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5998 (locate_and_pad_parm): Call arg size_in_bytes instead
6000 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
6001 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
6002 * target.def (empty_record_p, warn_parameter_passing_abi): New target
6004 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
6005 (std_gimplify_va_arg_expr): Skip empty records. Call
6006 arg_size_in_bytes instead size_in_bytes.
6007 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
6008 * tree-core.h (tree_type_common): Add empty_flag.
6009 (tree_decl_common): Update comments.
6010 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6012 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6013 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6015 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6016 * tree.c (default_is_empty_type): New function.
6017 (default_is_empty_record): New function.
6018 (arg_int_size_in_bytes): New function.
6019 (arg_size_in_bytes): New function.
6020 (get_ultimate_context): New function.
6021 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6022 TRANSLATION_UNIT_WARN_EMPTY_P.
6023 (default_is_empty_record, arg_int_size_in_bytes,
6024 arg_size_in_bytes, get_ultimate_context): Declare.
6026 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6028 * config/arm/arm.c (cmse_clear_registers): New function.
6029 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6030 code by call to cmse_clear_registers.
6031 (cmse_nonsecure_entry_clear_before_return): Likewise.
6033 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6035 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6036 (vdot_s32, vdotq_s32): New.
6037 (vdot_lane_u32, vdotq_lane_u32): New.
6038 (vdot_lane_s32, vdotq_lane_s32): New.
6041 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6044 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6045 for unsigned values with fewer HWIs than the precision.
6046 (test_overflow): New function.
6047 (wide_int_cc_tests): Call it.
6049 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6051 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6052 has the same mode class as Pmode.
6054 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6056 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6057 padding_bits_to_clear_ptr.
6058 (cmse_nonsecure_entry_clear_before_return): Likewise.
6060 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6062 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6063 auto_sbitap instead of integer bitfield to control register needing
6066 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6068 PR tree-optimization/83044
6069 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6070 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6071 that need the upper bound. Subtract offset from
6072 get_addr_base_and_unit_offset only if positive and subtract it
6073 before division by eltsize rather than after it.
6076 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6077 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6078 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6082 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6083 expand_binop with *mul_widen_optab, make sure at least one of the
6084 operands doesn't have VOIDmode.
6087 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6089 PR rtl-optimization/82044
6090 PR tree-optimization/82042
6091 * dse.c (record_store): Check for overflow.
6092 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6093 add_wild_read instead of clear_rhs_from_active_local_stores on
6094 overflow and log it into dump_file.
6096 2017-11-22 Richard Biener <rguenther@suse.de>
6098 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6099 fake edges to exit when looking for a place to insert.
6100 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6102 (insert_into_preds_of_block): Commit edge insertion immediately,
6103 assert that doesn't require new BBs.
6104 (fini_pre): Release expressions.
6105 (pass_pre::execute): Shuffle things around a bit, if the fn
6106 is too large do not compute AVAIL either as this is really the
6109 2017-11-22 Richard Biener <rguenther@suse.de>
6111 PR tree-optimization/83089
6112 * tree-if-conv.c (pass_if_conversion::execute): If anything
6113 changed reset SCEV and free the number of iteration estimates.
6115 2017-11-21 Martin Sebor <msebor@redhat.com>
6117 PR tree-optimization/82945
6118 * calls.h (warn_nonstring_bound): Remove unused function.
6120 2017-11-21 Martin Sebor <msebor@redhat.com>
6122 PR tree-optimization/82945
6123 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6124 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6125 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6127 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6128 * calls.h (get_attr_nonstring_decl): Declare new function.
6129 * doc/extend.texi (attribute nonstring): Update.
6130 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6131 get_attr_nonstring_decl and handle it.
6132 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6133 detection of nul-termination.
6134 (strlen_to_stridx): Change to a pointer.
6135 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6136 (pass_strlen::execute): Same.
6138 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6140 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6141 for the new option -mprefer-vector-width=[none|128|256|512].
6142 * config/i386/i386.c (ix86_target_string): remove old style options
6143 -mprefer-avx256 and make -mprefer-avx128 as alias.
6144 (ix86_option_override_internal): Apply defaults for the
6145 -mprefer-vector-width=[128|256] option.
6146 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6147 Implement macros to work with -mprefer-vector-width=.
6148 * config/i386/i386.opt: Implemented option
6149 -mprefer-vector-width=[none|128|256|512].
6150 * doc/invoke.texi: Documentation for
6151 -mprefer-vector-width=[none|128|256|512].
6153 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6155 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6157 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6159 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6161 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6163 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6165 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6166 * expr.c (expand_expr_real_2): Likewise.
6167 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6168 fold_binary_loc): Likewise.
6169 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6170 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6171 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6172 MINUS_EXPR transformations.
6173 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6174 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6175 * tree-inline.c (estimate_operator_cost): Likewise.
6176 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6177 op_symbol_code): Likewise.
6178 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6179 * vr-values.c (extract_range_from_binary_expr): Likewise.
6180 * varasm.c (initializer_constant_valid_p_1): Likewise.
6181 * tree.def: New tree code POINTER_DIFF_EXPR.
6183 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6185 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6186 integer suffix to movbe mnemonic.
6187 (*bswaphi2_movbe): Ditto.
6188 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6190 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6193 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6194 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6195 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6197 2017-11-21 Martin Liska <mliska@suse.cz>
6199 * tree-inline.c (expand_call_inline): Remove not needed
6202 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6203 Jakub Jelinek <jakub@redhat.com>
6206 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6207 Remove static keyword from f variable.
6209 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6211 PR tree-optimization/83086
6212 * gimple-ssa-store-merging.c
6213 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6214 rather than n.base_addr.
6216 2017-11-21 Martin Liska <mliska@suse.cz>
6218 PR rtl-optimization/82044
6219 PR tree-optimization/82042
6220 * dse.c (check_mem_read_rtx): Check for overflow.
6222 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6224 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6227 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6229 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6230 and its corresponding call arg location note.
6232 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6233 Andrew Burgess <andrew.burgess@embecosm.com>
6235 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6237 (arc_return_slot_offset): Likewise.
6238 (arc_eh_return_address_location): New declaration.
6239 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6240 (MUST_SAVE_REGISTER): Add exception handler case.
6241 (MUST_SAVE_RETURN_ADDR): Likewise.
6242 (arc_frame_pointer_required): Likewise.
6243 (arc_frame_pointer_needed): New function.
6244 (arc_compute_frame_size): Changed.
6245 (arc_expand_prologue): Likewise.
6246 (arc_expand_epilogue): Likewise.
6247 (arc_initial_elimination_offset): Likewise.
6248 (arc_return_slot_offset): Delete.
6249 (arc_eh_return_address_location): New function.
6250 (arc_builtin_setjmp_frame_value): Likewise.
6251 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6252 (EH_RETURN_STACKADJ_RTX): Define.
6253 (EH_RETURN_HANDLER_RTX): Likewise.
6254 * config/arc/arc.md (eh_return): Delete.
6256 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6258 * print-tree.h (debug_vec_tree): Remove prototype.
6259 * gdbinit.in (pvt): Remove macro.
6261 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6263 PR tree-optimization/83047
6264 * gimple-ssa-store-merging.c
6265 (imm_store_chain_info::output_merged_store): If the loads with the
6266 same vuse are in different basic blocks, for load_gsi pick a load
6267 location that is dominated by the other loads.
6270 * config/i386/i386.c (ix86_memmodel_check): Start
6271 -Winvalid-memory-model diagnostics with lowercase letter.
6274 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6275 set in_cold_section_p to the partition containing loc_list->first.
6276 When seeing loc_list->last_before_switch node, update secname and
6277 perform range_across_switch second partition handling only after that.
6280 * run-rtl-passes.c: Include debug.h.
6281 (run_rtl_passes): Call debug_hooks->assembly_start.
6282 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6286 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6287 OPTAB_DIRECT in calls to expand_simple_binop.
6289 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6292 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6294 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6297 * spellcheck.h (best_match::blithely_get_best_candidate): New
6300 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6302 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6303 with lower case letter.
6305 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6307 * config/i386/i386.md (bswaphi2): New expander.
6308 (*bswaphi2_movbe): New insn pattern.
6309 (bswaphi -> rorhi pepehole2): New peephole pattern.
6311 2017-11-20 Jeff Law <law@redhat.com>
6313 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6314 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6315 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6316 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6317 gimple-ssa-evrp-analyze.h.
6319 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6320 set BB_VISITED here.
6321 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6323 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6325 * tree-predcom.c: Add general comment on Store-Store chains.
6326 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6328 (get_chain_last_ref_at): Rename into...
6329 (get_chain_last_write_at): ...this.
6330 (get_chain_last_write_before_load): New function.
6331 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6332 CT_STORE_STORE when write reference is added.
6333 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6334 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6335 (initialize_root_vars_store_elim_1): Ditto.
6336 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6337 definition is created.
6338 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6339 chain by replacing it with dominant stored value.
6341 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6343 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6345 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6347 * vr-values.c (extract_range_from_binary_expr): Use a full range
6350 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6352 * config/arm/arm.md (R4_REGNUM): Define constant.
6353 (nonsecure_call_internal): Remove r4 clobber.
6354 (nonsecure_call_value_internal): Likewise.
6355 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6356 clobber and resequence match_operands.
6357 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6358 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6359 (nonsecure_call_value_reg_thumb2): Likewise.
6361 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6363 PR tree-optimization/78821
6364 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6365 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6366 to the address of the base rather than the base itself.
6367 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6368 (find_bswap_or_nop_finalize): New function.
6369 (find_bswap_or_nop): Use it.
6370 (bswap_replace): Return a tree rather than bool, change first
6371 argument from gimple * to gimple_stmt_iterator, allow inserting
6372 into an empty sequence, allow ins_stmt to be NULL - then emit
6373 all stmts into gsi. Fix up MEM_REF address gimplification.
6374 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6375 (struct store_immediate_info): Add N and INS_STMT non-static
6377 (store_immediate_info::store_immediate_info): Initialize them
6378 from newly added ctor args.
6379 (merged_store_group::apply_stores): Formatting fixes. Sort by
6381 (stmts_may_clobber_ref_p): For stores call also
6382 refs_anti_dependent_p.
6383 (gather_bswap_load_refs): New function.
6384 (imm_store_chain_info::try_coalesce_bswap): New method.
6385 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6386 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6387 (imm_store_chain_info::output_merged_store): Fail if number of
6388 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6389 and NOP_EXPR rhs_code.
6390 (pass_store_merging::process_store): Compute n and ins_stmt, if
6391 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6392 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6394 (pass_store_merging::execute): Calculate dominators.
6396 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6397 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6398 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6399 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6400 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6401 class pass_optimize_bswap, bswap_replace,
6402 pass_optimize_bswap::execute): Moved to ...
6403 * gimple-ssa-store-merging.c: ... this file.
6404 Include optabs-tree.h.
6405 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6406 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6407 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6408 anonymous namespace, remove static keywords.
6409 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6410 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6412 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6415 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6417 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6419 * vec.h (debug_helper): New function.
6420 (DEFINE_DEBUG_VEC): New macro.
6421 * hash-set.h (debug_helper): New function.
6422 (DEFINE_DEBUG_HASH_SET): New macro.
6423 * cfg.c (debug_slim (edge)): New function.
6424 Call DEFINE_DEBUG_VEC for edges.
6425 Call DEFINE_DEBUG_HASH_SET for edges.
6426 * cfghooks.c (debug_slim (basic_block)): New function.
6427 Call DEFINE_DEBUG_VEC for basic blocks.
6428 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6429 * print-tree.c (debug_slim): New function to handle trees.
6430 Call DEFINE_DEBUG_VEC for trees.
6431 Call DEFINE_DEBUG_HASH_SET for trees.
6432 (debug (vec<tree, va_gc>) &): Remove.
6433 (debug (<vec<tree, va_gc>) *): Remove.
6434 * print-rtl.c (debug_slim): New function to handle const_rtx.
6435 Call DEFINE_DEBUG_VEC for rtx_def.
6436 Call DEFINE_DEBUG_VEC for rtx_insn.
6437 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6438 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6439 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6440 (debug (vec<rtx_insn *> *ptr): Remove.
6441 (debug_insn_vector): Remove.
6442 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6444 2017-11-20 Tom de Vries <tom@codesourcery.com>
6446 PR rtl-optimization/82020
6447 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6448 IF_THEN_ELSE condition.
6450 2017-11-19 Jeff Law <law@redhat.com>
6452 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6453 of degenerates resulting from ignoring an edge.
6455 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6458 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6460 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6463 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6464 for uninitialied counts.
6466 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6469 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6471 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6474 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6476 2017-11-19 Tom de Vries <tom@codesourcery.com>
6478 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6481 2017-11-19 Tom de Vries <tom@codesourcery.com>
6483 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6486 2017-11-19 Tom de Vries <tom@codesourcery.com>
6488 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6491 2017-11-19 Tom de Vries <tom@codesourcery.com>
6493 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6495 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6497 2017-11-19 Tom de Vries <tom@codesourcery.com>
6499 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6501 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6502 semicolon after MCORE_EXPORT_NAME call.
6504 2017-11-19 Tom de Vries <tom@codesourcery.com>
6507 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6508 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6510 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6512 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6513 (cgraph_edge::clone): Cleanup updating of profile.
6514 * ipa-cp.c (update_profiling_info): Likewise.
6515 * ipa-inline-transform.c (inline_transform): Likewise.
6516 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6517 * ipa-split.c (execute_split_functions): Do not split when function
6519 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6520 * profile-count.c (profile_count::dump): Add global0.
6521 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6523 (profile_count::to_sreal_scale): Likewise.
6524 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6525 (profile_count::combine_with_ipa_count): New function.
6526 * profile-count.h (profile_guessed_global0adjusted): New.
6527 (profile_count::adjusted_zero): New.
6528 (profile_count::global0adjusted): New.
6529 (profile_count::combine_with_ipa_count): New.
6530 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6531 correct profile of return block of split functions.
6532 (copy_cfg_body): Remove unused profile_count.
6533 (copy_body): Likewise.
6534 (expand_call_inline): Update.
6535 (tree_function_versioning): Update.
6537 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6539 * hash-set.h (hash_set::empty): New.
6540 * tree-ssa-threadbackward.h: Delete.
6541 * tree-ssa-threadbackward.c (class thread_jumps): New.
6542 Move max_threaded_paths into class.
6543 (fsm_find_thread_path): Remove arguments that are now in class.
6544 (profitable_jump_thread_path): Rename to...
6545 (thread_jumps::profitable_jump_thread_path): ...this.
6546 (convert_and_register_jump_thread_path): Rename to...
6547 (thread_jumps::convert_and_register_current_path): ...this.
6548 (check_subpath_and_update_thread_path): Rename to...
6549 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6550 (register_jump_thread_path_if_profitable): Rename to...
6551 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6552 (handle_phi): Rename to...
6553 (thread_jumps::handle_phi): ...this.
6554 (handle_assignment): Rename to...
6555 (thread_jumps::handle_assignment): ...this.
6556 (fsm_find_control_statement_thread_paths): Rename to...
6557 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6558 (find_jump_threads_backwards): Rename to...
6559 (thread_jumps::find_jump_threads_backwards): ...this.
6560 Initialize path local data.
6561 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6562 from within thread_jumps class.
6563 (pass_early_thread_jumps::execute): Same.
6565 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6567 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6569 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6571 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6572 to canonical location.
6574 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6577 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6579 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6581 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6583 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6584 rcrt1.o%s/grcrt1.o%s for -static-pie.
6586 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6588 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6589 ix86_shift_rotate_cost): Break out from ...
6590 (ix86_rtx_costs): ... here.
6591 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6594 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6596 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6597 when entry block was promoted unlikely.
6598 (estimate_bb_frequencies): Increase frequency scale.
6599 * profile-count.h (profile_count): Export precision info.
6601 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6603 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6604 disturbin profile of entry block.
6606 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6608 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6611 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6613 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6614 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6616 2017-11-17 Jeff Law <law@redhat.com>
6618 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6619 from evrp_dom_walker class. Various methods moved into new class.
6620 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6621 (evrp_range_analyzer::enter): New method.
6622 (evrp_range_analyzer::leave): New method.
6623 (evrp_dom_walker): Remove delegators no longer needed by this class.
6624 Replace vr_values data member with evrp_range_analyzer
6626 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6627 method extracted from evrp_dom_walker::before_dom_children.
6628 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6629 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6631 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6632 Add private copy constructor and move assignment operators.
6633 Privatize methods and class data where trivially possible.
6634 (evrp_dom_walker::cleanup): New function, extracted from
6635 execute_early_vrp. Simplify access to class data.
6637 * vr-values.h (get_output_for_vrp): Prototype.
6638 * vr-values.c (get_output_for_vrp): New function extracted from
6639 vrp_visit_assignment_or_call and extract_range_from_stmt.
6640 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6642 2017-11-17 Luis Machado <luis.machado@linaro.org>
6644 * config/aarch64/aarch64.c
6645 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6646 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6647 tune_params::AUTOPREFETCHER_WEAK.
6649 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6652 * config/arm/arm.c (arm_valid_target_attribute_rec):
6653 Parse "arch=" and "+<ext>".
6654 (arm_valid_target_attribute_tree): Re-init global options.
6655 (arm_option_override): Make non-static.
6656 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6657 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6658 (__ARM_FEATURE_CRC32): Support undef.
6659 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6660 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6662 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6664 * gdbinit.in (break-on-diagnostic): New command.
6666 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6668 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6669 return to indirect jump for EH return if control-flow
6670 protection is enabled. Change explicit 'false' argument in
6671 pro_epilogue_adjust_stack with a value of
6673 * config/i386/i386.md (simple_return_indirect_internal):
6674 Remove SImode restriction to support 64-bit.
6676 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6678 * combine.c (added_notes_insn): New.
6679 (try_combine): Handle added_notes_insn like added_links_insn.
6680 Rewrite return value code.
6681 (distribute_notes): Set added_notes_insn to the earliest insn we added
6684 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6686 PR rtl-optimization/82621
6687 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6688 dest of one of those SETs is unused.
6690 2017-11-17 Richard Biener <rguenther@suse.de>
6693 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6694 * tree-pretty-print.c (dump_generic_node): Handle
6695 annot_expr_parallel_kind.
6696 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6697 * gimplify.c (gimple_boolify): Likewise.
6699 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6701 * config.gcc (extra_headers): Add cet.h for x86 targets.
6702 * config/i386/cet.h: New file.
6703 * doc/install.texi: Add --enable-cet/--disable-cet.
6705 2017-11-17 Richard Biener <rguenther@suse.de>
6707 PR tree-optimization/83017
6708 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6709 (gen_parallel_loop): Properly count iterations.
6710 (parallelize_loops): Handle loop->can_be_parallel independent
6711 of flag_loop_parallelize_all. Make static profitability test match
6713 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6714 * invoke.texi (parloops-min-per-thread): Document.
6716 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6718 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6719 upstreaming review comments.
6721 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6723 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6724 with fast unaligned access.
6725 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6727 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6729 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6730 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6732 2017-11-17 Richard Biener <rguenther@suse.de>
6734 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6735 folding of references.
6737 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6740 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6741 of replacing call to strncmp with corresponding call to strcmp when
6744 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6746 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6747 option prefer-avx256 for skylake-avx512 configuration.
6748 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6749 (get_builtin_code_for_version): Ditto.
6751 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6752 Monk Chiang <sh.chiang04@gmail.com>
6754 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6755 (FIXED_REGISTERS): Reserve more register numbers.
6756 (CALL_USED_REGISTERS): Likewise.
6757 (REG_ALLOC_ORDER): Likewise.
6758 (REG_CLASS_CONTENTS): Likewise.
6759 (REGISTER_NAMES): Likewise.
6761 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6762 Kito Cheng <kito.cheng@gmail.com>
6764 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6766 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6768 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6770 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6771 parameter list for vec_splats.
6773 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6775 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6777 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6779 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6782 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6783 with the required version of automake 1.11.6.
6785 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6787 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6789 (DU_C2_3_power9): Correct reservation combinations.
6790 (FP_DIV_power9, VEC_DIV_power9): New.
6791 (power9-alu): Split out rotate/shift...
6792 (power9-rot): ...to here, correct dispatch resource.
6793 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6795 (power9-fp): Correct latency.
6796 (power9-sdiv): Add div/sqrt resource.
6797 (power9-ddiv): Correct latency, add div/sqrt resource.
6798 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
6799 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
6801 (power9-qpdiv, power9-qpmul): Adjust resource usage.
6803 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
6805 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
6806 switch statement mapping KF built-ins to TF built-ins if we don't
6807 have the proper ISA 3.0 assembler support.
6809 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6811 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
6812 (lower_emutls_function_body): Do not compute it.
6814 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6816 * ipa-split.c (split_bb_info): Turn time to sreal.
6817 (split_point): Likewise.
6818 (dump_split_point): Likewise.
6819 (fine_split_points): Likewise.
6820 (execute_split_functions): Only zero split_bbs; turn time to sreals.
6822 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6824 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6827 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6829 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6831 (expensive_function_p): Remove useless assert.
6832 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6834 2017-11-16 Martin Sebor <msebor@redhat.com>
6836 PR tree-optimization/82588
6837 PR tree-optimization/82583
6838 * tree-vrp.c (check_array_ref): Handle flexible array members,
6839 string literals, and inner indices.
6840 (search_for_addr_array): Add detail to diagnostics.
6842 2017-11-16 Nathan Sidwell <nathan@acm.org>
6846 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6847 (SET_DECL_ASSEMBLER_NAME): Forward to
6848 overwrite_decl_assembler_name.
6849 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6850 (overwrite_decl_assembler_name): Declare.
6851 * tree.c (overwrite_decl_assembler_name): New.
6852 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
6853 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
6854 (LANG_HOOKS_INITIALIZER): Add it.
6855 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
6856 * langhooks.c (lhd_set_decl_assembler_name): Use
6857 SET_DECL_ASSEMBLER_NAME.
6858 (lhd_overwrite_decl_assembler_name): Default implementation.
6860 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6861 Jackson Woodruff <jackson.woodruff@arm.com>
6863 PR tree-optimization/71026
6864 * match.pd: Canonicalize constant multiplies in division.
6866 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6868 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
6869 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
6870 Remove OPT_fomit_frame_pointer entry.
6871 * common/config/arc/arc-common.c: Likewise.
6872 * common/config/arm/arm-common.c: Likewise.
6873 * common/config/avr/avr-common.c: Likewise.
6874 * common/config/c6x/c6x-common.c: Likewise.
6875 * common/config/cr16/cr16-common.c: Likewise.
6876 * common/config/cris/cris-common.c: Likewise.
6877 * common/config/epiphany/epiphany-common.c: Likewise.
6878 * common/config/fr30/fr30-common.c: Likewise.
6879 * common/config/frv/frv-common.c: Likewise.
6880 * common/config/ia64/ia64-common.c: Likewise.
6881 * common/config/iq2000/iq2000-common.c: Likewise.
6882 * common/config/lm32/lm32-common.c: Likewise.
6883 * common/config/m32r/m32r-common.c: Likewise.
6884 * common/config/mcore/mcore-common.c: Likewise.
6885 * common/config/microblaze/microblaze-common.c: Likewise.
6886 * common/config/mips/mips-common.c: Likewise.
6887 * common/config/mmix/mmix-common.c: Likewise.
6888 * common/config/mn10300/mn10300-common.c: Likewise.
6889 * common/config/nios2/nios2-common.c: Likewise.
6890 * common/config/pa/pa-common.c: Likewise.
6891 * common/config/pdp11/pdp11-common.c: Likewise.
6892 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6893 * common/config/riscv/riscv-common.c: Likewise.
6894 * common/config/rs6000/rs6000-common.c: Likewise.
6895 * common/config/rx/rx-common.c: Likewise.
6896 * common/config/s390/s390-common.c: Likewise.
6897 * common/config/sh/sh-common.c: Likewise.
6898 * common/config/sparc/sparc-common.c: Likewise.
6899 * common/config/tilegx/tilegx-common.c: Likewise.
6900 * common/config/tilepro/tilepro-common.c: Likewise.
6901 * common/config/v850/v850-common.c: Likewise.
6902 * common/config/visium/visium-common.c: Likewise.
6903 * common/config/xstormy16/xstormy16-common.c: Likewise.
6904 * common/config/xtensa/xtensa-common.c: Likewise.
6905 * invoke.texi (-fomit-frame-pointer): Update documentation.
6907 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6909 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6911 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6913 * cfg.c (scale_bbs_frequencies_int,
6914 cale_bbs_frequencies_gcov_type): Remove.
6915 * cfg.h (scale_bbs_frequencies_int,
6916 cale_bbs_frequencies_gcov_type): Remove.
6918 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6920 * tree-ssa-loop-manip.c
6921 (scale_dominated_blocks_in_loop): Update to profile counts.
6922 (tree_transform_and_unroll_loop): Likewise.
6924 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6926 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6927 scale_bbs_frequencies_int.
6929 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6931 * final.c (compute_alignments): Use counts rather than frequencies.
6933 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6935 * cfgloopanal.c: Include sreal.h
6936 (average_num_loop_insns): Use counts and sreal for accounting.
6938 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6940 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6943 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6944 Kito Cheng <kito.cheng@gmail.com>
6946 * config/nds32/constraints.md: Provide more constraints.
6947 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6948 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6949 support constraints usage.
6951 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6953 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6955 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6956 Kito Cheng <kito.cheng@gmail.com>
6958 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6959 * config/nds32/nds32.opt: Refine the layout.
6960 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6961 TARGET_EXT_STRING): Support new options.
6962 * config/nds32/nds32.h: Likewise.
6963 * config/nds32/nds32.md: Likewise.
6964 * config/nds32/nds32-predicates.c: Likewise.
6965 * config/nds32/constraints.md: Likewise.
6966 * common/config/nds32/nds32-common.c: Likewise.
6968 2017-11-16 Julia Koval <julia.koval@intel.com>
6971 * config/i386/gfniintrin.h: Add sse check.
6972 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6974 2017-11-16 Julia Koval <julia.koval@intel.com>
6976 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6977 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6978 (ix86_handle_option): Handle -mavx512vbmi2.
6979 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6980 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6981 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6982 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6983 (ix86_valid_target_attribute_inner_p): Ditto.
6984 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6985 * config/i386/i386.opt (mavx512vbmi2): New option.
6986 * doc/invoke.texi: Add new option.
6988 2017-11-16 Julia Koval <julia.koval@intel.com>
6990 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6991 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6992 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6993 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6994 _mm512_gf2p8mul_epi8): New intrinsics.
6995 * config/i386/i386-builtin-types.def
6996 (V64QI_FTYPE_V64QI_V64QI): New type.
6997 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6998 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6999 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
7000 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
7001 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
7002 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
7004 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
7006 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
7007 explicitly as a stream of bytes.
7009 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7011 * config/rs6000/altivec.h (vec_xst_be): New #define.
7012 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7013 and externalize from *altivec_vperm_<mode>_internal.
7014 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7016 (XL_BE_V8HI): Likewise.
7017 (XL_BE_V4SI): Likewise.
7018 (XL_BE_V4SI): Likewise.
7019 (XL_BE_V2DI): Likewise.
7020 (XL_BE_V4SF): Likewise.
7021 (XL_BE_V2DF): Likewise.
7022 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7023 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7024 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7025 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7026 VSX_BUILTIN_VEC_XST_BE.
7027 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7028 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7030 (altivec_init_builtins): Replace conditional calls to def_builtin
7031 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7032 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7033 with unconditional calls. Remove calls to def_builtin for
7034 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7035 __builtin_vec_xst_be.
7036 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7037 to define_expand, and add alternate RTL generation for P8.
7038 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7039 vsx_ld_elemrev_v8hi.
7040 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7041 add alternate RTL generation for P8.
7042 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7043 vsx_ld_elemrev_v16qi.
7044 (vsx_st_elemrev_v8hi): Convert define_insn
7045 to define_expand, and add alternate RTL generation for P8.
7046 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7047 vsx_st_elemrev_v8hi.
7048 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7049 add alternate RTL generation for P8.
7050 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7051 vsx_st_elemrev_v16qi.
7053 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7056 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7057 TARGET_AVX512ER check.
7058 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7059 neither -mzeroupper nor -mno-zeroupper is used and
7060 TARGET_EMIT_VZEROUPPER is set.
7061 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7062 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7064 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7066 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7067 folding of vector compares.
7068 (fold_build_vec_cmp): New helper function.
7069 (fold_compare_helper): New helper function.
7070 (builtin_function_type): Add compare builtins to the list of functions
7071 having unsigned arguments. Cosmetic updates to comment indentation.
7072 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7073 the not+eq combination.
7075 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7077 PR tree-optimization/82726
7078 PR tree-optimization/70754
7079 * tree-predcom.c (order_drefs_by_pos): New function.
7080 (combine_chains): Move code setting has_max_use_after to...
7081 (try_combine_chains): ...here. New parameter. Sort combined chains
7082 according to position information.
7083 (tree_predictive_commoning_loop): Update call to above function.
7084 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7086 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7088 PR tree-optimization/82726
7090 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7092 PR tree-optimization/70754
7093 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7094 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7095 combined stmt before it if not NULL.
7096 (combine_chains): Process refs reversely and compute dominance point
7100 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7102 PR tree-optimization/79663
7103 * tree-predcom.c (combine_chains): Process refs in reverse order
7104 only for ZERO length chains, and add explaining comment.
7106 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7108 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7109 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7110 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7112 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7114 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7116 2017-11-15 Martin Liska <mliska@suse.cz>
7118 * tree-cfg.c (pass_warn_function_return::execute):
7119 Compare warn_return_type for greater than zero.
7121 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7125 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7126 to return true on Xeon and not on Xeon Phi.
7127 (ix86_check_avx256_register): Changed to ...
7128 (ix86_check_avx_upper_register): ... this. Add extra check for
7129 VALID_AVX512F_REG_OR_XI_MODE.
7130 (ix86_avx_u128_mode_needed): Changed
7131 ix86_check_avx256_register to ix86_check_avx_upper_register.
7132 (ix86_check_avx256_stores): Changed to ...
7133 (ix86_check_avx_upper_stores): ... this. Changed
7134 ix86_check_avx256_register to ix86_check_avx_upper_register.
7135 (ix86_avx_u128_mode_after): Changed
7136 avx_reg256_found to avx_upper_reg_found. Changed
7137 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7138 (ix86_avx_u128_mode_entry): Changed
7139 ix86_check_avx256_register to ix86_check_avx_upper_register.
7140 (ix86_avx_u128_mode_exit): Ditto.
7141 * config/i386/i386.h: (host_detect_local_cpu): New define.
7143 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7145 * config/arm/xgene1.md (xgene1): Split into automatons
7146 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7147 (xgene1_f_load): Adjust reservations and/or types.
7148 (xgene1_f_store): Likewise.
7149 (xgene1_load_pair): Likewise.
7150 (xgene1_store_pair): Likewise.
7151 (xgene1_fp_load1): Likewise.
7152 (xgene1_load1): Likewise.
7153 (xgene1_store1): Likewise.
7154 (xgene1_move): Likewise.
7155 (xgene1_alu): Likewise.
7156 (xgene1_simd): Likewise.
7157 (xgene1_bfm): Likewise.
7158 (xgene1_neon_load1): Likewise.
7159 (xgene1_neon_store1): Likewise.
7160 (xgene1_neon_logic): Likewise.
7161 (xgene1_neon_st1): Likewise.
7162 (xgene1_neon_ld1r): Likewise.
7163 (xgene1_alu_cond): Added.
7164 (xgene1_shift_reg): Likwise.
7165 (xgene1_bfx): Likewise.
7166 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7168 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7171 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7173 (can_widen_mult_without_libcall): New function.
7174 (expand_mul_overflow): If only checking unsigned mul overflow,
7175 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7176 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7177 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7178 (expand_DIVMOD): Formatting fix.
7179 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7180 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7181 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7183 PR tree-optimization/82977
7184 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7185 constructed temporary to strlen_to_stridx.put.
7187 2017-11-15 Martin Liska <mliska@suse.cz>
7189 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7190 * configure: Regenerate.
7192 2017-11-15 Martin Liska <mliska@suse.cz>
7195 * config/sh/sh-mem.cc: Use proper probability for
7198 2017-11-14 Jeff Law <law@redhat.com>
7200 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7201 the red zone for stack_clash_protection_final_dynamic_probe targets
7202 when the total dynamic stack size is zero bytes.
7204 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7205 blocks is post order.
7207 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7209 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7210 * final.c (rest_of_clean_state): Set it for the
7211 -fcompare-debug dump.
7212 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7213 class when TDF_COMPARE_DEBUG is set.
7215 * dwarf2out.c (gen_producer_string): Discard
7218 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7221 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7222 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7223 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7224 (__TGMATH_REAL_2_3): Remove macros.
7225 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7226 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7228 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7230 (remquo): Define using __TGMATH_REAL_3.
7232 2017-11-14 Jeff Law <law@redhat.com>
7234 * vr-values.c: New file with contents extracted from tree-vrp.c.
7235 * Makefile.in (OBJS): Add vr-values.o
7236 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7237 (set_value_range, set_and_canonicalize_value_range): Likewise.
7238 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7239 (value_range_constant_singleton, symbolic_range_p): Likewise.
7240 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7241 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7242 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7243 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7244 (find_case_label_range, find_case_label_index): Likewise.
7245 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7246 (range_int_cst_singleton_p, value_inside_range): Likewise.
7247 (get_single_symbol): Likewise.
7248 (switch_update): Move structure definition here.
7249 (to_remove_edges, to_update_switch_stmts): Provide externs.
7250 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7251 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7252 (vrp_val_is_min, set_value_range): Likewise.
7253 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7254 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7255 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7256 (range_is_nonnull, range_int_cst_p): Likewwise.
7257 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7258 (get_single_symbol, operand_less_p): Likewise
7259 (compare_values_warnv, compare_values): Likewise.
7260 (value_inside_range, value_range_constant_singleton): Likewise.
7261 (zero_nonzero_bitgs_from_vr): Likewise.
7262 (extract_range_from_binary_expr_1): Likewise.
7263 (overflow_comparison_p): Likewise.
7264 (to_remove_edges, to_update_switch_stmts): Likewise.
7265 (find_case_label-index, find_case_label_range): Likewise.
7266 (switch_update, set_value_range_to_nonnegative): Remove.
7267 (set_value_range_to_truthvalue): Likewise.
7268 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7269 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7270 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7271 (find_case_label_ranges, test_for_singularity): Likewise.
7272 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7273 (x_vr_values): Move to its remaining use site.
7275 2017-11-10 Jeff Law <law@redhat.com>
7277 * vr-values.h (VR_INITIALIZER): Move #define here.
7278 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7279 * Makefile.in (OBJS): Add tree-evrp.o
7280 * tree-vrp.h (assert_info): Move structure definition here.
7281 (set_value_range_to_varying): Prototype.
7282 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7283 (infer_value_range, register_edge_assert_for): Likewise.
7284 (stmt_interesting_for_vrp): Likewise.
7285 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7286 (set_value_range_to_varying): No longer static.
7287 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7288 (infer_value_range, register_edge_assert_for): Likewise.
7290 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7292 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7293 generate the XXBRD instruction.
7295 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7296 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7297 as being compatible if -mabi=ieeelongdouble.
7298 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7299 to setup float128 built-ins with hardware support.
7300 (BU_FLOAT128_HW_2): Likewise.
7301 (BU_FLOAT128_HW_3): Likewise.
7302 (BU_FLOAT128_HW_VSX_1): Likewise.
7303 (BU_FLOAT128_HW_VSX_2): Likewise.
7304 (scalar_extract_expq): Change float128 built-in functions to
7305 accommodate having both KFmode and TFmode functions. Use the
7306 KFmode variant as the default.
7307 (scalar_extract_sigq): Likewise.
7308 (scalar_test_neg_qp): Likewise.
7309 (scalar_insert_exp_q): Likewise.
7310 (scalar_insert_exp_qp): Likewise.
7311 (scalar_test_data_class_qp): Likewise.
7312 (sqrtf128_round_to_odd): Delete processing the round to odd
7313 built-in functions as special built-in functions, and define them
7314 as float128 built-ins. Use the KFmode variant as the default.
7315 (truncf128_round_to_odd): Likewise.
7316 (addf128_round_to_odd): Likewise.
7317 (subf128_round_to_odd): Likewise.
7318 (mulf128_round_to_odd): Likewise.
7319 (divf128_round_to_odd): Likewise.
7320 (fmaf128_round_to_odd): Likewise.
7321 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7322 support for KFmode and TFmode xststdcqp calls.
7323 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7324 point, switch the built-in handlers for the get/set float128
7325 exponent, get float128 mantissa, float128 test built-ins, and the
7326 float128 round to odd built-in functions. Eliminate creating the
7327 float128 round to odd built-in functions as special built-ins.
7328 (rs6000_init_builtins): Eliminate special creation of the float128
7329 round to odd built-in functions.
7330 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7331 function insns to support both TFmode and KFmode variants.
7332 (xsxsigqp_<mode>): Likewise.
7333 (xsiexpqpf_<mode>): Likewise.
7334 (xsiexpqp_<mode>): Likewise.
7335 (xststdcqp_<mode>): Likewise.
7336 (xststdcnegqp_<mode>): Likewise.
7337 (xststdcqp_<mode>): Likewise.
7339 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7341 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7342 unused path_in_freq_ptr parameter.
7343 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7345 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7347 * ipa-inline.c (edge_badness): Dump sreal frequency.
7348 (compute_inlined_call_time): Match natural implementaiton ...
7349 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7350 forgotten division by CGRAPH_FREQ_BASE.
7352 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7354 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7355 Solaris 11. Update comment.
7356 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7358 * config/sol2.h (STARTFILE_SPEC): Likewise.
7359 * configure: Regenerate.
7361 2017-11-14 Carl Love <cel@us.ibm.com>
7363 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7364 le_ and be_ prefixes to swap* variables. Remove
7365 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7368 2017-11-14 Jason Merrill <jason@redhat.com>
7370 Support GTY((cache)) on hash_map.
7371 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7372 (ggc_cache_remove): Override it instead of ggc_mx.
7373 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7374 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7375 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7376 (simple_cache_map_traits): Override maybe_mx.
7377 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7378 (hash_map): Friend gt_cleare_cache.
7379 (gt_cleare_cache): New.
7380 * tree.h (tree_cache_traits): New hash_map traits class.
7381 (tree_cache_map): New typedef.
7383 2017-11-14 Richard Biener <rguenther@suse.de>
7385 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7386 paramter and handling.
7387 (cleanup_control_flow_bb): Likewise.
7388 (cleanup_control_flow_pre): New helper performing a DFS walk
7389 to call cleanup_control_flow_bb in PRE order.
7390 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7391 via cleanup_control_flow_pre.
7393 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7395 * config/aarch64/aarch64-simd.md
7396 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7397 (*aarch64_simd_bsl<mode>_alt): Likewise.
7398 (aarch64_simd_bsldi_internal): New.
7399 (aarch64_simd_bsldi_alt): Likewise.
7401 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7403 * tracer.c (better_p): Do not compare frequencies.
7404 * reg-stack.c (better_edge): Likewise.
7405 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7408 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7410 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7411 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7412 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7413 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7414 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7415 * predict.c (maybe_hot_frequency_p): Inline to ...
7416 (maybe_hot_count_p): ... here; rewrite to counts.
7417 (counts_to_freqs): Rename to ...
7418 (update_max_bb_count): ... this one.
7419 (expensive_function_p): Use counts.
7420 (estimate_bb_frequencies): Update.
7421 (rebuild_frequencies): Update.
7422 * predict.h (counts_to_freqs): Rename to ...
7423 (update_max_bb_count): ... this one.
7424 * profile.c (compute_branch_probabilities): Add debug info
7425 * tree-inline.c (expand_call_inline): Update debug info.
7426 (optimize_inline_calls): Use update_max_bb_count..
7427 (tree_function_versioning): Use update_max_bb_count..
7428 * value-prof.c (gimple_value_profile_transformations):
7429 Do not use update_max_bb_count.
7431 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7433 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7434 always use frequencies.
7436 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7438 * bb-reorder.c: Remove frequencies from comments.
7439 (better_edge_p): Use profile counts.
7440 (find_traces): Dump profile counts.
7441 (rotate_loop): Use profile counts.
7442 (find_traces_1_round): Likewise.
7443 (connect_better_edge_p): Use counts instead of probabilities for
7445 (copy_bb_p): Drop early check for non-0 frequency.
7446 (sanitize_hot_paths): Update comments.
7448 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7450 * ipa-split.c (struct split_point): Add count.
7451 (consider_split): Do not compute incoming frequency; compute incoming
7452 count and store it to split_point.
7453 (split_function): Set count of the call to split part correctly.
7455 2017-11-13 Carl Love <cel@us.ibm.com>
7457 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7459 2017-11-13 Tom Tromey <tom@tromey.com>
7461 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7463 2017-11-13 Carl Love <cel@us.ibm.com>
7465 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7466 Add support for builtins:
7467 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7468 vector {un,}signed {char,int,short},
7469 vector {un,}signed {char,int,short}) arguments.
7470 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7471 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7472 Add BU_P9V_AV_2 expansions for the builtins.
7473 * config/rs6000/altivec.h (vec_first_match_index,
7474 vec_first_mismatch_index, vec_first_match_or_eos_index,
7475 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7476 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7477 new extern declaration.
7478 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7479 * config/rs6000/vsx.md (first_match_index_<mode>,
7480 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7481 first_mismatch_or_eos_index_<mode>): Add define expand.
7482 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7483 * doc/extend.texi: Update the built-in documenation file for the new
7486 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7488 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7489 into the min/max operations for _Float<N> and _Float<N>X types.
7491 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7494 * dwarf2out.c (do_eh_frame): New static variable.
7495 (dwarf2out_begin_prologue): Set it.
7496 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7498 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7500 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7502 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7504 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7507 * debug.h (dwarf2out_do_eh_frame): Declare.
7508 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7509 (dwarf2out_do_frame): Use it.
7510 (dwarf2out_do_cfi_asm): Likewise.
7511 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7512 (dwarf2out_assembly_start): Likewise.
7513 (dwarf2out_begin_prologue): Fix comment.
7514 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7515 if the target needs either debug or unwind DWARF2 info.
7516 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7517 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7518 -ftrapping-math, -ftrapv and -fwrapv.
7520 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7522 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7523 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7524 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7525 (estimate_edge_size_and_time): Likewise.
7526 (ipa_merge_fn_summary_after_inlining): Likewise.
7527 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7528 (compute_uninlined_call_time): Use sreal_frequency.
7529 (compute_inlined_call_time): Likewise.
7530 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7531 * profile-count.c: Include sreal.h.
7532 (profile_count::to_sreal_scale): New.
7533 * profile-count.h: Forward declare sreal.
7534 (profile_count::to_sreal_scale): Declare.
7536 2017-11-13 Nathan Sidwell <nathan@acm.org>
7538 * diagnostic.c (maybe_line_and_column): New.
7539 (diagnostic_get_location_text): Use it.
7540 (diagnostic_report_current_module): Likewise.
7541 (test_diagnostic_get_location_text): Add tests.
7543 2017-11-13 Luis Machado <luis.machado@linaro.org>
7545 * doc/md.texi (Specifying processor pipeline description): Fix
7546 incorrect latency for the div instruction example.
7548 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7550 PR tree-optimization/78821
7551 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7552 that bit_not_p is the same.
7553 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7554 (split_group): Count precisely bit_not_p bits in each statement.
7555 (invert_op): New function.
7556 (imm_store_chain_info::output_merged_store): Use invert_op to
7557 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7558 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7560 2017-11-13 Martin Liska <mliska@suse.cz>
7562 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7563 (struct source_info): Likewise.
7564 (add_branch_counts): Likewise.
7565 (add_line_counts): Likewise.
7566 (function_summary): Likewise.
7567 (output_intermediate_line): Likewise.
7568 (generate_results): Likewise.
7570 2017-11-13 Martin Liska <mliska@suse.cz>
7572 * gcov.c (struct block_info): Remove typedef for block_t.
7573 (struct line_info): Likewise.
7574 (line_info::has_block): Likewise.
7575 (EXIT_BLOCK): Likewise.
7576 (unblock): Likewise.
7577 (circuit): Likewise.
7578 (get_cycles_count): Likewise.
7579 (process_file): Likewise.
7580 (read_graph_file): Likewise.
7581 (solve_flow_graph): Likewise.
7582 (find_exception_blocks): Likewise.
7583 (add_line_counts): Likewise.
7584 (accumulate_line_info): Likewise.
7585 (output_line_details): Likewise.
7587 2017-11-13 Martin Liska <mliska@suse.cz>
7589 * gcov.c (struct arc_info): Remove typedef for arc_t.
7590 (struct line_info): Likewise.
7591 (add_branch_counts): Likewise.
7592 (output_branch_count): Likewise.
7593 (function_info::~function_info): Likewise.
7594 (circuit): Likewise.
7595 (output_intermediate_line): Likewise.
7596 (read_graph_file): Likewise.
7597 (solve_flow_graph): Likewise.
7598 (find_exception_blocks): Likewise.
7599 (add_line_counts): Likewise.
7600 (accumulate_line_info): Likewise.
7601 (output_line_details): Likewise.
7602 (output_function_details): Likewise.
7604 2017-11-13 Martin Liska <mliska@suse.cz>
7606 * gcov.c (struct function_info): Remove typedef for function_t.
7607 (struct source_info): Likewise.
7608 (source_info::get_functions_at_location): Likewise.
7609 (solve_flow_graph): Likewise.
7610 (find_exception_blocks): Likewise.
7611 (add_line_counts): Likewise.
7612 (output_intermediate_file): Likewise.
7613 (process_file): Likewise.
7614 (generate_results): Likewise.
7615 (release_structures): Likewise.
7616 (read_graph_file): Likewise.
7617 (read_count_file): Likewise.
7618 (accumulate_line_counts): Likewise.
7619 (output_lines): Likewise.
7621 2017-11-13 Martin Liska <mliska@suse.cz>
7623 * gcov.c (function_info::function_info): Remove num_counts
7624 and add vector<gcov_type>.
7625 (function_info::~function_info): Use the vector.
7626 (process_file): Likewise.
7627 (read_graph_file): Likewise.
7628 (read_count_file): Likewise.
7629 (solve_flow_graph): Likewise.
7631 2017-11-13 Martin Liska <mliska@suse.cz>
7633 * gcov.c (function_info::is_artificial): New function.
7634 (process_file): Erase all artificial early.
7635 (generate_results): Skip as all artificial are already
7638 2017-11-13 Martin Liska <mliska@suse.cz>
7640 * gcov.c (read_graph_file): Store to global vector of functions.
7641 (read_count_file): Iterate the vector.
7642 (process_file): Likewise.
7643 (generate_results): Likewise.
7644 (release_structures): Likewise.
7646 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7648 PR tree-optimization/82954
7649 * gimple-ssa-store-merging.c
7650 (imm_store_chain_info::coalesce_immediate_stores): If
7651 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7653 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7655 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7656 Upddate call to ENDIAN_LANE_N.
7657 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7658 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7659 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7660 and use aarch64_endian_lane_rtx.
7661 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7663 2017-11-12 Tom de Vries <tom@codesourcery.com>
7665 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7667 2017-11-12 Tom de Vries <tom@codesourcery.com>
7669 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7671 2017-11-12 Tom de Vries <tom@codesourcery.com>
7673 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7675 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7676 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7677 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7679 2017-11-11 Martin Sebor <msebor@redhat.com>
7682 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7685 * prefic.c (translate_name): Replace strncpy with memcpy to
7686 avoid -Wstringop-truncation.
7688 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7690 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7692 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7694 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7696 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7698 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7701 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7703 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7704 enable generating XXBRH if the value is in a vector register.
7705 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7706 value is in a vector register.
7707 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7708 register to register bswap64's instead of doing the GPR sequence
7709 used on previous machines.
7710 (bswapdi2_xxbrd): New insn.
7711 (bswapdi2_reg): Disallow on ISA 3.0.
7712 (register to register bswap64 splitter): Do not split the insn on
7713 ISA 3.0 systems that use XXBRD.
7715 2017-11-10 Martin Sebor <msebor@redhat.com>
7718 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7720 (find_subframework_file): Same.
7722 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7724 * auto-profile.c (afdo_indirect_call): Drop frequency.
7725 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7726 (cgraph_node::create_edge): Drop frequency argument.
7727 (cgraph_node::create_indirect_edge): Drop frequency argument.
7728 (cgraph_edge::make_speculative): Drop frequency arguments.
7729 (cgraph_edge::resolve_speculation): Do not update frequencies
7730 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7731 (cgraph_node::dump): Check consistency in IPA mode.
7732 (cgraph_edge::maybe_hot_p): Use IPA counter.
7733 (cgraph_edge::verify_count_and_frequency): Rename to ...
7734 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7735 (cgraph_node::verify_node): Update.
7736 * cgraph.h (struct cgraph_edge): Drop frequency.
7737 (cgraph_edge::frequency): New function.
7738 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7740 (cgraph_edge::rebuild_edges): Likewise.
7741 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7742 (duplicate_thunk_for_node): Do not pass frequency.
7743 (cgraph_node::create_clone): Scale only counts.
7744 (cgraph_node::create_virtual_clone): Do not pass frequency.
7745 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7746 (cgraph_node::create_version_clone): Do not pass frequency.
7747 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7748 (cgraph_node::expand_thunk): Do not pass frequency.
7749 (cgraph_node::create_wrapper): Do not pass frequency.
7750 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7752 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7753 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7754 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7755 (gather_caller_stats): Use frequency function.
7756 (ipcp_cloning_candidate_p): Use frequency function.
7757 (ipcp_propagate_stage): Use frequency function.
7758 (get_info_about_necessary_edges): Use frequency function.
7759 (update_profiling_info): Update only IPA profile.
7760 (update_specialized_profile): Use frequency functoin.
7761 (perhaps_add_new_callers): Update only IPA profile.
7762 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7763 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7764 (dump_ipa_call_summary): Use frequency function.
7765 (estimate_edge_size_and_time): Use frequency function.
7766 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7767 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7768 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7769 (update_noncloned_counts): ... ths one; scale counts only.
7770 (clone_inlined_nodes): Do not scale frequency.
7771 (inline_call): Do not pass frequency.
7772 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7773 (compute_inlined_call_time): Use IPA profile.
7774 (want_inline_small_function_p): Use IPA profile.
7775 (want_inline_self_recursive_call_p): Use IPA profile.
7776 (edge_badness): Use IPA profile.
7777 (lookup_recursive_calls): Use IPA profile.
7778 (recursive_inlining): Do not pass frequency.
7779 (resolve_noninline_speculation): Do not update frequency.
7780 (inline_small_functions): Collect max of IPA profile.
7781 (dump_overall_stats): Dump IPA porfile.
7782 (dump_inline_stats): Dump IPA porfile.
7783 (ipa_inline): Collect IPA stats.
7784 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7785 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7786 (ipa_propagate_frequency): Use frequency function.
7787 (ipa_profile): Cleanup.
7788 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7789 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7790 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7791 (input_node): Do not stream frequency.
7792 (input_edge): Do not stream frequency.
7793 (merge_profile_summaries): Scale only IPA profiles.
7794 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7795 * predict.c (drop_profile): Do not recompute frequency.
7796 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7797 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7798 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
7799 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
7800 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
7801 * tree-inline.c (copy_bb): Do not scale frequency.
7802 (expand_call_inline): Do not scale frequency.
7803 (tree_function_versioning): Do not scale frequency.
7804 * ubsan.c (ubsan_create_edge): Do not pass frequency.
7806 2017-11-10 Julia Koval <julia.koval@intel.com>
7808 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
7809 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
7810 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
7811 (_mm256_maskz_gf2p8affine_epi64_epi8)
7812 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
7813 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
7814 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
7815 (__builtin_ia32_vgf2p8affineqb_v32qi)
7816 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
7817 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
7819 2017-11-10 Tamar Christina <tamar.christina@arm.com>
7823 (arm_option_override): Refactor.
7824 (arm_option_reconfigure_globals): New.
7825 (arm_options_perform_arch_sanity_checks): New.
7826 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7828 (arm_options_perform_arch_sanity_checks): Likewise
7830 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7832 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7833 (power9-qpmul): New.
7834 * rs6000/rs6000.md ("type" attr): Add qmul.
7835 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7836 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7837 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7839 2017-11-10 Martin Sebor <msebor@redhat.com>
7842 * builtins.c (compute_objsize): Handle arrays that
7843 compute_builtin_object_size likes to fail for. Make extern.
7844 * builtins.h (compute_objsize): Declare.
7845 (check_strncpy_sizes): New function.
7846 (expand_builtin_strncpy): Call check_strncpy_sizes.
7847 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7848 -Wstringop-truncation.
7849 (gimple_fold_builtin_strncat): Same.
7850 * gimple.c (gimple_build_call_from_tree): Set call location.
7851 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
7852 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
7853 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
7854 (handle_builtin_strlen): Use strlen_to_stridx.
7855 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
7857 Use strlen_to_stridx.
7858 (pass_strlen::execute): Release strlen_to_stridx.
7859 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7860 (-Wstringop-truncation): Document new option.
7862 2017-11-10 Martin Liska <mliska@suse.cz>
7864 PR gcov-profile/82702
7865 * gcov.c (main): Handle intermediate files in a different
7867 (get_gcov_intermediate_filename): New function.
7868 (output_gcov_file): Remove support of intermediate files.
7869 (generate_results): Allocate intermediate file.
7870 (release_structures): Clean-up properly fn_end.
7871 (output_intermediate_file): Start iterating with line 1.
7873 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7875 PR tree-optimization/82929
7876 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7877 ops_swapped_p non-static data member.
7878 (store_immediate_info::store_immediate_info): Clear it.
7879 (imm_store_chain_info::coalesce_immediate_stores): If swapping
7880 ops set ops_swapped_p.
7881 (count_multiple_uses): Handle ops_swapped_p.
7883 2017-11-10 Martin Liska <mliska@suse.cz>
7885 * coverage.c (coverage_init): Stream information about
7886 support of has_unexecuted_blocks.
7887 * doc/gcov.texi: Document that.
7888 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7889 * gcov.c (read_graph_file): Likewise.
7890 (output_line_beginning): Fix a small issue with
7893 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7895 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7896 reference of trivial component.
7898 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7901 * gimple-ssa-store-merging.c
7902 (pass_store_merging::terminate_all_aliasing_chains): For
7903 gimple_store_p stmts also call refs_output_dependent_p.
7905 PR rtl-optimization/82913
7906 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7909 2017-11-09 Jeff Law <law@redhat.com>
7911 * vr-values.h: New file with vr_values class.
7912 * tree-vrp.c: Include vr-values.h
7913 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7914 data objects into the vr_values class.
7915 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7916 (get_value_range): Make it a member function within vr_values class.
7917 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7918 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7919 (op_with_constant_singleton_value_range): Likewise.
7920 (extract_range_for_var_from_comparison_expr): Likewise.
7921 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7922 (extract_range_from_binary_expr): Likewise.
7923 (extract_range_from_unary_expr): Likewise.
7924 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7925 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7926 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7927 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7928 (compare_name_with_value, compare_names): Likewise.
7929 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7930 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7931 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7932 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7933 (extract_range_from_phi_node): Likewise.
7934 (simplify_truth_ops_using_ranges): Likewise.
7935 (simplify_div_or_mod_using_ranges): Likewise.
7936 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7937 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7938 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7939 (simplify_float_conversion_using_ranges): Likewise.
7940 (simplify_internal_call_using_ranges): Likewise.
7941 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7942 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7943 poitner via x_vr_values for calls into gimple folder.
7944 (vrp_initialize_lattice): Make this the vr_values ctor.
7945 (vrp_free_lattice): Make this the vr_values dtor.
7946 (set_vr_value): New function.
7947 (class vrp_prop): Add vr_values data member. Add various member
7948 functions as well as member functions that delegate to vr_values.
7949 (check_array_ref): Make a member function within vrp_prop class.
7950 (search_for_addr_array, vrp_initialize): Likewise.
7951 (vrp_finalize): Likewise. Revamp to avoid direct access to
7952 vr_value, values_propagated, etc.
7953 (check_array_bounds): Extract vrp_prop class instance pointer from
7954 walk info structure. Use it to call member functions.
7955 (check_all_array_refs): Make a member function within vrp_prop class.
7956 Smuggle class instance pointer via walk info structure.
7957 (x_vr_values): New local static.
7958 (vrp_valueize): Use x_vr_values to get class instance.
7959 (vr_valueize_1): Likewise.
7960 (class vrp_folder): Add vr_values data member. Add various member
7961 functions as well as member functions that delegate to vr_values.
7962 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7963 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7964 class instance from vr_values. Use it to call member functions.
7965 (vrp_dom_walker): Add vr_values data member.
7966 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7967 instance via x_vr_values.
7968 (identify_jump_threads): Accept vr_values as argument. Store
7969 it into the walker structure.
7970 (evrp_dom_walker): Add vr_values class data member. Add various
7972 (evrp_dom_walker::try_find_new_range): Use vr_values data
7973 member to access the memory allocator.
7974 (evrp_dom_walker::before_dom_children): Store vr_values class
7975 instance into the vrp_folder class.
7976 (evrp_dom_walker::push_value_range): Rework to avoid direct
7977 access to num_vr_values and vr_value.
7978 (evrp_dom_walker::pop_value_range): Likewise.
7979 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7980 Use vr_values to get to dump_all_value_ranges member function.
7981 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7982 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7983 Pass vr_values class instance down to identify_jump_threads.
7984 Remove call to vrp_free_lattice.
7985 (debug_all_value_ranges): Remove.
7987 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7988 (vrp_folder): Likewise.
7990 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7991 Get it from the existing bitmap instead.
7992 (vrp_intersect_ranges_1): Likewise.
7994 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7996 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7998 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7999 set bit_not_p to it.
8000 (imm_store_chain_info::coalesce_immediate_stores): Break group
8001 if bit_not_p is different.
8002 (count_multiple_uses, split_group,
8003 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
8004 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
8005 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
8006 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
8007 to store_immediate_info ctor.
8009 2017-11-09 Jim Wilson <jimw@sifive.com>
8011 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8012 (scan_prog_file): Likewise.
8014 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8016 * bb-reorder.c (max_entry_frequency): Remove.
8017 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8018 connect_traces, push_to_next_round_p): Remove prototypes.
8019 (find_traces_1_round): Use counts only.
8020 (push_to_next_round_p): Likewise.
8021 (find_traces): Likewise.
8022 (rotate_loop): Likewise.
8023 (find_traces_1_round): Likewise.
8024 (connect_traces): Likewise.
8025 (edge_order): Likewise.
8027 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8029 * config/arm/arm.c (output_return_instruction): Add comments to
8030 indicate requirement for cmse_nonsecure_entry return to account
8031 for the size of clearing instruction output here.
8032 (thumb_exit): Likewise.
8033 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8034 return in hardfloat mode.
8036 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8038 * config/rs6000/rs6000.c (machine_function): Add a bool,
8039 "toc_is_wrapped_separately".
8040 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8041 if it wasn't explicitly set or unset, we are optimizing for speed, and
8042 doing separate shrink-wrapping.
8043 (rs6000_get_separate_components): Enable the TOC component if
8044 saving the TOC register in the prologue.
8045 (rs6000_components_for_bb): Handle the TOC component.
8046 (rs6000_emit_prologue_components): Store the TOC register where needed.
8047 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8048 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8050 2017-11-09 Martin Jambor <mjambor@suse.cz>
8052 * ipa-param-manipulation.c: New file.
8053 * ipa-param-manipulation.h: Likewise.
8054 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8055 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8056 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8057 (ipa_parm_adjustment): Likewise.
8058 (ipa_parm_adjustment_vec): Likewise.
8059 (ipa_get_vector_of_formal_parms): Moved declaration to
8060 ipa-param-manipulation.h.
8061 (ipa_get_vector_of_formal_parm_types): Likewise.
8062 (ipa_modify_formal_parameters): Likewise.
8063 (ipa_modify_call_arguments): Likewise.
8064 (ipa_combine_adjustments): Likewise.
8065 (ipa_dump_param_adjustments): Likewise.
8066 (ipa_modify_expr): Likewise.
8067 (ipa_get_adjustment_candidate): Likewise.
8068 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8069 ipa-param-manipulation.c.
8070 (ipa_get_vector_of_formal_parm_types): Likewise.
8071 (ipa_modify_formal_parameters): Likewise.
8072 (ipa_modify_call_arguments): Likewise.
8073 (ipa_modify_expr): Likewise.
8074 (get_ssa_base_param): Likewise.
8075 (ipa_get_adjustment_candidate): Likewise.
8076 (index_in_adjustments_multiple_times_p): Likewise.
8077 (ipa_combine_adjustments): Likewise.
8078 (ipa_dump_param_adjustments): Likewise.
8079 * tree-sra.c: Also include ipa-param-manipulation.h
8080 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8083 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8084 Alan Hayward <alan.hayward@arm.com>
8085 David Sherwood <david.sherwood@arm.com>
8087 * doc/sourcebuild.texi (vect_masked_store): Document.
8089 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8090 Alan Hayward <alan.hayward@arm.com>
8091 David Sherwood <david.sherwood@arm.com>
8093 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8095 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8096 Alan Hayward <alan.hayward@arm.com>
8097 David Sherwood <david.sherwood@arm.com>
8099 * doc/sourcebuild.texi (vect_variable_length): Document.
8101 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8102 Alan Hayward <alan.hayward@arm.com>
8103 David Sherwood <david.sherwood@arm.com>
8105 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8107 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8108 Alan Hayward <alan.hayward@arm.com>
8109 David Sherwood <david.sherwood@arm.com>
8111 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8113 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8114 Alan Hayward <alan.hayward@arm.com>
8115 David Sherwood <david.sherwood@arm.com>
8117 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8118 previously undocumented selectors.
8119 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8121 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8123 * doc/rtl.texi (const_vector): Say that elements can be
8124 const_wide_ints too.
8125 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8126 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8127 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8128 * optabs.c (expand_vector_broadcast): Likewise.
8130 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8131 Alan Hayward <alan.hayward@arm.com>
8132 David Sherwood <david.sherwood@arm.com>
8134 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8135 scaled index even if the unscaled address was invalid.
8136 Don't increase the complexity of using a scale in that case.
8138 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8139 Alan Hayward <alan.hayward@arm.com>
8140 David Sherwood <david.sherwood@arm.com>
8142 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8143 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8144 UNITS_PER_WORD bytes.
8145 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8146 based on REGMODE_NATURAL_SIZE of the inner mode.
8147 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8148 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8149 * expmed.c (lowpart_bit_field_p): Divide the value up into
8150 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8151 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8152 whether something is likely to occupy more than one register.
8154 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8157 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8158 profile_count::adjust_for_ipa_scaling.
8159 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8160 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8162 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8164 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8166 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8167 (split_group): Add total_orig and total_new arguments, estimate the
8168 number of statements related to the store group without store merging
8169 and with store merging.
8170 (imm_store_chain_info::output_merged_store): Adjust split_group
8171 callers, punt if estimated number of statements with store merging
8172 is not smaller than estimated number of statements without it.
8174 (handled_load): Remove has_single_use checks.
8175 (pass_store_merging::process_store): Likewise.
8177 2017-11-09 Richard Biener <rguenther@suse.de>
8179 PR tree-optimization/82902
8180 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8182 2017-11-09 Martin Liska <mliska@suse.cz>
8185 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8188 2017-11-09 Martin Liska <mliska@suse.cz>
8190 PR tree-optimization/82669
8191 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8193 2017-11-09 Martin Liska <mliska@suse.cz>
8195 PR gcov-profile/48463
8196 * coverage.c (coverage_begin_function): Output also end locus
8197 of a function and information whether the function is
8199 * gcov-dump.c (tag_function): Parse and print the information.
8200 * gcov.c (INCLUDE_MAP): Add include.
8201 (INCLUDE_SET): Likewise.
8202 (struct line_info): Move earlier in the source file because
8203 of vector<line_info> in function_info structure.
8204 (line_info::line_info): Likewise.
8205 (line_info::has_block): Likewise.
8206 (struct source_info): Add new member index.
8207 (source_info::get_functions_at_location): New function.
8208 (function_info::group_line_p): New function.
8209 (output_intermediate_line): New function.
8210 (output_intermediate_file): Use the mentioned function.
8211 (struct function_start): New.
8212 (struct function_start_pair_hash): Likewise.
8213 (process_file): Add code that identifies group functions.
8214 Assign lines either to global or function scope.
8215 (generate_results): Skip artificial functions.
8216 (find_source): Assign index for each source file.
8217 (read_graph_file): Read new flag artificial and end_line.
8218 (add_line_counts): Assign it either to global of function scope.
8219 (accumulate_line_counts): Isolate core of the function to
8220 accumulate_line_info and call it for both function and global
8222 (accumulate_line_info): New function.
8223 (output_line_beginning): Fix GNU coding style.
8224 (print_source_line): New function.
8225 (output_line_details): Likewise.
8226 (output_function_details): Likewise.
8227 (output_lines): Iterate both source (global) scope and function
8229 (struct function_line_start_cmp): New class.
8230 * doc/gcov.texi: Reflect changes in documentation.
8232 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8235 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8236 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8237 and similarly for not instead of neg.
8239 2017-11-08 Andi Kleen <ak@linux.intel.com>
8241 * config/i386/i386.opt: Add -mforce-indirect-call.
8242 * config/i386/predicates.md: Check for flag_force_indirect_call.
8243 * doc/invoke.texi: Document -mforce-indirect-call
8245 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8247 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8249 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8250 config/riscv/riscv.c (predict.h): New include.
8251 (riscv_slow_unaligned_access_p): No longer static.
8252 (riscv_block_move_straight): Add require.
8253 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8255 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8258 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8259 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8260 nonimmediate_operand predicate for operand 1 instead of
8263 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8265 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8267 * config/aarch64/constraints.md (Uml): New constraint.
8268 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8271 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8273 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8274 of two vec_duplicates into a vec_concat.
8276 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8278 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8279 Simplify vec_merge of vec_duplicate and vec_concat.
8280 * config/aarch64/constraints.md (Utq): New constraint.
8281 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8284 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8286 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8287 Simplify vec_merge of vec_duplicate and const_vector.
8288 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8290 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8291 mode iterator. Update predicate on operand 1 to
8292 handle non-const_vec constants. Delete constraints.
8293 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8295 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8297 PR tree-optimization/78821
8298 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8300 (store_operand_info::store_operand_info): Initialize it to false.
8301 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8302 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8303 store in the group, and if chain_info is non-NULL, to ignore altogether
8305 (compatible_load_p): Fail if bit_not_p does not match.
8306 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8307 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8308 (pass_store_merging::process_store): Adjust
8309 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8310 call terminate_all_aliasing_chains newly when adding a store into
8311 a chain with non-NULL chain_info.
8313 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8315 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8317 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8319 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8321 (aarch64_layout_frame): Initialise emit_frame_chain.
8322 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8323 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8325 2017-11-08 Martin Liska <mliska@suse.cz>
8327 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8328 of gimple_call_internal_p.
8330 2017-11-07 Tom de Vries <tom@codesourcery.com>
8332 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8334 2017-11-07 Tom de Vries <tom@codesourcery.com>
8336 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8339 2017-11-08 Martin Liska <mliska@suse.cz>
8342 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8344 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8346 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8348 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8351 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8352 (*cmp<mode>_ccz_1): New insn with $k alternative.
8355 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8356 RTX_COMM_COMPARE as commutative as well.
8357 (ix86_binary_operator_ok): Formatting fix.
8358 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8359 *<code><mode>3<mask_name><round_saeonly_name>,
8360 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8361 *<s>mul<mode>3_highpart<mask_name>,
8362 *vec_widen_umult_even_v16si<mask_name>,
8363 *vec_widen_umult_even_v8si<mask_name>,
8364 *vec_widen_umult_even_v4si<mask_name>,
8365 *vec_widen_smult_even_v16si<mask_name>,
8366 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8367 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8368 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8369 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8370 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8371 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8372 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8373 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8374 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8375 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8376 ix86_binary_operator_ok. Formatting fixes.
8377 (*<plusminus_insn><mode>3<mask_name><round_name>,
8378 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8381 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8383 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8384 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8385 eq and ne if TARGET_ISEL.
8386 (cmp): New code_iterator.
8387 (UNS, UNSU_, UNSIK): New code_attrs.
8388 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8389 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8391 ("eq<mode>3"): ... this.
8392 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8394 ("ne<mode>3"): ... this.
8396 2017-11-07 Julia Koval <julia.koval@intel.com>
8399 * common/config/i386/i386-common.c
8400 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8401 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8402 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8403 * config/i386/i386.opt: Ditto.
8404 * config/i386/i386.c (ix86_target_string): Ditto.
8405 (ix86_option_override_internal): Ditto.
8406 (ix86_init_mpx_builtins): Move MPX to args2.
8407 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8408 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8409 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8410 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8411 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8412 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8413 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8415 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8418 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8419 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8420 (zero-extendsidi peephole2): Remove peephole.
8422 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8425 * stor-layout.c: Include attribs.h.
8426 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8427 explicit lookup of "aligned" attribute.
8429 2017-11-07 Andrew Waterman <andrew@sifive.com>
8431 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8432 (riscv_expand_block_move): Likewise.
8433 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8435 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8436 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8437 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8438 (riscv_adjust_block_mem): Likewise.
8439 (riscv_block_move_loop): Likewise.
8440 (riscv_expand_block_move): Likewise.
8441 * config/riscv/riscv.md (movmemsi): New pattern.
8443 2017-11-07 Michael Clark <michaeljclark@mac.com>
8445 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8446 (MUSL_DYNAMIC_LINKER): Likewise.
8448 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8450 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8451 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8454 2017-11-07 Richard Biener <rguenther@suse.de>
8456 * match.pd: Fix build.
8458 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8459 Jackson Woodruff <jackson.woodruff@arm.com>
8461 PR tree-optimization/71026
8462 * match.pd: Canonicalize negate in division.
8464 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8467 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8469 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8471 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8472 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8474 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8476 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8477 non-scalar integral types.
8478 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8479 (-(A-B), -(~A)): New transformations.
8481 2017-11-07 Tom de Vries <tom@codesourcery.com>
8483 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8484 semicolon after "do {} while (0)".
8485 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8486 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8487 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8488 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8489 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8491 2017-11-07 Tom de Vries <tom@codesourcery.com>
8493 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8494 after "do {} while (0)".
8495 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8496 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8497 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8498 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8499 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8501 2017-11-07 Tom de Vries <tom@codesourcery.com>
8504 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8506 (arm_rtx_costs_internal): Add missing semicolon after
8507 HANDLE_NARROW_SHIFT_ARITH call.
8509 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8511 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8512 disable isel if it was not set explicitly.
8514 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8516 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8517 (add_type_attribute) likewise.
8519 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8521 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8522 of struct ix86_frame.
8523 (ix86_initial_elimination_offset): Likewise.
8524 (ix86_expand_split_stack_prologue): Likewise.
8526 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8528 * tree-vrp.h (enum value_range_type): Update stale comment.
8530 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8531 Alan Hayward <alan.hayward@arm.com>
8532 David Sherwood <david.sherwood@arm.com>
8534 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8535 (aarch64_expand_vec_perm_const): Take the number of units too.
8536 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8537 (aarch64_expand_vec_perm_const): Likewise.
8538 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8539 (vec_perm<mode>): Update accordingly.
8541 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8542 Alan Hayward <alan.hayward@arm.com>
8543 David Sherwood <david.sherwood@arm.com>
8545 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8546 Take the number of units too.
8547 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8548 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8549 but check for a vector mode before rather than after the call.
8550 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8551 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8552 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8553 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8554 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8555 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8556 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8557 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8558 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8559 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8560 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8561 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8562 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8563 (aarch64_sqdmull2_n<mode>): Update accordingly.
8565 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8566 Alan Hayward <alan.hayward@arm.com>
8567 David Sherwood <david.sherwood@arm.com>
8569 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8570 the number of units too.
8571 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8572 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8573 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8574 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8575 (vec_store_lanesxi<mode>): Update accordingly.
8577 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8578 Alan Hayward <alan.hayward@arm.com>
8579 David Sherwood <david.sherwood@arm.com>
8581 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8582 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8583 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8584 of units rather than the mode.
8585 * config/aarch64/iterators.md (nunits): New mode attribute.
8586 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8587 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8588 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8589 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8590 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8591 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8592 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8593 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8594 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8595 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8596 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8597 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8598 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8599 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8600 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8601 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8602 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8603 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8604 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8605 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8606 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8607 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8608 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8609 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8610 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8611 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8612 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8613 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8614 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8615 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8616 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8617 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8618 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8619 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8620 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8621 (aarch64_simd_vec_setv2di): Likewise.
8623 2017-11-06 Carl Love <cel@us.ibm.com>
8625 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8627 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8628 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8630 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8632 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8633 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8634 Add power 8 macro expansions.
8635 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8636 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8637 power 8 instructions. (VSX_XXBR): Add iterator.
8639 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8641 * config/arm/arm.md (predicable_short_it): Change default to "no",
8642 improve documentation, remove uses that are identical to the default.
8643 (enabled_for_depr_it): Rename to enabled_for_short_it.
8644 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8645 Remove default uses.
8646 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8647 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8648 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8649 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8651 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8654 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8655 float128 helper macros, which are no longer used after deleting
8656 the old 'q' built-in functions, and moving the round to odd
8657 built-in functions to being special built-in functions.
8658 (BU_FLOAT128_2): Likewise.
8659 (BU_FLOAT128_1_HW): Likewise.
8660 (BU_FLOAT128_2_HW): Likewise.
8661 (BU_FLOAT128_3_HW): Likewise.
8662 (FABSQ): Delete old 'q' built-in functions.
8663 (COPYSIGNQ): Likewise.
8664 (SQRTF128_ODD): Move round to odd built-in functions to be
8665 special built-in functions, so that we can handle
8666 -mabi=ieeelongdouble.
8667 (TRUNCF128_ODD): Likewise.
8668 (ADDF128_ODD): Likewise.
8669 (SUBF128_ODD): Likewise.
8670 (MULF128_ODD): Likewise.
8671 (DIVF128_ODD): Likewise.
8672 (FMAF128_ODD): Likewise.
8673 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8674 built-in names to 'f128'.
8675 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8676 old 'q' built-in functions, as the machine independent code for
8677 'f128' built-in functions handles this.
8678 (rs6000_expand_builtin): Add expansion for float128 round to odd
8679 functions, keying off on -mabi=ieeelongdouble of whether to use
8680 the KFmode or TFmode variant.
8681 (rs6000_init_builtins): Initialize the _Float128 round to odd
8683 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8684 _Float128 'q' built-in functions are now mapped into the new
8685 'f128' built-in functions.
8687 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8689 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8690 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8692 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8695 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8696 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8697 * toplev.c: Include "ipa-fnsummary.h".
8698 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8700 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8702 PR tree-optimization/82838
8703 * gimple-ssa-store-merging.c
8704 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8705 on a separate gimple_seq which is then appended to seq.
8707 2017-11-06 Jeff Law <law@redhat.com>
8710 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8711 (get_probe_interval): New functions.
8712 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8713 (ix86_adjust_stack_and_probe): Likewise.
8714 (output_adjust_stack_and_probe): Likewise.
8715 (ix86_emit_probe_stack_range): Likewise.
8716 (ix86_expand_prologue): Likewise.
8718 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8720 PR tree-optimization/82816
8721 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8722 if the modes of the two types are the same.
8723 (convert_plusminus_to_widen): Likewise.
8725 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8727 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8729 (usadv16qi): New define_expand.
8730 (usadv8hi): New define_expand.
8732 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8735 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8737 (inline_transform): Likewise.
8738 * predict.c (counts_to_freqs): Remove useless conditional.
8739 * profile-count.h (profile_count::apply_scale): Move sanity check.
8740 * tree-inline.c (copy_bb): Always scale.
8741 (copy_cfg_body): Likewise.
8743 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8746 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8748 (*cmp_ite0): Add enabled_for_depr_it attribute.
8749 (*cmp_ite1): Likewise.
8751 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8753 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8756 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8758 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8759 return the value by pointer.
8760 (extract_range_from_multiplicative_op_1): Update accordingly.
8761 Return as soon as an operation fails.
8763 2017-11-05 Tom de Vries <tom@codesourcery.com>
8766 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8767 (DEF_SANITIZER_BUILTIN): ... here.
8768 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8769 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8771 2017-11-05 Tom de Vries <tom@codesourcery.com>
8774 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8776 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8777 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8778 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8780 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8781 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8782 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8784 2017-11-05 Tom de Vries <tom@codesourcery.com>
8787 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8790 2017-11-04 Michael Clark <michaeljclark@mac.com>
8792 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8793 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8795 (*addsi3_extended): Likewise.
8796 (*addsi3_extended2): Likewise.
8797 (<optab>si3): Likewise.
8798 (<optab>di3): Likewise.
8799 (<optab><mode>3): Likewise.
8800 (<*optabe>si3_internal): Likewise.
8801 (zero_extendqi<SUPERQI:mode>2): Likewise.
8802 (*add<mode>hi3): Likewise.
8803 (*xor<mode>hi3): Likewise.
8804 (<optab>di3): Likewise.
8805 (*<optab>si3_extend): Likewise.
8806 (*sge<u>_<X:mode><GPR:mode>): Likewise.
8807 (*slt<u>_<X:mode><GPR:mode>): Likewise.
8808 (*sle<u>_<X:mode><GPR:mode>): Likewise.
8810 2017-11-04 Andrew Waterman <andrew@sifive.com>
8812 * config/riscv/riscv.c (riscv_option_override): Conditionally set
8813 TARGET_STRICT_ALIGN based upon -mtune argument.
8815 2017-11-04 Andrew Waterman <andrew@sifive.com>
8817 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
8819 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
8821 * config/i386/i386.c (choose_basereg): Use optional scratch
8822 register and add assertion.
8823 (x86_emit_outlined_ms2sysv_save): Use scratch register when
8824 needed, and don't allocate stack.
8825 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
8826 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8827 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8829 2017-11-03 Jeff Law <law@redhat.com>
8831 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8832 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8833 to probe at the start of a noreturn function.
8835 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8837 PR tree-optimization/78821
8838 * gimple-ssa-store-merging.c: Update the file comment.
8839 (MAX_STORE_ALIAS_CHECKS): Define.
8840 (struct store_operand_info): New type.
8841 (store_operand_info::store_operand_info): New constructor.
8842 (struct store_immediate_info): Add rhs_code and ops data members.
8843 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8844 and op1r arguments to the ctor, initialize corresponding data members.
8845 (struct merged_store_group): Add load_align_base and load_align
8847 (merged_store_group::merged_store_group): Initialize them.
8848 (merged_store_group::do_merge): Update them.
8849 (merged_store_group::apply_stores): Pick the constant for
8850 encode_tree_to_bitpos from one of the two operands, or skip
8851 encode_tree_to_bitpos if neither operand is a constant.
8852 (class pass_store_merging): Add process_store method decl. Remove
8853 bool argument from terminate_all_aliasing_chains method decl.
8854 (pass_store_merging::terminate_all_aliasing_chains): Remove
8855 var_offset_p argument and corresponding handling.
8856 (stmts_may_clobber_ref_p): New function.
8857 (compatible_load_p): New function.
8858 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
8859 if there is overlap and rhs_code is not INTEGER_CST. For
8860 non-overlapping stores terminate group if rhs is not mergeable.
8861 (get_alias_type_for_stmts): Change first argument from
8862 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8863 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8864 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
8866 (get_location_for_stmts): Change first argument from
8867 auto_vec<gimple *> & to vec<gimple *> &.
8868 (struct split_store): Remove orig_stmts data member, add orig_stores.
8869 (split_store::split_store): Create orig_stores rather than orig_stmts.
8870 (find_constituent_stmts): Renamed to ...
8871 (find_constituent_stores): ... this. Change second argument from
8872 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
8873 to info structures rather than the statements.
8874 (split_group): Rename ALLOW_UNALIGNED argument to
8875 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
8876 it. Adjust find_constituent_stores caller.
8877 (imm_store_chain_info::output_merged_store): Handle rhs_code other
8878 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
8879 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
8880 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
8881 (mem_valid_for_store_merging): New function.
8882 (handled_load): New function.
8883 (pass_store_merging::process_store): New method.
8884 (pass_store_merging::execute): Use process_store method. Adjust
8885 terminate_all_aliasing_chains caller.
8887 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8889 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8890 Return true for more constants, symbols and label references.
8891 (aarch64_valid_floating_const): Remove unused function.
8893 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8896 * config/aarch64/aarch64.c (aarch64_layout_frame):
8897 Undo forcing of LR at bottom of frame.
8899 2017-11-03 Jeff Law <law@redhat.com>
8902 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8903 for int_registers_saved.
8905 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8906 extracted from tree-ssa-dom.c.
8907 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8908 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8909 (record_equivalences_from_incoming_edge): Add additional argument
8910 to single_pred_edge_ignoring_loop_edges call.
8911 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8912 (uncprop_dom_walker::before_dom_children): Add additional argument
8913 to single_pred_edge_ignoring_loop_edges call.
8914 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8915 single_pred_edge_ignoring_loop_edges rather than open coding.
8916 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8918 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8920 * match.pd (-(-A)): Rewrite.
8922 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8924 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8925 (rs6000_emit_int_cmove): New declaration.
8926 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8927 (rs6000_emit_sISEL): Delete.
8928 (rs6000_emit_int_cmove): Make non-static.
8929 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8930 instead of rs6000_emit_sISEL.
8932 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8934 * asan.c (create_cond_insert_point): Maintain profile.
8935 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8937 * basic-block.h (struct basic_block_def): Remove frequency.
8938 (EDGE_FREQUENCY): Use to_frequency
8939 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8941 (find_traces): Update to use to_frequency.
8942 (find_traces_1_round): Likewise; use only IPA counts.
8943 (bb_to_key): Likewise.
8944 (connect_traces): Use IPA counts only.
8945 (copy_bb_p): Update to use to_frequency.
8946 (fix_up_crossing_landing_pad): Likewise.
8947 (sanitize_hot_paths): Likewise.
8948 * bt-load.c (basic_block_freq): Likewise.
8949 * cfg.c (init_flow): Set count_max to uninitialized.
8950 (check_bb_profile): Remove frequencies; check counts.
8951 (dump_bb_info): Do not dump frequencies.
8952 (update_bb_profile_for_threading): Update counts only.
8953 (scale_bbs_frequencies_int): Likewise.
8954 (MAX_SAFE_MULTIPLIER): Remove.
8955 (scale_bbs_frequencies_gcov_type): Update counts only.
8956 (scale_bbs_frequencies_profile_count): Update counts only.
8957 (scale_bbs_frequencies): Update counts only.
8958 * cfg.h (struct control_flow_graph): Add count-max.
8959 (update_bb_profile_for_threading): Update prototype.
8960 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8961 (find_many_sub_basic_blocks): Likewise.
8962 * cfgcleanup.c (try_forward_edges): Likewise.
8963 (try_crossjump_to_edge): Likewise.
8964 * cfgexpand.c (expand_gimple_cond): Likewise.
8965 (expand_gimple_tailcall): Likewise.
8966 (construct_init_block): Likewise.
8967 (construct_exit_block): Likewise.
8968 * cfghooks.c (verify_flow_info): Check consistency of counts.
8969 (dump_bb_for_graph): Do not dump frequencies.
8970 (split_block_1): Do not update frequencies.
8971 (split_edge): Do not update frequencies.
8972 (make_forwarder_block): Do not update frequencies.
8973 (duplicate_block): Do not update frequencies.
8974 (account_profile_record): Do not update frequencies.
8975 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8976 for global heuristics.
8977 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8978 (expected_loop_iterations_unbounded): Use counts only.
8979 * cfgloopmanip.c (scale_loop_profile): Simplify.
8980 (create_empty_loop_on_edge): Simplify
8982 (duplicate_loop_to_header_edge): Simplify
8983 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8984 (update_br_prob_note): Take care of removing note when profile
8986 (relink_block_chain): Do not dump frequency.
8987 (rtl_account_profile_record): Use to_frequency.
8988 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8989 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8990 (cgraph_update_edges_for_call_stmt_node): Likewise.
8991 (cgraph_edge::verify_count_and_frequency): Update.
8992 (cgraph_node::verify_node): Temporarily disable frequency verification.
8993 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8994 to_cgraph_frequency.
8995 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8996 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8998 (cgraph_node::expand_thunk): Update profile.
8999 * except.c (dw2_build_landing_pads): Do not update frequency.
9000 * final.c (compute_alignments): Use to_frequency.
9001 (dump_basic_block_info): Do not dump frequency.
9002 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
9003 (dump_gimple_bb_header): Do not dump frequency.
9004 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
9006 * gimple-streamer-in.c (input_bb): Do not stream frequency.
9007 * gimple-streamer-out.c (output_bb): Do not stream frequency.
9008 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9009 (init_before_recovery): Do not update frequency.
9010 (sched_create_recovery_edges): Do not update frequency.
9011 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9012 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9013 (ipa_cp_c_finalize): Set max_count to uninitialized.
9014 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9015 (param_change_prob): Use counts.
9016 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9018 * ipa-split.c (consider_split): Use to_frequency.
9019 (split_function): Use to_frequency.
9020 * ira-build.c (loop_compare_func): Likewise.
9021 (mark_loops_for_removal): Likewise.
9022 (mark_all_loops_for_removal): Likewise.
9023 * loop-doloop.c (doloop_modify): Do not update frequency.
9024 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9026 * lto-streamer-in.c (input_function): Update count_max.
9027 * omp-expand.c (expand_omp_taskreg): Update count_max.
9028 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9029 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9030 (maybe_hot_count_p): Use ipa counts only.
9031 (maybe_hot_bb_p): Simplify.
9032 (maybe_hot_edge_p): Simplify.
9033 (probably_never_executed): Do not take frequency argument.
9034 (probably_never_executed_bb_p): Do not pass frequency.
9035 (probably_never_executed_edge_p): Likewise.
9036 (combine_predictions_for_bb): Check that profile is nonzero.
9037 (propagate_freq): Do not set frequency.
9038 (drop_profile): Simplify.
9039 (counts_to_freqs): Simplify.
9040 (expensive_function_p): Use to_frequency.
9041 (propagate_unlikely_bbs_forward): Simplify.
9042 (determine_unlikely_bbs): Simplify.
9043 (estimate_bb_frequencies): Add hack to silence graphite issues.
9044 (compute_function_frequency): Use ipa counts.
9045 (pass_profile::execute): Update.
9046 (rebuild_frequencies): Use counts only.
9047 (force_edge_cold): Use counts only.
9048 * profile-count.c (profile_count::dump): Dump new count types.
9049 (profile_count::differs_from_p): Check compatiblity.
9050 (profile_count::to_frequency): New function.
9051 (profile_count::to_cgraph_frequency): New function.
9052 * profile-count.h (struct function): Declare.
9053 (enum profile_quality): Add profile_guessed_local and
9054 profile_guessed_global0.
9055 (class profile_proability): Decrease number of bits to 29;
9056 update from_reg_br_prob_note and to_reg_br_prob_note.
9057 (class profile_count: Update comment; decrease number of bits
9058 to 61. Check compatibility.
9059 (profile_count::compatible_p): New private member function.
9060 (profile_count::ipa_p): New member function.
9061 (profile_count::operator<): Handle global zero correctly.
9062 (profile_count::operator>): Handle global zero correctly.
9063 (profile_count::operator<=): Handle global zero correctly.
9064 (profile_count::operator>=): Handle global zero correctly.
9065 (profile_count::nonzero_p): New member function.
9066 (profile_count::force_nonzero): New member function.
9067 (profile_count::max): New member function.
9068 (profile_count::apply_scale): Handle IPA scalling.
9069 (profile_count::guessed_local): New member function.
9070 (profile_count::global0): New member function.
9071 (profile_count::ipa): New member function.
9072 (profile_count::to_frequency): Declare.
9073 (profile_count::to_cgraph_frequency): Declare.
9074 * profile.c (OVERLAP_BASE): Delete.
9075 (compute_frequency_overlap): Delete.
9076 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9077 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9078 * sched-ebb.c (rank): Use counts only.
9079 * shrink-wrap.c (handle_simple_exit): Use counts only.
9080 (try_shrink_wrapping): Use counts only.
9081 (place_prologue_for_one_component): Use counts only.
9082 * tracer.c (find_best_predecessor): Use to_frequency.
9083 (find_trace): Use to_frequency.
9084 (tail_duplicate): Use to_frequency.
9085 * trans-mem.c (expand_transaction): Do not update frequency.
9086 * tree-call-cdce.c: Do not update frequency.
9087 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9088 (gimple_merge_blocks): Likewise.
9089 (gimple_split_edge): Likewise.
9090 (gimple_duplicate_sese_region): Likewise.
9091 (gimple_duplicate_sese_tail): Likewise.
9092 (move_sese_region_to_fn): Likewise.
9093 (gimple_account_profile_record): Likewise.
9094 (insert_cond_bb): Likewise.
9095 * tree-complex.c (expand_complex_div_wide): Likewise.
9096 * tree-eh.c (lower_resx): Update profile.
9097 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9099 (initialize_cfun): Do not initialize frequencies
9100 (freqs_to_counts): Delete.
9101 (copy_cfg_body): Ignore count parameter.
9102 (copy_body): Update.
9103 (expand_call_inline): Update count_max.
9104 (optimize_inline_calls): Update count_max.
9105 (tree_function_versioning): Update count_max.
9106 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9107 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9109 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9110 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9111 (try_peel_loop): Likewise.
9112 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9114 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9115 (tree_transform_and_unroll_loop): Do not use frequencies
9116 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9117 Use reliable prediction only.
9118 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9119 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9120 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9121 probability scaling.
9122 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9123 not update frequency
9124 (any_remaining_duplicated_blocks): Likewise.
9125 (update_profile): Likewise.
9126 (estimated_freqs_path): Delete.
9127 (freqs_to_counts_path): Delete.
9128 (clear_counts_path): Delete.
9129 (ssa_fix_duplicate_block_edges): Likewise.
9130 (duplicate_thread_path): Likewise.
9131 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9132 * tree-tailcall.c (decrease_profile): Do not update frequency.
9133 (eliminate_tail_call): Likewise.
9134 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9135 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9136 (optimize_mask_stores): Likewise.
9137 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9138 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9139 (ubsan_expand_ptr_ifn): Update profile.
9140 * value-prof.c (gimple_ic): Simplify.
9141 * value-prof.h (gimple_ic): Update prototype.
9142 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9143 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9145 (want_inline_self_recursive_call_p): Likewise.
9146 (resolve_noninline_speculation): Only cummulate defined counts.
9147 (inline_small_functions): Use nonzero_p.
9148 (ipa_inline): Do not access freed node.
9150 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9152 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9153 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9155 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9157 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9158 non-legitimate address.
9160 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9162 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9163 (*lt0_<mode>di, *lt0_<mode>si): New.
9165 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9167 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9168 TARGET_PAIRED_FLOAT.
9170 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9171 Jim Wilson <jim.wilson@linaro.org>
9173 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9174 * config/aarch64/aarch64-tune.md: Regenerated.
9175 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9176 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9178 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9180 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9181 (arc_expand_prologue): Restore blink for millicode.
9182 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9184 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9187 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9188 gen_vec_duplicate after forcing the scalar into a register.
9190 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9192 * combine (try_combine): Print the insns input to try_combine to the
9195 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9198 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9199 Remove second argument from aarch64_process_target_attr call.
9200 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9202 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9204 (aarch64_handle_attr_arch): Remove second argument.
9205 (aarch64_handle_attr_cpu): Ditto.
9206 (aarch64_handle_attr_tune): Ditto.
9207 (aarch64_handle_attr_isa_flags): Ditto.
9208 (aarch64_process_one_target_attr): Ditto.
9209 (aarch64_process_target_attr): Ditto.
9210 (aarch64_option_valid_attribute_p): Remove second argument.
9211 on aarch64_process_target_attr call.
9213 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9215 * diagnostic.c: Include "selftest-diagnostic.h".
9216 (selftest::assert_location_text): New function.
9217 (selftest::test_diagnostic_get_location_text): New function.
9218 (selftest::diagnostic_c_tests): Call it.
9220 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9222 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9223 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9224 (class selftest::test_diagnostic_context): Move to...
9225 * selftest-diagnostic.c: New file.
9226 * selftest-diagnostic.h: New file.
9228 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9230 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9231 offset range for FT32B.
9232 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9233 * config/ft32/ft32.md: Add TARGET_NOPM.
9234 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9235 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9237 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9239 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9241 2017-11-02 Jeff Law <law@redhat.com>
9243 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9244 virtual keyword on FINAL OVERRIDE members.
9246 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9247 virtuals together. Add virtual destructor.
9248 (substitute_and_fold_engine): Similarly.
9250 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9252 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9254 2017-11-02 Richard Biener <rguenther@suse.de>
9256 PR tree-optimization/82795
9257 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9259 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9261 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9263 (gcc_GAS_CHECK_FEATURE): Remove.
9264 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9265 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9266 * configure: Regenerate.
9268 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9270 * config/arc/arc.c (hwloop_optimize): Account for empty
9273 2017-11-02 Richard Biener <rguenther@suse.de>
9276 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9277 Truncate ARRAY_REF index and element size.
9279 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9281 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9283 2017-11-01 Jeff Law <law@redhat.com>
9285 * tree-ssa-ccp.c (ccp_folder): New class derived from
9286 substitute_and_fold_engine.
9287 (ccp_folder::get_value): New member function.
9288 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9289 (ccp_fold_stmt): Remove prototype.
9290 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9291 * tree-ssa-copy.c (copy_folder): New class derived from
9292 substitute_and_fold_engine.
9293 (copy_folder::get_value): Renamed from get_value.
9294 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9295 * tree-vrp.c (vrp_folder): New class derived from
9296 substitute_and_fold_engine.
9297 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9298 (vrp_folder::get_value): New member function.
9299 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9300 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9301 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9302 provide a class interface to folder/substitute routines.
9303 (ssa_prop_fold_stmt_fn): Remove typedef.
9304 (ssa_prop_get_value_fn): Likewise.
9305 (subsitute_and_fold): Remove prototype.
9306 (replace_uses_in): Likewise.
9307 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9308 Renamed from replace_uses_in. Call the virtual member function
9309 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9310 (substitute_and_fold_dom_walker): Remove initialization of
9311 data member entries for calbacks. Add substitute_and_fold_engine
9312 member and initialize it.
9313 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9314 member functions for get_value, replace_phi_args_in c
9315 replace_uses_in, and fold_stmt calls.
9316 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9317 substitute_and_fold. Remove assert. Update ctor call.
9319 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9320 (ssa_prop_visit_phi_fn): Likewise.
9321 (class ssa_propagation_engine): New class to provide an interface
9323 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9325 (ssa_prop_visit_phi): Likewise.
9326 (ssa_propagation_engine::simulate_stmt): Moved into class.
9327 Call visit_phi/visit_stmt from the class rather than via
9328 file scoped static variables.
9329 (ssa_propagation_engine::simulate_block): Moved into class.
9330 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9331 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9332 set file scoped statics for the visit_stmt/visit_phi callbacks.
9333 * tree-complex.c (complex_propagate): New class derived from
9334 ssa_propagation_engine.
9335 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9336 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9337 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9339 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9340 ssa_propagation_engine.
9341 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9342 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9343 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9344 * tree-ssa-copy.c (copy_prop): New class derived from
9345 ssa_propagation_engine.
9346 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9347 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9348 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9349 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9350 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9351 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9352 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9354 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9356 PR rtl-optimization/82778
9357 PR rtl-optimization/82597
9358 * compare-elim.c (struct comparison): Add in_a_setter field.
9359 (find_comparison_dom_walker::before_dom_children): Remove killed
9360 bitmap and df_simulate_find_defs call, instead walk the defs.
9361 Compute last_setter and initialize in_a_setter. Merge definitions
9362 with first initialization for a few variables.
9363 (try_validate_parallel): Use insn_invalid_p instead of
9364 recog_memoized. Return insn rather than just the pattern.
9365 (try_merge_compare): Fix up comment. Don't uselessly test if
9366 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9368 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9371 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9372 Alan Hayward <alan.hayward@arm.com>
9373 David Sherwood <david.sherwood@arm.com>
9375 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9376 aarch64_hard_regno_nregs to get the number of registers
9379 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9380 Alan Hayward <alan.hayward@arm.com>
9381 David Sherwood <david.sherwood@arm.com>
9383 * config/aarch64/constraints.md (Upl): Rename to...
9385 * config/aarch64/aarch64.md
9386 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9389 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9390 Alan Hayward <alan.hayward@arm.com>
9391 David Sherwood <david.sherwood@arm.com>
9393 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9394 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9397 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9398 Alan Hayward <alan.hayward@arm.com>
9399 David Sherwood <david.sherwood@arm.com>
9401 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9402 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9403 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9405 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9406 of a single element.
9407 * config/aarch64/iterators.md: Add a comment above the permute
9408 unspecs to say that they are generated directly by
9409 aarch64_expand_vec_perm_const.
9410 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9412 2017-11-01 Nathan Sidwell <nathan@acm.org>
9414 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9416 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9418 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9419 and medany code models, and describe what they do.
9421 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9423 Revert accidental duplicate:
9425 * combine.c (can_change_dest_mode): Reject changes in
9426 REGMODE_NATURAL_SIZE.
9428 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9430 PR rtl-optimization/64682
9431 PR rtl-optimization/69567
9432 PR rtl-optimization/69737
9433 PR rtl-optimization/82683
9434 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9435 register mentioned in the note, drop the note, unless it came from I3,
9436 in which case it should go to I3 again.
9438 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9440 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9441 and return false if not.
9442 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9444 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9446 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9447 range is known to be empty.
9449 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9450 Alan Hayward <alan.hayward@arm.com>
9451 David Sherwood <david.sherwood@arm.com>
9453 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9454 and CONST_VECTOR_NUNITS instead of computing the number of units from
9455 the byte sizes of the vector and element.
9456 (simplify_binary_operation_1): Likewise.
9457 (simplify_const_binary_operation): Likewise.
9458 (simplify_ternary_operation): Likewise.
9460 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9461 Alan Hayward <alan.hayward@arm.com>
9462 David Sherwood <david.sherwood@arm.com>
9464 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9465 (int_mem_offset): ...this new function.
9466 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9467 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9468 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9471 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9472 Alan Hayward <alan.hayward@arm.com>
9473 David Sherwood <david.sherwood@arm.com>
9475 * lower-subreg.c (interesting_mode_p): New function.
9476 (compute_costs, find_decomposable_subregs, decompose_register)
9477 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9478 (resolve_clobber, dump_choices): Use it.
9480 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9481 Alan Hayward <alan.hayward@arm.com>
9482 David Sherwood <david.sherwood@arm.com>
9484 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9486 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9487 Alan Hayward <alan.hayward@arm.com>
9488 David Sherwood <david.sherwood@arm.com>
9490 * alias.c (find_base_value, find_base_term): Only process integer
9491 truncations. Check the precision rather than the size.
9493 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9494 Alan Hayward <alan.hayward@arm.com>
9495 David Sherwood <david.sherwood@arm.com>
9497 * machmode.h (is_narrower_int_mode): New function
9498 * optabs.c (expand_float, expand_fix): Use it.
9499 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9501 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9502 Alan Hayward <alan.hayward@arm.com>
9503 David Sherwood <david.sherwood@arm.com>
9505 * rtl.h (narrower_subreg_mode): New function.
9506 * ira-color.c (update_costs_from_allocno): Use it.
9508 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9509 Alan Hayward <alan.hayward@arm.com>
9510 David Sherwood <david.sherwood@arm.com>
9512 * optabs-query.h (convert_optab_p): New function, split out from...
9513 (convert_optab_handler): ...here.
9514 (widening_optab_handler): Delete.
9515 (find_widening_optab_handler): Remove permit_non_widening parameter.
9516 (find_widening_optab_handler_and_mode): Likewise. Provide an
9517 override that operates on mode class wrappers.
9518 * optabs-query.c (widening_optab_handler): Delete.
9519 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9520 parameter. Assert that the two modes are the same class and that
9521 the "from" mode is narrower than the "to" mode. Use
9522 convert_optab_handler instead of widening_optab_handler.
9523 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9524 instead of widening_optab_handler.
9525 * expr.c (expand_expr_real_2): Update calls to
9526 find_widening_optab_handler.
9527 * optabs.c (expand_widen_pattern_expr): Likewise.
9528 (expand_binop_directly): Take the insn_code as a parameter.
9529 (expand_binop): Only call find_widening_optab_handler for
9530 conversion optabs; use optab_handler otherwise. Update calls
9531 to find_widening_optab_handler and expand_binop_directly.
9532 Use convert_optab_handler instead of widening_optab_handler.
9533 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9534 find_widening_optab_handler and use scalar_mode rather than
9536 (convert_plusminus_to_widen): Likewise.
9538 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9539 Alan Hayward <alan.hayward@arm.com>
9540 David Sherwood <david.sherwood@arm.com>
9542 * machmode.h (fixed_size_mode): New class.
9543 * rtl.h (get_pool_mode): Return fixed_size_mode.
9544 * gengtype.c (main): Add fixed_size_mode.
9545 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9546 (get_raw_arg_mode): Likewise.
9547 * doc/tm.texi: Regenerate.
9548 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9549 * targhooks.c (default_get_reg_raw_mode): Likewise.
9550 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9551 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9552 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9553 (msp430_get_raw_result_mode): Likewise.
9554 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9555 * dbxout.c (dbxout_parms): Require fixed-size modes.
9556 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9557 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9558 * omp-low.c (lower_oacc_reductions): Likewise.
9559 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9560 (simplify_subreg): Update accordingly.
9561 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9562 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9563 that aren't fixed-size.
9564 (get_pool_mode): Return a fixed_size_mode.
9565 (output_constant_pool_2): Take a fixed_size_mode.
9567 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9568 Alan Hayward <alan.hayward@arm.com>
9569 David Sherwood <david.sherwood@arm.com>
9571 * doc/rtl.texi (vec_series): Document.
9572 (const): Say that the operand can be a vec_series.
9573 * rtl.def (VEC_SERIES): New rtx code.
9574 * rtl.h (const_vec_series_p_1): Declare.
9575 (const_vec_series_p): New function.
9576 * emit-rtl.h (gen_const_vec_series): Declare.
9577 (gen_vec_series): Likewise.
9578 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9579 (gen_vec_series): Likewise.
9580 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9581 * simplify-rtx.c (simplify_unary_operation): Handle negations
9583 (simplify_binary_operation_series): New function.
9584 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9585 (test_vector_ops_series): New function.
9586 (test_vector_ops): Call it.
9587 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9588 gen_const_vec_series.
9589 (altivec_lvsr): Likewise.
9590 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9592 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9593 Alan Hayward <alan.hayward@arm.com>
9594 David Sherwood <david.sherwood@arm.com>
9596 * doc/rtl.texi (const): Update description of address constants.
9597 Say that vector constants are allowed too.
9598 * common.md (E, F): Use CONSTANT_P instead of checking for
9600 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9601 checking for CONST_VECTOR.
9602 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9604 * expr.c (expand_expr_real_2): Check for vector modes instead
9605 of checking for CONST_VECTOR.
9606 * rtl.h (const_vec_p): New function.
9607 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9608 (unwrap_const_vec_duplicate): Handle them here too.
9610 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9611 David Malcolm <dmalcolm@redhat.com>
9612 Alan Hayward <alan.hayward@arm.com>
9613 David Sherwood <david.sherwood@arm.com>
9615 * rtl.h (vec_duplicate_p): New function.
9616 * selftest-rtl.c (assert_rtx_eq_at): New function.
9617 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9618 (assert_rtx_eq_at): Declare.
9619 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9620 * selftest-run-tests.c (selftest::run_tests): Call it.
9621 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9622 (simplify_unary_operation_1): Recursively handle vector duplicates.
9623 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9625 (simplify_subreg): Handle subregs of vector duplicates.
9626 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9627 (selftest::simplify_rtx_c_tests): New functions.
9629 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9630 Alan Hayward <alan.hayward@arm.com>
9631 David Sherwood <david.sherwood@arm.com>
9633 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9634 (gen_vec_duplicate): Likewise.
9635 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9637 (gen_const_vector): ...here.
9638 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9639 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9640 whose elements are all equal.
9641 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9642 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9643 (simplify_relational_operation): Likewise.
9644 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9646 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9647 (aarch64_expand_vector_init): Likewise.
9648 * config/arm/arm.c (neon_vdup_constant): Likewise.
9649 (neon_expand_vector_init): Likewise.
9650 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9651 (arm_block_set_unaligned_vect): Likewise.
9652 (arm_block_set_aligned_vect): Likewise.
9653 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9654 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9655 (expand_vec_perm_even_odd_pack): Likewise.
9656 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9657 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9658 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9659 gen_const_vec_duplicate.
9660 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9661 * config/mips/mips.c (mips_gen_const_int_vector): Use
9662 gen_const_vec_duplicate.
9663 (mips_expand_vector_init): Use CONST0_RTX.
9664 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9665 (define_split): Use gen_const_vec_duplicate.
9666 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9667 (define_split): Use gen_const_vec_duplicate.
9668 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9669 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9670 * config/spu/spu.c (spu_const): Likewise.
9672 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9673 Alan Hayward <alan.hayward@arm.com>
9674 David Sherwood <david.sherwood@arm.com>
9676 * combine.c (can_change_dest_mode): Reject changes in
9677 REGMODE_NATURAL_SIZE.
9679 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9681 * configure.ac (--enable-libssp): New.
9682 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9683 trying to determine target-specific default. Adjust indentation.
9684 * configure: Regenerated.
9685 * doc/install.texi (Configuration): Expand --disable-libssp
9688 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9690 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9693 2017-10-31 Martin Jambor <mjambor@suse.cz>
9696 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9698 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9700 * auto-profile.c (autofdo_source_profile::read): Use
9701 UNKNOWN_LOCATION rather than 0.
9702 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9703 (warning_at): ...this overload.
9704 (warning_at_rich_loc_n): Rename to...
9705 (warning_n): ...this overload.
9706 (error_at_rich_loc): Rename to...
9707 (error_at): ...this overload.
9708 (pedwarn_at_rich_loc): Rename to...
9709 (pedwarn): ...this overload.
9710 (permerror_at_rich_loc): Rename to...
9711 (permerror): ...this overload.
9712 (inform_at_rich_loc): Rename to...
9713 (inform): ...this overload.
9714 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9715 (diagnostic_n_impl_richloc): Rename to...
9716 (diagnostic_n_impl): ...this rich_location *-based decl.
9717 (inform_at_rich_loc): Rename to...
9718 (inform): ...this, and add an assertion.
9719 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9720 (warning_at_rich_loc): Rename to...
9721 (warning_at): ...this, and add an assertion.
9722 (warning_at_rich_loc_n): Rename to...
9723 (warning_n): ...this, and add an assertion.
9724 (warning_n): Update location_t-based implementation for removal of
9725 location_t-based diagnostic_n_impl.
9726 (pedwarn_at_rich_loc): Rename to...
9727 (pedwarn): ...this, and add an assertion.
9728 (permerror_at_rich_loc): Rename to...
9729 (permerror): ...this, and add an assertion.
9730 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9731 (error_at_rich_loc): Rename to...
9732 (error_at): ...this, and add an assertion.
9733 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9734 (driver::do_spec_on_infiles): Likewise.
9735 * substring-locations.c (format_warning_va): Update for renaming
9736 of inform_at_rich_loc.
9738 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9740 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9741 _Float<N>X built-in functions so that the variant without the
9742 "__builtin_" prefix is only enabled for the GNU C and Objective C
9743 languages when they are in non-strict ANSI/ISO mode.
9744 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9745 * target.def (floatn_builtin_p): Add a target hook to control
9746 whether _Float<N> and _Float<N>X built-in functions without the
9747 "__builtin_" prefix are enabled, and return true for C and
9748 Objective C in the default hook. Include langhooks.h in
9750 * targhooks.h (default_floatn_builtin_p): Likewise.
9751 * targhooks.c (default_floatn_builtin_p): Likewise.
9752 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9753 floatn_builtin_p target hook.
9754 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9756 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9757 Eric Botcazou <ebotcazou@adacore.com>
9759 PR rtl-optimization/81803
9760 * lra-constraints.c (curr_insn_transform): Also reload the whole
9761 register for a strict subreg no wider than a word if this is for
9762 a WORD_REGISTER_OPERATIONS target.
9764 2017-10-31 Jason Merrill <jason@redhat.com>
9766 * gdbinit.in: Skip over inlines from timevar.h.
9768 2017-10-31 Martin Liska <mliska@suse.cz>
9770 * doc/gcov.texi: Document new option.
9771 * gcov.c (print_usage): Likewise print it.
9772 (process_args): Support the argument.
9773 (format_count): New function.
9774 (format_gcov): Use the function.
9776 2017-10-31 Martin Liska <mliska@suse.cz>
9778 * gcov.c (struct name_map): do not use typedef.
9779 Define operator== and operator<.
9780 (name_search): Remove.
9781 (name_sort): Remove.
9782 (main): Do not allocate names.
9783 (process_file): Add vertical space.
9784 (generate_results): Use std::find.
9785 (release_structures): Do not release memory.
9786 (find_source): Use std::find.
9788 2017-10-31 Martin Liska <mliska@suse.cz>
9790 * gcov.c (struct line_info): Remove it's typedef.
9791 (line_info::line_info): Add proper ctor.
9792 (line_info::has_block): Do not use a typedef.
9793 (struct source_info): Do not use typedef.
9794 (circuit): Likewise.
9795 (get_cycles_count): Likewise.
9796 (output_intermediate_file): Iterate via vector iterator.
9797 (add_line_counts): Use std::vector methods.
9798 (accumulate_line_counts): Likewise.
9799 (output_lines): Likewise.
9801 2017-10-31 Martin Liska <mliska@suse.cz>
9803 * gcov.c (struct source_info): Remove typedef.
9804 (source_info::source_info): Add proper ctor.
9805 (accumulate_line_counts): Use struct, not it's typedef.
9806 (output_gcov_file): Likewise.
9807 (output_lines): Likewise.
9808 (main): Do not allocate an array.
9809 (output_intermediate_file): Use size of vector container.
9810 (process_file): Resize the vector.
9811 (generate_results): Do not preallocate, use newly added vector
9813 (release_structures): Do not release sources.
9814 (find_source): Use vector methods.
9815 (add_line_counts): Do not use typedef.
9817 2017-10-31 Martin Liska <mliska@suse.cz>
9819 * doc/gcov.texi: Document that.
9820 * gcov.c (add_line_counts): Mark lines with a non-executed
9822 (output_line_beginning): Handle such lines.
9823 (output_lines): Pass new argument.
9824 (output_intermediate_file): Print it in intermediate format.
9826 2017-10-31 Martin Liska <mliska@suse.cz>
9828 * color-macros.h: New file.
9829 * diagnostic-color.c: Factor out color related to macros to
9831 * doc/gcov.texi: Document -k option.
9832 * gcov.c (INCLUDE_STRING): Include string.h.
9833 (print_usage): Add -k option.
9834 (process_args): Parse it.
9835 (pad_count_string): New function.
9836 (output_line_beginning): Likewise.
9837 (DEFAULT_LINE_START): New macro.
9838 (output_lines): Support color output.
9840 2017-10-31 Martin Liska <mliska@suse.cz>
9842 PR gcov-profile/82633
9843 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9844 their interaction with GCOV infrastructure.
9845 * configure.ac: Add -fkeep-{inline,static}-functions to
9847 * configure: Regenerate.
9849 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
9852 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
9854 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
9857 * config/rs6000/rs6000.md (allocate_stack): Force update interval
9858 into a register if it does not fit into an immediate offset field.
9860 2017-10-31 Olivier Hainque <hainque@adacore.com>
9862 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9864 2017-10-31 Julia Koval <julia.koval@intel.com>
9866 * config.gcc: Add gfniintrin.h.
9867 * config/i386/gfniintrin.h: New.
9868 * config/i386/i386-builtin-types.def
9869 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
9870 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
9871 __builtin_ia32_vgf2p8affineinvqb_v32qi,
9872 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
9873 __builtin_ia32_vgf2p8affineinvqb_v16qi,
9874 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
9875 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
9876 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
9877 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
9878 V64QI_FTYPE_V64QI_V64QI_INT): New types.
9879 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
9880 * config/i386/immintrin.h: Include gfniintrin.h.
9881 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
9883 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9885 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9887 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9889 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9890 (arm_ashldi3_1bit): Remove pattern.
9891 (ashrdi3): Remove shift by 1 expansion.
9892 (arm_ashrdi3_1bit): Remove pattern.
9893 (lshrdi3): Remove shift by 1 expansion.
9894 (arm_lshrdi3_1bit): Remove pattern.
9895 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9896 cost of ashldi3 by 1.
9897 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9898 (<shift>di3_neon): Likewise.
9900 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9902 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9903 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9904 and (*aarch64_simd_mov<VQ:mode>).
9905 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9906 pattern alternative.
9907 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9908 attributes to match pattern alternative.
9910 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9912 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9913 * config/rs6000/emmintrin.h: New file.
9914 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9916 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9918 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9919 * (movdi_vfp_cortexa8): Remove pattern.
9921 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9923 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9925 (Specific, alpha*-dec-osf5.1): Remove.
9926 (Specific, mips-sgi-irix5): Remove.
9927 (Specific, mips-sgi-irix6): Remove.
9929 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9932 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9933 arguments to clear_bit_region_be.
9935 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9937 * gimplify.c: Include memmodel.h.
9939 2017-10-30 Martin Jambor <mjambor@suse.cz>
9941 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9942 condition whether loop should be executed at all.
9944 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9946 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9947 gimple folding of vec_madd() intrinsics.
9948 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9949 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9950 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9952 2017-10-30 Richard Biener <rguenther@suse.de>
9954 PR tree-optimization/82762
9956 2017-10-23 Richard Biener <rguenther@suse.de>
9958 PR tree-optimization/82129
9960 2017-08-01 Richard Biener <rguenther@suse.de>
9962 PR tree-optimization/81181
9963 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9964 (compute_antic): ... end of iteration here.
9966 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9968 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9969 -std=iso9899:2017 and -std=gnu17.
9970 * doc/standards.texi (C Language): Document C17 support.
9971 * doc/cpp.texi (Overview): Mention -std=c17.
9972 (Standard Predefined Macros): Document C11 and C17 values of
9973 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9974 * doc/extend.texi (Inline): Do not list individual options for
9975 standards newer than C99.
9976 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9978 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9981 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9983 * asan.c (asan_finish_file): Align asan globals array by shadow
9986 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9989 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9990 (struct store_immediate_info): Add bitregion_start and bitregion_end
9992 (store_immediate_info::store_immediate_info): Add brs and bre
9993 arguments and initialize bitregion_{start,end} from those.
9994 (struct merged_store_group): Add bitregion_start, bitregion_end,
9995 align_base and mask fields. Drop unnecessary struct keyword from
9996 struct store_immediate_info. Add do_merge method.
9997 (clear_bit_region_be): Use memset instead of loop storing zeros.
9998 (merged_store_group::do_merge): New method.
9999 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
10000 stores as long as the surrounding bitregions have no gaps.
10001 (merged_store_group::merge_overlapping): Use do_merge.
10002 (merged_store_group::apply_stores): Test that bitregion_{start,end}
10003 is byte aligned, rather than requiring that start and width are
10004 byte aligned. Drop unnecessary struct keyword from
10005 struct store_immediate_info. Allocate and populate also mask array.
10006 Make start of the arrays relative to bitregion_start rather than
10007 start and size them according to bitregion_{end,start} difference.
10008 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10009 struct store_immediate_info.
10010 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10011 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10012 struct keyword from struct store_immediate_info.
10013 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10014 between stores as long as the surrounding bitregions have no gaps.
10016 (struct split_store): Add orig non-static data member.
10017 (split_store::split_store): Initialize orig to false.
10018 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10019 if there is exactly a single original stmt. Change stmts argument
10020 to pointer from reference, if NULL, don't push anything to it. Add
10021 first argument, use it to optimize skipping over orig stmts that
10022 are known to be before bitpos already. Simplify.
10023 (split_group): Return unsigned int count how many stores are or
10024 would be needed rather than a bool. Add allow_unaligned argument.
10025 Change split_stores argument from reference to pointer, if NULL,
10026 only do a dry run computing how many stores would be produced.
10027 Rewritten algorithm to use both alignment and misalign if
10028 !allow_unaligned and handle bitfield stores with gaps.
10029 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10030 from bitregion_start instead of start. Compute allow_unaligned
10031 here, if true, do 2 split_group dry runs to compute which one
10032 produces fewer stores and prefer aligned if equal. Punt if
10033 new count is bigger or equal than original before emitting any
10034 statements, rather than during that. Remove no longer needed
10035 new_ssa_names tracking. Replace num_stmts with
10036 split_stores.length (). Use 32-bit stack allocated entries
10037 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10038 if possible. Handle bitfields with gaps.
10039 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10040 Compute bitregion_{start,end} for the stores and construct
10041 store_immediate_info with that. Formatting fixes.
10043 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10046 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10047 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10049 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10051 * gimplify.c: Include tm_p.h.
10053 * common.opt (gcoff): Re-add as ignored option.
10054 (gcoff1, gcoff2, gcoff3): Likewise.
10056 * Makefile.in (OBJS): Delete sdbout.o.
10057 (GTFILES): Delete $(srcdir)/sdbout.c.
10058 * debug.h: Delete sdb_debug_hooks.
10059 * final.c: Delete sdbout.h include.
10060 (final_scan_insn): Delete SDB_DEBUG check.
10061 (rest_of_clean_state): Likewise.
10062 * output.h: Delete sdb_begin_function_line.
10063 * sdbout.c: Delete.
10064 * sdbout.h: Delete.
10065 * toplev.c: Delete sdbout.h include.
10066 (process_options): Delete SDB_DEBUG check.
10067 * tree-core.h (tree_type_common): Delete pointer field of
10069 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10070 TYPE_SYMTAB_POINTER.
10071 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10072 (TYPE_SYMTAB_IS_POINTER): Delete.
10073 (TYPE_SYMTAB_IS_DIE): Renumber.
10074 * xcoffout.c: Refer to former sdbout.c file.
10075 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10077 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10078 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10079 (Debugging Options): Delete -gcoff.
10080 (-gstabs): Delete SDB reference.
10082 (-gcoff@var{level}): Delete.
10083 * doc/passes.texi (Debugging information output): Delete SDB and
10084 sdbout.c references.
10085 * doc/tm.texi: Regenerate.
10086 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10087 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10089 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10090 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10091 (SDB_DEBUGGING_INFO): Delete.
10092 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10093 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10094 * target.def (output_source_filename): Delete COFF reference.
10096 * common.opt (gcoff): Delete.
10097 (gxcoff+): Update Negative chain.
10098 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10100 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10101 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10102 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10103 and SDB references.
10104 (expand_function_start): Change sdb reference to past tense.
10105 (expand_function_end): Change sdb reference to past tense.
10106 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10107 * opts.c (debug_type_names): Delete coff entry.
10108 (common_handle_option): Delete OPT_gcoff case.
10109 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10111 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10112 * config/cris/cris.h: Delete SDB reference in comment.
10113 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10114 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10115 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10116 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10118 (ix86_expand_prologue): Likewise.
10119 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10120 * config/ia64/ia64.h: Likewise.
10121 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10122 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10124 * config/mmix/mmix.h: Likewise.
10125 * config/nds32/nds32.c: Likewise.
10126 * config/stormy/storym16.h: Likewise.
10127 * config/visium/visium.h: Likewise.
10128 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10130 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10132 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10133 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10134 FRAME_POINTER_REGNUM point at high end of local var area.
10136 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10138 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10139 Move comment around. Do not reset best_edge for a copiable
10140 destination if the copy would cause a partition change.
10141 (better_edge_p): Remove redundant check.
10143 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10145 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10147 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10149 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10150 for math functions that have _Float<N> and _Float<N>X variants.
10151 (mathfn_built_in_2): Add support for math functions that have
10152 _Float<N> and _Float<N>X variants.
10153 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10154 (expand_builtin_mathfn_ternary): Add support for fma with
10155 _Float<N> and _Float<N>X variants.
10156 (expand_builtin): Likewise.
10157 (fold_builtin_3): Likewise.
10158 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10159 create math function _Float<N> and _Float<N>X variants as external
10161 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10162 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10163 the __builtin_ prefix and if not strict ansi, without the prefix.
10164 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10165 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10166 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10167 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10168 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10169 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10170 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10171 function signatures for fma _Float<N> and _Float<N>X variants.
10172 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10173 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10174 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10175 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10176 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10177 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10178 * gencfn-macros.c (print_case_cfn): Add support for math functions
10179 that have _Float<N> and _Float<N>X variants.
10180 (print_define_operator_list): Likewise.
10181 (fltfn_suffixes): Likewise.
10183 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10184 for math functions that have _Float<N> and _Float<N>X variants.
10185 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10186 and _Float<N>X variants.
10187 (COPYSIGN): Likewise.
10190 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10191 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10193 (integer_valued_read_call_p): Likewise.
10194 * fold-const-call.c (fold_const_call_ss): Likewise.
10195 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10196 _Float<N> and _Float<N>X variants.
10197 (fold_const_call_ssss): Add support for fma _Float<N> and
10198 _Float<N>X variants.
10199 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10200 support for copysign and fma _Float<N> and _Float<N>X variants.
10201 (backprop::process_builtin_call_use): Likewise.
10202 * tree-call-cdce.c (can_test_argument_range); Add support for
10203 sqrt _Float<N> and _Float<N>X variants.
10204 (edom_only_function): Likewise.
10205 (get_no_error_domain): Likewise.
10206 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10207 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10208 copysign _Float<N> and _Float<N>X variants.
10209 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10210 handled by machine independent code.
10211 (FMAF128): Likewise.
10212 * doc/cpp.texi (Common Predefined Macros): Document defining
10213 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10214 fma _Float<N> and _Float<N>X variants.
10216 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10219 * config/i386/i386-modes.def (CCFPU): Remove definition.
10220 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10221 (ix86_cc_modes_compatible): Ditto.
10222 (ix86_expand_carry_flag_compare): Ditto.
10223 (ix86_expand_int_movcc): Ditto.
10224 (ix86_expand_int_addcc): Ditto.
10225 (ix86_reverse_condition): Ditto.
10226 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10227 Return true/false for unordered/ordered fp comparisons.
10228 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10229 (ix86_prepare_fp_compare_args): Update for rename.
10230 (ix86_expand_fp_compare): Update for rename. Generate unordered
10231 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10232 (ix86_expand_sse_compare_and_jump): Ditto.
10233 * config/i386/predicates.md (fcmov_comparison_operator):
10234 Remove CCFPU mode handling.
10235 (ix86_comparison_operator): Ditto.
10236 (ix86_carry_flag_operator): Ditto.
10237 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10238 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10239 (*cmpu<mode>_cc_i387): Ditto.
10240 (FPCMP): Remove mode iterator.
10241 (unord): Remove mode attribute.
10242 (unord_subst): New define_subst transformation
10243 (unord): New define_subst attribute.
10244 (unordered): Ditto.
10245 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10246 (*cmpi<unord>xf_i387): Ditto.
10247 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10248 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10249 using unord_subst transformation.
10250 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10251 (round_saeonly): Also handle CCFP mode.
10252 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10253 Remove UNSPEC_SAHF unspec handling.
10255 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10257 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10259 2017-10-27 Jeff Law <law@redhat.com>
10261 * gimple-ssa-sprintf.c: Include domwalk.h.
10262 (class sprintf_dom_walker): New class, derived from dom_walker.
10263 (sprintf_dom_walker::before_dom_children): New function.
10264 (struct call_info): Moved into sprintf_dom_walker class
10265 (compute_formath_length, handle_gimple_call): Likewise.
10266 (sprintf_length::execute): Call the dominator walker rather
10267 than walking the statements.
10269 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10270 gimple statement locations.
10271 (check_array_bounds): Corresponding changes. Get the statement's
10272 location directly from wi->stmt.
10274 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10277 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10279 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10281 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10282 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10284 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10287 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10288 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10289 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10290 maybe_get_pool_constant.
10291 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10294 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10296 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10297 2.26 caveat. Update gas and gld versions.
10298 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10301 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10303 * cgraph.h (set_malloc_flag): Declare.
10304 * cgraph.c (set_malloc_flag_1): New function.
10305 (set_malloc_flag): Likewise.
10306 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10307 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10309 (read_ipa_call_summary): Add support for reading is_return_callee.
10310 (write_ipa_call_summary): Stream is_return_callee.
10311 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10312 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10313 ipa-prop.h, ipa-fnsummary.h.
10314 (pure_const_names): Change to static.
10315 (malloc_state_e): Define.
10316 (malloc_state_names): Define.
10317 (funct_state_d): Add field malloc_state.
10318 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10319 (check_retval_uses): New function.
10320 (malloc_candidate_p): Likewise.
10321 (analyze_function): Add support for malloc attribute.
10322 (pure_const_write_summary): Stream malloc_state.
10323 (pure_const_read_summary): Add support for reading malloc_state.
10324 (dump_malloc_lattice): New function.
10325 (propagate_malloc): New function.
10326 (warn_function_malloc): New function.
10327 (ipa_pure_const::execute): Call propagate_malloc and
10328 ipa_free_fn_summary.
10329 (pass_local_pure_const::execute): Add support for malloc attribute.
10330 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10331 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10333 2017-10-27 Martin Liska <mliska@suse.cz>
10335 PR gcov-profile/82457
10336 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10337 for fork-like functions to be properly instrumented.
10339 2017-10-27 Richard Biener <rguenther@suse.de>
10341 PR middle-end/81659
10342 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10343 info when we redirected EH.
10345 2017-10-26 Michael Collison <michael.collison@arm.com>
10347 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10349 (<optab>_trunchf<GPI:mode>2: New pattern.
10350 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10351 * config/aarch64/iterators.md (wv): New mode attribute.
10352 (vf, VF): New mode attributes.
10353 (vgp, VGP): New mode attributes.
10354 (s): Update attribute with SImode and DImode prefixes.
10356 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10358 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10359 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10360 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10361 (nios2_option_overide): Initialize it. Don't allow R0-relative
10362 addressing with PIC.
10363 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10364 (nios2_symbolic_constant_p): Likewise.
10365 (nios2_legitimate_address_p): Likewise.
10366 (nios2_r0rel_section_name_p): New.
10367 (nios2_symbol_ref_in_r0rel_data_p): New.
10368 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10369 (r0rel_constant_p): New.
10370 (nios2_print_operand_address): Handle r0rel_constant_p.
10371 (nios2_cdx_narrow_form_p): Likewise.
10372 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10373 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10374 (Nios II Options): Document -mr0rel-sec.
10376 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10378 * config/nios2/nios2.c: Include xregex.h.
10379 (nios2_gprel_sec_regex): New.
10380 (nios2_option_overide): Initialize it. Don't allow GP-relative
10381 addressing with PIC.
10382 (nios2_small_section_name_p): Check for regex match.
10383 * config/nios2/nios2.opt (mgprel-sec=): New option.
10384 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10385 (Nios II Options): Document -mgprel-sec.
10387 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10389 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10391 2017-10-26 Tom de Vries <tom@codesourcery.com>
10393 PR tree-optimization/82707
10394 * gimple.c (gimple_copy): Fix unsharing of
10395 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10397 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10399 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10400 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10401 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10402 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10403 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10404 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10405 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10406 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10407 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10408 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10409 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10410 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10411 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10412 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10413 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10414 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10415 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10417 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10419 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10421 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10422 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10423 warning to rs6000.c. Remove the Undocumented flag, since it has
10425 (-mabi=ibmlongdouble): Likewise.
10426 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10427 already set, set the default format for long double.
10428 (rs6000_debug_reg_global): Print whether long double is IBM or
10430 (rs6000_option_override_internal): Rework setting long double
10431 format. Only warn if the user is changing the long double default
10432 and they did not use -Wno-psabi.
10433 * doc/invoke.texi (PowerPC options): Update the documentation for
10434 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10436 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10437 Alan Hayward <alan.hayward@arm.com>
10438 David Sherwood <david.sherwood@arm.com>
10440 * rtl.h (wider_subreg_mode): New function.
10441 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10442 rather than an unsigned int *.
10443 * ira-color.c (regno_max_ref_width): Replace with...
10444 (regno_max_ref_mode): ...this new variable.
10445 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10446 Use wider_subreg_mode.
10447 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10448 rather than an unsigned int *.
10449 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10450 (process_alt_operands): Likewise.
10451 (invariant_p): Likewise.
10452 * lra-spills.c (assign_mem_slot): Likewise.
10453 (add_pseudo_to_slot): Likewise.
10454 * lra.c (collect_non_operand_hard_regs): Likewise.
10455 (add_regs_to_insn_regno_info): Likewise.
10456 * reload1.c (regno_max_ref_width): Replace with...
10457 (regno_max_ref_mode): ...this new variable.
10458 (reload): Update accordingly. Update call to
10459 ira_sort_regnos_for_alter_reg.
10460 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10461 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10462 (scan_paradoxical_subregs): Likewise.
10464 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10466 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10467 (aarch64_frame): Add emit_frame_chain boolean.
10468 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10469 Move eh_return case to aarch64_layout_frame.
10470 (aarch64_layout_frame): Initialize emit_frame_chain.
10471 (aarch64_expand_prologue): Use emit_frame_chain.
10473 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10475 * config/aarch64/aarch64.c (aarch64_layout_frame):
10476 Ensure LR is always stored at the bottom of the callee-saves.
10477 Remove rarely used frame layout which saves callee-saves at top of
10478 frame, so the store of LR can be used as a valid probe in all cases.
10480 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10482 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10483 Improve unaligned TImode/TFmode base/offset split.
10485 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10486 Alan Hayward <alan.hayward@arm.com>
10487 David Sherwood <david.sherwood@arm.com>
10489 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10490 * combine.c (find_single_use_1): Likewise.
10491 (expand_field_assignment): Likewise.
10492 (move_deaths): Likewise.
10493 * lra-constraints.c (simplify_operand_subreg): Likewise.
10494 (curr_insn_transform): Likewise.
10495 * lra.c (collect_non_operand_hard_regs): Likewise.
10496 (add_regs_to_insn_regno_info): Likewise.
10497 * rtlanal.c (reg_referenced_p): Likewise.
10498 (covers_regno_no_parallel_p): Likewise.
10500 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10502 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10503 Don't print any bits outside the precision of the value.
10504 * wide-int.cc (test_printing): Add some new tests.
10506 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10508 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10509 supports -xbrace_comment option.
10510 * configure: Regenerate.
10511 * config.in: Regenerate.
10512 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10513 (ASM_CPU_SPEC): Use it.
10515 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10517 * target.def (static_rtx_alignment): New hook.
10518 * targhooks.h (default_static_rtx_alignment): Declare.
10519 * targhooks.c (default_static_rtx_alignment): New function.
10520 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10521 * doc/tm.texi: Regenerate.
10522 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10523 instead of targetm.constant_alignment. Remove call to
10524 set_mem_attributes.
10525 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10526 (cris_preferred_mininum_alignment): New function, split out from...
10527 (cris_constant_alignment): ...here.
10528 (cris_static_rtx_alignment): New function.
10529 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10531 (ix86_constant_alignment): ...here.
10532 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10533 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10534 (mmix_static_rtx_alignment): New function.
10535 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10536 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10538 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10541 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10542 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10544 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10546 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10547 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10548 mark as a sign-extending load.
10549 (local_pic_load_u): Define.
10551 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10553 PR middle-end/82062
10554 * fold-const.c (operand_equal_for_comparison_p): Also return true
10555 if ARG0 is a simple variant of ARG1 with narrower precision.
10556 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10558 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10560 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10562 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10563 scatter_static, scatter_per_elt.
10564 * x86-tune-costs.h: Add new cost entries.
10566 2017-10-25 Richard Biener <rguenther@suse.de>
10568 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10569 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10570 class pass_fre): Move to ...
10571 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10572 class pass_fre): ... here and adjust for statistics.
10574 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10577 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10578 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10579 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10580 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10583 2017-10-25 Richard Biener <rguenther@suse.de>
10585 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10586 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10587 eliminate_push_avail, eliminate_insert): Move inside...
10588 (class eliminate_dom_walker): ... this class in preparation
10590 (fini_eliminate): Remove by merging with ...
10591 (eliminate): ... this function. Adjust for class changes.
10592 (pass_pre::execute): Remove fini_eliminate call.
10593 (pass_fre::execute): Likewise.
10595 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10598 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10599 (VPERMI2, VPERMI2I): New mode iterators.
10600 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10601 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10603 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10604 mode iterator. Remove 3 old define_insn patterns.
10605 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10606 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10607 VPERMI2 mode iterator, remove the other two expanders.
10608 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10609 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10610 instructions, remove the other two patterns.
10611 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10612 mode iterator, remove the other two patterns.
10613 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10614 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10615 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10616 and adjust argument order accordingly.
10617 (ix86_expand_vec_perm): Adjust caller.
10618 (expand_vec_perm_1): Likewise.
10619 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10620 (expand_vec_perm_vpermt2_vpshub2): ... this.
10621 (ix86_expand_vec_perm_const_1): Adjust caller.
10622 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10625 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10626 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10627 (vec_shl_<mode>): Remove unused expander.
10628 (avx512bw_<shift_insn><mode>3): New define_insn.
10629 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10630 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10632 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10635 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10638 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10640 PR rtl-optimization/82396
10641 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10642 (autopref_multipass_init): Simplify initialization.
10643 (autopref_rank_data): Simplify sort order.
10644 * gcc/sched-int.h (autopref_multipass_data_): Remove
10645 multi_mem_insn_p, min_offset and max_offset.
10647 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10649 PR middle-end/60580
10650 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10651 Check special value of flag_omit_frame_pointer.
10652 (aarch64_can_eliminate): Likewise.
10653 (aarch64_override_options_after_change_1): Simplify handling of
10654 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10656 2017-10-24 Richard Biener <rguenther@suse.de>
10658 PR tree-optimization/82697
10659 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10660 zero for conditional load and unconditional store.
10662 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10664 * doc/install.texi: Document bootstrap-cet.
10666 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10669 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10670 ENDBR instruction at function entrance if function is only
10673 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10676 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10677 patterns to better describe from which operation the CF is computed.
10678 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10679 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10680 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10681 is 0, use _0 suffixed expanders instead of emitting a comparison
10684 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10686 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10687 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10689 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10691 PR middle-end/82569
10692 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10693 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10694 * loop-iv.c (iv_get_reaching_def): Likewise.
10695 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10696 variable is promoted and the partition contains undefined values.
10698 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10700 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10702 (nios2_address_cost): Define.
10703 (nios2_legitimize_address): Recognize (exp + constant) directly.
10704 (TARGET_ADDRESS_COST): Define.
10706 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10708 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10709 (nios2_symbolic_memory_operand_p): Declare.
10710 (nios2_split_large_constant): Declare.
10711 (nios2_split_symbolic_memory_operand): Declare.
10712 * config/nios2/nios2.c: Adjust includes.
10713 (nios2_symbolic_constant_allowed): New.
10714 (nios2_symbolic_constant_p): New.
10715 (nios2_plus_symbolic_constant_p): New.
10716 (nios2_valid_addr_expr_p): Recognize addresses involving
10717 symbolic constants.
10718 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10719 (nios2_symbolic_memory_operand_p): New.
10720 (nios2_large_constant_p): New.
10721 (nios2_split_large_constant): New.
10722 (nios2_split_plus_large_constant): New.
10723 (nios2_split_symbolic_memory_operand): New.
10724 (nios2_legitimize_address): Code refactoring. Handle addresses
10725 involving symbolic constants.
10726 (nios2_emit_move_sequence): Likewise.
10727 (nios2_print_operand): Improve error output.
10728 (nios2_print_operand_address): Handle LO_SUM.
10729 (nios2_cdx_narrow_form_p): Likewise.
10730 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10731 operands involving symbolic constants.
10732 (movhi_internal, movsi_internal): Likewise.
10733 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10734 (extendhisi2, extendqi<mode>2): Likewise.
10736 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10738 * tree-pass.h (PROP_rtl_split_insns): Define.
10739 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10741 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10743 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10745 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10748 * target.def (const_not_ok_for_debug_p): Default to
10749 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10750 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10751 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10752 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10753 which targetm.const_not_ok_for_debug_p returned true.
10754 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10756 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10758 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10759 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10760 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10762 (ix86_const_not_ok_for_debug_p): New function.
10763 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10764 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10766 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10769 * system.h: Conditionally include "unique-ptr.h" if
10770 INCLUDE_UNIQUE_PTR is defined.
10771 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10772 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10774 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10776 * config/rl78/rl78.md: New define_expand "subdi3".
10778 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10781 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10782 DF_REF_INSN if DF_REF_INSN_INFO is false.
10784 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10786 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10787 xmm_move instead of sse_move.
10788 (sse_store_index): New function.
10789 (ix86_register_move_cost): Be more sensible about mismatch stall;
10790 model AVX moves correctly; make difference between sse->integer and
10792 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10793 moves; make difference between SSE and AVX.
10794 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10795 and zmm_move. Increase size of sse load and store tables;
10796 add unaligned load and store tables; add ssemmx_to_integer.
10797 * x86-tune-costs.h: Update all entries according to real
10798 move latencies from Agner Fog's manual and chip documentation.
10800 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10803 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
10804 * config/i386/constraints.md (Wf): New constraint.
10805 * config/i386/i386.md (UNSPEC_SBB): New unspec.
10806 (cmp<dwi>_doubleword): Removed.
10807 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
10808 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
10809 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
10810 expand with cmp<dwi>_doubleword. For LTU and GEU use
10811 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
10813 * common.opt (gcolumn-info): Enable by default.
10814 * doc/invoke.texi (gcolumn-info): Document new default.
10816 2017-10-23 Richard Biener <rguenther@suse.de>
10818 PR tree-optimization/82672
10819 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10820 Fold the stmt if we propagated into it.
10822 2017-10-23 Richard Biener <rguenther@suse.de>
10824 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
10825 (bitmap_remove_expr_from_set): ... this. All callers call this
10826 for non-constant values.
10827 (bitmap_set_subtract): Rename to...
10828 (bitmap_set_subtract_expressions): ... this. Adjust and
10830 (bitmap_set_contains_value): Remove superfluous check.
10831 (bitmap_set_replace_value): Inline into single caller ...
10832 (bitmap_value_replace_in_set): ... here and simplify.
10833 (dependent_clean): Merge into ...
10834 (clean): ... this using an overload. Adjust.
10835 (prune_clobbered_mems): Adjust.
10836 (compute_antic_aux): Likewise.
10837 (compute_partial_antic_aux): Likewise.
10839 2017-10-23 Richard Biener <rguenther@suse.de>
10841 PR tree-optimization/82129
10843 2017-08-01 Richard Biener <rguenther@suse.de>
10845 PR tree-optimization/81181
10846 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10847 (compute_antic): ... end of iteration here.
10849 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10851 * target.def (starting_frame_offset): New hook.
10852 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
10853 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
10854 * doc/tm.texi.in: Regenerate.
10855 * hooks.h (hook_hwi_void_0): Declare.
10856 * hooks.c (hook_hwi_void_0): New function.
10857 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
10858 STARTING_FRAME_OFFSET.
10859 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10860 * reload1.c (reload): Likewise.
10861 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10862 instead of STARTING_FRAME_OFFSET.
10863 * function.c (try_fit_stack_local): Likewise.
10864 (assign_stack_local_1): Likewise
10865 (instantiate_virtual_regs): Likewise.
10866 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
10867 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
10868 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
10869 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
10870 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
10871 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
10872 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
10873 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
10874 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
10875 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
10876 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
10877 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
10878 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
10879 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
10880 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
10881 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
10882 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
10883 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10884 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10885 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10886 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10887 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10888 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10889 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10890 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10891 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10892 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10893 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10894 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10895 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10896 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10897 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10898 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10899 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10900 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10901 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10902 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10903 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10904 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10905 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10906 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10907 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10908 return a HOST_WIDE_INT.
10909 (avr_builtin_setjmp_frame_value): Use it instead of
10910 STARTING_FRAME_OFFSET.
10911 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10912 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10913 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10915 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10916 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10917 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10918 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10919 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10920 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10921 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10922 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10923 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10924 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10925 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10926 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10928 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10929 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10930 * config/mips/mips.c (mips_compute_frame_info): Refer to
10931 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10932 (mips_starting_frame_offset): New function.
10933 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10934 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10935 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10936 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10937 and return a HOST_WIDE_INT.
10938 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10939 (mmix_initial_elimination_offset): Refer to
10940 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10941 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10942 * config/pa/pa.c (pa_starting_frame_offset): New function.
10943 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10944 (pa_expand_prologue): Likewise.
10945 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10946 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10947 !FRAME_GROWS_DOWNWARD handling to...
10948 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10949 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10950 !FRAME_GROWS_DOWNWARD handling to...
10951 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10952 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10953 !FRAME_GROWS_DOWNWARD handling to...
10954 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10955 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10957 (rs6000_starting_frame_offset): New function.
10958 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10959 !FRAME_GROWS_DOWNWARD handling to...
10960 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10961 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10962 !FRAME_GROWS_DOWNWARD handling to...
10963 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10964 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10965 !FRAME_GROWS_DOWNWARD handling to...
10966 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10967 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10968 (rs6000_starting_frame_offset): New function.
10969 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10970 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10971 * config/vax/vax.c (vax_starting_frame_offset): New function.
10972 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10973 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10974 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10975 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10976 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10977 * system.h (STARTING_FRAME_OFFSET): Poison.
10979 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10981 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10982 SCALAR_TYPE_MODE instead of TYPE_MODE.
10984 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10985 Alan Hayward <alan.hayward@arm.com>
10986 David Sherwood <david.sherwood@arm.com>
10988 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10990 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10991 Alan Hayward <alan.hayward@arm.com>
10992 David Sherwood <david.sherwood@arm.com>
10994 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10996 2017-10-23 Richard Biener <rguenther@suse.de>
10998 PR tree-optimization/82129
10999 * tree-ssa-pre.c (bitmap_set_and): Remove.
11000 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
11001 canonicalizing expressions in the set to those with lowest
11002 ID rather than taking that from the first edge.
11004 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11006 * combine.c (rtx_equal_for_field_assignment_p): Use
11007 byte_lowpart_offset.
11009 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11010 Alan Hayward <alan.hayward@arm.com>
11011 David Sherwood <david.sherwood@arm.com>
11013 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11014 to a SUBREG_PROMOTED_VAR.
11016 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11017 Alan Hayward <alan.hayward@arm.com>
11018 David Sherwood <david.sherwood@arm.com>
11020 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11021 (expand_debug_source_expr): Likewise.
11022 * combine.c (combine_simplify_rtx): Likewise.
11023 * cse.c (fold_rtx): Likewise.
11024 * optabs.c (expand_float): Likewise.
11025 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11026 (simplify_binary_operation_1): Likewise.
11028 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11029 Alan Hayward <alan.hayward@arm.com>
11030 David Sherwood <david.sherwood@arm.com>
11032 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11033 (record_promoted_value): Likewise.
11034 * expr.c (expand_expr_real_2): Likewise.
11035 * ree.c (update_reg_equal_equiv_notes): Likewise.
11036 (combine_set_extension): Likewise.
11037 * rtlanal.c (low_bitmask_len): Likewise.
11038 * simplify-rtx.c (neg_const_int): Likewise.
11039 (simplify_binary_operation_1): Likewise.
11041 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11042 Alan Hayward <alan.hayward@arm.com>
11043 David Sherwood <david.sherwood@arm.com>
11045 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11046 * regcprop.c (maybe_mode_change): Likewise.
11047 * reload1.c (alter_reg): Likewise.
11049 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11051 * inchash.h (inchash::hash::add_wide_int): New function.
11052 * lto-streamer-out.c (hash_tree): Use it.
11054 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11056 * inchash.h (inchash::hash::add_wide_int): Rename to...
11057 (inchash::hash::add_hwi): ...this.
11058 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11059 (polymorphic_call_target_hasher::hash): Likewise.
11060 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11061 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11062 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11063 * lto-streamer-out.c (hash_tree): Likewise.
11064 * optc-save-gen.awk: Likewise.
11065 * tree.c (add_expr): Likewise.
11067 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11070 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11071 for ordered inequality comparisons even with TARGET_IEEE_FP.
11073 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11076 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11077 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11078 Expand with cmp<dwi>_doubleword.
11080 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11082 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11083 List CET intrinsics.
11084 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11085 specific to -fcf-protection option.
11087 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11089 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11090 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11091 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11092 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11093 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11094 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11095 (extra_objs): Add cet.o for Linux/x86 targets.
11096 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11097 * config/i386/cet.c: New file.
11098 * config/i386/cetintrin.h: Likewise.
11099 * config/i386/t-cet: Likewise.
11100 * config/i386/cpuid.h (bit_SHSTK): New.
11101 (bit_IBT): Likewise.
11102 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11103 pass IBT and SHSTK bits.
11104 * config/i386/i386-builtin-types.def
11105 (VOID_FTYPE_UNSIGNED_PVOID): New.
11106 (VOID_FTYPE_UINT64_PVOID): Likewise.
11107 * config/i386/i386-builtin.def: Add CET intrinsics.
11108 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11109 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11110 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11111 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11113 * config/i386/i386.c (rest_of_insert_endbranch): New.
11114 (pass_data_insert_endbranch): Likewise.
11115 (pass_insert_endbranch): Likewise.
11116 (make_pass_insert_endbranch): Likewise.
11117 (ix86_notrack_prefixed_insn_p): Likewise.
11118 (ix86_target_string): Add -mibt, -mshstk flags.
11119 (ix86_option_override_internal): Add flag_cf_protection
11121 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11122 (ix86_print_operand): Add 'notrack' prefix output.
11123 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11124 (ix86_expand_builtin): Expand CET intrinsics.
11125 (x86_output_mi_thunk): Add 'endbranch' instruction.
11126 * config/i386/i386.h (TARGET_IBT): New.
11127 (TARGET_IBT_P): Likewise.
11128 (TARGET_SHSTK): Likewise.
11129 (TARGET_SHSTK_P): Likewise.
11130 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11131 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11132 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11133 (builtin_setjmp_setup): New pattern.
11134 (builtin_longjmp): Likewise.
11135 (rdssp<mode>): Likewise.
11136 (incssp<mode>): Likewise.
11137 (saveprevssp): Likewise.
11138 (rstorssp): Likewise.
11139 (wrss<mode>): Likewise.
11140 (wruss<mode>): Likewise.
11141 (setssbsy): Likewise.
11142 (clrssbsy): Likewise.
11143 (nop_endbr): Likewise.
11144 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11146 * config/i386/immintrin.h: Include <cetintrin.h>.
11147 * config/i386/linux-common.h
11148 (file_end_indicate_exec_stack_and_cet): New prototype.
11149 (TARGET_ASM_FILE_END): New.
11151 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11153 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11154 latencies instead of having separate table; make difference between
11155 integer and float costs.
11156 * i386.h (processor_costs): Remove scalar_stmt_cost,
11157 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11158 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11160 * x86-tune-costs.h: Remove entries which has been removed in
11161 procesor_costs from all tables; make cond_taken_branch_cost
11162 and cond_not_taken_branch_cost COST_N_INSNS based.
11164 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11166 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11168 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11170 * config/i386/i386.md (isa): Remove fma_avx512f.
11171 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11172 <avx512>_fmadd_<mode>_mask3<round_name>,
11173 <avx512>_fmsub_<mode>_mask<round_name>,
11174 <avx512>_fmsub_<mode>_mask3<round_name>,
11175 <avx512>_fnmadd_<mode>_mask<round_name>,
11176 <avx512>_fnmadd_<mode>_mask3<round_name>,
11177 <avx512>_fnmsub_<mode>_mask<round_name>,
11178 <avx512>_fnmsub_<mode>_mask3<round_name>,
11179 <avx512>_fmaddsub_<mode>_mask<round_name>,
11180 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11181 <avx512>_fmsubadd_<mode>_mask<round_name>,
11182 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11183 (*vec_widen_umult_even_v16si<mask_name>,
11184 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11185 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11187 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11189 * extend.texi: Add 'nocf_check' documentation.
11190 * gimple.texi: Add second parameter to
11191 gimple_build_call_from_tree.
11192 * invoke.texi: Add -fcf-protection documentation.
11193 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11195 2017-10-20 Richard Biener <rguenther@suse.de>
11197 PR tree-optimization/82473
11198 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11199 the largest input type.
11201 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11203 * c-attribs.c (handle_nocf_check_attribute): New function.
11204 (c_common_attribute_table): Add 'nocf_check' handling.
11205 * gimple-parser.c: Add second argument NULL to
11206 gimple_build_call_from_tree.
11207 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11208 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11210 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11211 * common.opt: Add fcf-protection flag.
11212 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11213 * flag-types.h: Add enum cf_protection_level.
11214 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11215 Add 'nocf_check' attribute propagation to gimple call.
11216 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11217 (gimple_build_call_from_tree): Update prototype.
11218 (gimple_call_nocf_check_p): New function.
11219 (gimple_call_set_nocf_check): Likewise.
11220 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11221 * ipa-icf.c: Add nocf_check attribute in statement hash.
11222 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11223 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11224 * toplev.c (process_options): Add flag_cf_protection handling.
11226 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11228 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11230 2017-10-20 Richard Biener <rguenther@suse.de>
11232 PR tree-optimization/82603
11233 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11234 remove false predicated stores.
11236 2017-10-20 Richard Biener <rguenther@suse.de>
11238 * graphite-isl-ast-to-gimple.c
11239 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11240 Remove return value and simplify, dump copied stmt after lhs
11242 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11243 Reduce dump verbosity.
11244 (gsi_insert_earliest): Likewise.
11245 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11246 * graphite.c (print_global_statistics): Adjust dumping.
11247 (print_graphite_scop_statistics): Likewise.
11248 (print_graphite_statistics): Do not dump loops here.
11249 (graphite_transform_loops): But here.
11251 2017-10-20 Nicolas Roche <roche@adacore.com>
11253 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11254 * configure: Regenerate.
11256 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11259 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11260 functions when optimizing replace GIMPLE_RETURN stmts with
11261 calls to __builtin_unreachable ().
11264 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11265 for -fsanitize=thread link of executables.
11266 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11267 link of executables.
11270 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11271 New mode iterators.
11272 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11273 define_insns for logical vector shifts to use VI248_AVX512BW
11274 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11275 condition, useless isa and prefix attributes. Change the first
11276 2 of these define_insns to ...
11277 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11278 define_insn for avx512vl.
11279 (<shift_insn><mode>3): ... and this, new define_insn without
11280 masking for non-avx512vl.
11283 * config/i386/sse.md (*andnot<mode>3,
11284 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11285 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11286 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11287 not applied use empty suffix even for TARGET_AVX512VL.
11288 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11289 is applied, supply evex,evex or evex,evex,evex instead of just
11292 2017-10-20 Julia Koval <julia.koval@intel.com>
11294 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11295 (OPTION_MASK_ISA_GFNI_UNSET): New.
11296 (ix86_handle_option): Handle OPT_mgfni.
11297 * config/i386/cpuid.h (bit_GFNI): New.
11298 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11299 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11300 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11301 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11302 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11303 * config/i386/i386.opt: Add mgfni.
11305 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11307 * config/msp430/msp430.c (msp430_option_override): Disable
11308 -fdelete-null-pointer-checks.
11309 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11311 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11313 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11314 of x87 and SSE instructions.
11316 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11318 * asan.c (create_cond_insert_point): Do not update edge count.
11319 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11320 (afdo_propagate_circuit): Likewise.
11321 (afdo_calculate_branch_prob): Likewise.
11322 (afdo_annotate_cfg): Likewise.
11323 * basic-block.h (struct edge_def): Remove count.
11324 (edge_def::count): New accessor.
11325 * bb-reorder.c (rotate_loop): Update.
11326 (find_traces_1_round): Update.
11327 (connect_traces): Update.
11328 (sanitize_hot_paths): Update.
11329 * cfg.c (unchecked_make_edge): Update.
11330 (make_single_succ_edge): Update.
11331 (check_bb_profile): Update.
11332 (dump_edge_info): Update.
11333 (update_bb_profile_for_threading): Update.
11334 (scale_bbs_frequencies_int): Update.
11335 (scale_bbs_frequencies_gcov_type): Update.
11336 (scale_bbs_frequencies_profile_count): Update.
11337 (scale_bbs_frequencies): Update.
11338 * cfganal.c (connect_infinite_loops_to_exit): Update.
11339 * cfgbuild.c (compute_outgoing_frequencies): Update.
11340 (find_many_sub_basic_blocks): Update.
11341 * cfgcleanup.c (try_forward_edges): Update.
11342 (try_crossjump_to_edge): Update
11343 * cfgexpand.c (expand_gimple_cond): Update
11344 (expand_gimple_tailcall): Update
11345 (construct_exit_block): Update
11346 * cfghooks.c (verify_flow_info): Update
11347 (redirect_edge_succ_nodup): Update
11348 (split_edge): Update
11349 (make_forwarder_block): Update
11350 (duplicate_block): Update
11351 (account_profile_record): Update
11352 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11353 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11354 * cfgloopmanip.c (scale_loop_profile): Update.
11356 (lv_adjust_loop_entry_edge): Update.
11357 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11358 (force_nonfallthru_and_redirect): Update.
11359 (purge_dead_edges): Update.
11360 (rtl_flow_call_edges_add): Update.
11361 * cgraphunit.c (init_lowered_empty_function): Update.
11362 (cgraph_node::expand_thunk): Update.
11363 * gimple-pretty-print.c (dump_probability): Update.
11364 (dump_edge_probability): Update.
11365 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11366 * haifa-sched.c (sched_create_recovery_edges): Update.
11367 * hsa-gen.c (convert_switch_statements): Update.
11368 * ifcvt.c (dead_or_predicable): Update.
11369 * ipa-inline-transform.c (inline_transform): Update.
11370 * ipa-split.c (split_function): Update.
11371 * ipa-utils.c (ipa_merge_profiles): Update.
11372 * loop-doloop.c (add_test): Update.
11373 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11374 * lto-streamer-in.c (input_cfg): Update.
11375 (input_function): Update.
11376 * lto-streamer-out.c (output_cfg): Update.
11377 * modulo-sched.c (sms_schedule): Update.
11378 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11379 * predict.c (maybe_hot_edge_p): Update.
11380 (unlikely_executed_edge_p): Update.
11381 (probably_never_executed_edge_p): Update.
11382 (dump_prediction): Update.
11383 (drop_profile): Update.
11384 (propagate_unlikely_bbs_forward): Update.
11385 (determine_unlikely_bbs): Update.
11386 (force_edge_cold): Update.
11387 * profile.c (compute_branch_probabilities): Update.
11388 * reg-stack.c (better_edge): Update.
11389 * shrink-wrap.c (handle_simple_exit): Update.
11390 * tracer.c (better_p): Update.
11391 * trans-mem.c (expand_transaction): Update.
11392 (split_bb_make_tm_edge): Update.
11393 * tree-call-cdce.c: Update.
11394 * tree-cfg.c (gimple_find_sub_bbs): Update.
11395 (gimple_split_edge): Update.
11396 (gimple_duplicate_sese_region): Update.
11397 (gimple_duplicate_sese_tail): Update.
11398 (gimple_flow_call_edges_add): Update.
11399 (insert_cond_bb): Update.
11400 (execute_fixup_cfg): Update.
11401 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11402 * tree-complex.c (expand_complex_div_wide): Update.
11403 * tree-eh.c (lower_resx): Update.
11404 (unsplit_eh): Update.
11405 (cleanup_empty_eh_move_lp): Update.
11406 * tree-inline.c (copy_edges_for_bb): Update.
11407 (freqs_to_counts): Update.
11408 (copy_cfg_body): Update.
11409 * tree-ssa-dce.c (remove_dead_stmt): Update.
11410 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11411 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11412 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11413 (unloop_loops): Update.
11414 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11415 * tree-ssa-loop-split.c (connect_loops): Update.
11416 (split_loop): Update.
11417 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11418 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11419 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11420 * tree-ssa-reassoc.c (branch_fixup): Update.
11421 * tree-ssa-tail-merge.c (replace_block_by): Update.
11422 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11423 (compute_path_counts): Update.
11424 (update_profile): Update.
11425 (recompute_probabilities): Update.
11426 (update_joiner_offpath_counts): Update.
11427 (estimated_freqs_path): Update.
11428 (freqs_to_counts_path): Update.
11429 (clear_counts_path): Update.
11430 (ssa_fix_duplicate_block_edges): Update.
11431 (duplicate_thread_path): Update.
11432 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11433 (case_bit_test_cmp): Update.
11434 (collect_switch_conv_info): Update.
11435 (gen_inbound_check): Update.
11436 (do_jump_if_equal): Update.
11437 (emit_cmp_and_jump_insns): Update.
11438 * tree-tailcall.c (decrease_profile): Update.
11439 (eliminate_tail_call): Update.
11440 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11441 (vect_do_peeling): Update.
11442 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11443 * ubsan.c (ubsan_expand_null_ifn): Update.
11444 (ubsan_expand_ptr_ifn): Update.
11445 * value-prof.c (gimple_divmod_fixed_value): Update.
11446 (gimple_mod_pow2): Update.
11447 (gimple_mod_subtract): Update.
11448 (gimple_ic): Update.
11449 (gimple_stringop_fixed_value): Update.
11451 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11454 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11456 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11458 PR rtl-optimization/82395
11459 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11460 based on non_spilled_static_chain_regno_p.
11462 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11464 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11465 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11466 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11468 2017-10-19 Martin Sebor <msebor@redhat.com>
11470 PR tree-optimization/82596
11471 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11473 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11475 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11476 (get_mem_refs_of_builtin_call): Likewise.
11477 * builtins.c (expand_builtin_apply): Adjust call to
11478 allocate_dynamic_stack_space.
11479 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11480 the third argument to allocate_dynamic_stack_space, otherwise -1.
11481 (expand_builtin): Deal with all alloca variants.
11482 (is_inexpensive_builtin): Likewise.
11483 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11484 * calls.c (special_function_p): Deal with all alloca variants.
11485 (initialize_argument_information): Adjust call to
11486 allocate_dynamic_stack_space.
11487 (expand_call): Likewise.
11488 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11489 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11490 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11491 use it for the stack usage computation.
11492 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11493 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11494 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11495 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11496 (in_loop_p): Remove first argument and useless check.
11497 (pass_walloca::execute): Remove useless test and adjust call to above.
11498 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11499 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11500 (gimplify_call_expr): Deal with all alloca variants.
11501 * hsa-gen.c (gen_hsa_alloca): Likewise.
11502 (gen_hsa_insns_for_call): Likewise.
11503 * ipa-pure-const.c (special_builtin_state): Likewise.
11504 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11505 * tree-object-size.c (alloc_object_size): Likewise.
11506 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11507 (call_may_clobber_ref_p_1): Likewise.
11508 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11509 (ccp_fold_stmt): Likewise.
11510 (optimize_stack_restore): Likewise.
11511 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11512 (mark_all_reaching_defs_necessary_1): Likewise.
11513 (propagate_necessity): Likewise.
11514 (eliminate_unnecessary_stmts): Likewise.
11515 * tree.c (build_common_builtin_nodes): Build
11516 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11517 (build_alloca_call_expr): New function.
11518 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11519 (CASE_BUILT_IN_ALLOCA): Likewise.
11520 (build_alloca_call_expr): Declare.
11521 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11523 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11526 * dwarf2out.c (new_die_raw): New static inline function.
11527 (new_die): Use it to create the DIE.
11528 (add_AT_external_die_ref): Likewise.
11529 (clone_die): Likewise.
11530 (clone_as_declaration): Likewise.
11531 (dwarf2out_vms_debug_main_pointer): Likewise.
11532 (base_type_die): Likewise. Remove early return for corner cases.
11533 Do not call add_pubtype on the DIE here.
11534 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11535 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11536 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11537 native order exists for base types, attach the DIE manually and call
11538 add_pubtype on it. Do not equate a reverse order DIE to the type.
11540 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11542 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11543 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11545 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11547 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11549 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11550 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11551 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11552 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11553 * opts.c (sanitizer_opts): Add builtin.
11554 * ubsan.c (instrument_builtin): New function.
11555 (pass_ubsan::execute): Call it.
11556 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11557 * doc/invoke.texi: Document -fsanitize=builtin.
11559 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11560 builtins, store max (log2 (align), 0) into uchar field instead of
11561 align into uptr field.
11562 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11563 store uchar 0 field instead of uptr 0 field.
11564 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11565 instead of passing one address of struct with 2 locations pass
11566 two addresses of structs with 1 location each.
11567 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11568 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11569 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11570 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11571 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11572 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11573 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11574 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11576 2017-10-19 Martin Liska <mliska@suse.cz>
11579 * file-find.c (remove_prefix): Remove.
11580 * file-find.h (remove_prefix): Likewise.
11581 * gcc-ar.c: Remove smartness of lookup.
11583 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11585 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11586 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11587 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11589 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11592 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11593 (setcc + and to xor + setcc): New peephole2.
11595 2017-10-19 Tom de Vries <tom@codesourcery.com>
11597 * doc/sourcebuild.texi (Test Directives, Variants of
11598 dg-require-support): Add dg-require-stack-size.
11600 2017-10-19 Martin Liska <mliska@suse.cz>
11603 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11604 that have a large alignment.
11605 (gimplify_target_expr): Likewise.
11607 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11609 PR rtl-optimization/82602
11610 * ira.c (rtx_moveable_p): Return false for volatile asm.
11612 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11615 * config/i386/i386-modes.def (CCGZ): New CC mode.
11616 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11617 * config/i386/predicates.md (ix86_comparison_operator):
11619 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11620 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11621 with double-word subtraction.
11622 (put_condition_code): Handle CCGZmode.
11624 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11626 * wide-int.cc (debug (const wide_int &)): New.
11627 (debug (const wide_int *)): New.
11628 (debug (const widest_int &)): New.
11629 (debug (const widest_int *)): New.
11631 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11633 PR middle-end/82556
11634 * lra-constraints.c (curr_insn_transform): Use non-input operand
11635 instead of output one for matched reload.
11637 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11639 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11640 (tree-ssa-loop-ivopts.h): New header file.
11641 (struct builtin_info): New fields.
11642 (classify_builtin_1): Compute and record base and offset parts for
11643 memset builtin partition by calling strip_offset.
11644 (offset_cmp, fuse_memset_builtins): New functions.
11645 (finalize_partitions): Fuse adjacent memset partitions by calling
11647 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11648 Expose the interface.
11649 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11651 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11653 PR tree-optimization/82574
11654 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11655 that data reference must be executed exactly once per iteration
11656 against the outermost loop in nest.
11657 (classify_partition): Update call to above function.
11659 2017-10-18 Richard Biener <rguenther@suse.de>
11661 PR tree-optimization/82591
11662 * graphite.c (graphite_transform_loops): Move code gen message
11664 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11665 Here. Handle scop_to_isl_ast failing.
11666 (scop_to_isl_ast): Limit the number of ISL operations.
11668 2017-10-18 Richard Biener <rguenther@suse.de>
11670 * graphite-isl-ast-to-gimple.c
11671 (translate_isl_ast_to_gimple::set_rename): Simplify.
11672 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11673 (graphite_copy_stmts_from_block): ... here.
11674 (copy_bb_and_scalar_dependences): Simplify.
11675 (add_parameters_to_ivs_params): Canonicalize.
11676 (generate_entry_out_of_ssa_copies): Simplify.
11677 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11678 by passing in ISL dimension.
11679 (parameter_index_in_region_1): Rename to ...
11680 (parameter_index_in_region): ... this.
11681 (extract_affine): Adjust assert, pass down parameter index.
11682 (add_param_constraints): Use range-info when available.
11683 (build_scop_context): Adjust.
11684 * sese.c (new_sese_info): Adjust.
11685 (free_sese_info): Likewise.
11686 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11687 Remove unused typedefs.
11688 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11690 2017-10-18 Martin Liska <mliska@suse.cz>
11692 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11694 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11696 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11697 (s390_sched_init): Do not reset s390_sched_state if we entered the
11698 current basic block via a fallthru edge and all others are unlikely.
11700 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11702 * config/s390/s390.c (NUM_SIDES): New variable.
11703 (LONGRUNNING_THRESHOLD): New variable.
11704 (LATENCY_FACTOR): New variable.
11705 (s390_sched_score): Decrease score for long-running instructions on
11707 (s390_sched_variable_issue): Perform bookkeeping for long-running
11710 2017-10-18 Richard Biener <rguenther@suse.de>
11712 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11713 Simplify with removal of the parameter rename map.
11714 (set_rename): Likewise.
11715 (should_copy_to_new_region): Likewise.
11716 (graphite_copy_stmts_from_block): Likewise.
11717 (copy_bb_and_scalar_dependences): Remove initialization of
11718 unused copied_bb_map.
11719 (copy_def): Remove.
11720 (copy_internal_parameters): Likewise.
11721 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11722 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11723 Use INTEGRAL_TYPE_P.
11724 (parameter_index_in_region_1): Rename to ...
11725 (assign_parameter_index_in_region): ... this. Assert we have
11726 a parameter we handle.
11727 (scan_tree_for_params): Adjust.
11728 * sese.h (parameter_rename_map_t): Remove.
11729 (struct sese_info_t): Remove unused parameter_rename_map and
11730 copied_bb_map members.
11731 * sese.c (new_sese_info): Adjust.
11732 (free_sese_info): Likewise.
11734 2017-10-18 Martin Liska <mliska@suse.cz>
11737 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11738 on an abnormal edge.
11740 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11742 * doc/invoke.texi (ffunction-sections and fdata-sections):
11745 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11747 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11748 the use statement can throw internally.
11750 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11752 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11753 any RTX present on the RHS of a SET.
11754 * compare-elim.c (try_eliminate_compare): Restore comment.
11756 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11758 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11759 may be also called on tcc_type nodes.
11760 * langhooks.c (lhd_tree_size): Likewise.
11762 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11764 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11765 format_warning_at_substring.
11766 (maybe_warn): Convert source_range * param to a location_t. Pass
11767 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11768 (format_directive): Remove code to extract source_ranges and
11769 source_range * in favor of just a location_t.
11770 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11772 * substring-locations.c (format_warning_va): Convert
11773 source_range * param to a location_t.
11774 (format_warning_at_substring): Likewise.
11775 * substring-locations.h (format_warning_va): Likewise.
11776 (format_warning_at_substring): Likewise.
11778 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11780 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11782 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11783 and scatter/gather ops.
11785 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11786 vec_gather_load and vec_scatter_store.
11787 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11788 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11789 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11790 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11791 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11792 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11794 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11796 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11797 Detect FTST instruction and handle its register pops. Only pop
11798 second operand if can_pop_second_op is true.
11799 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
11800 set can_pop_second_op to false in the compare_for_stack_reg call.
11802 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
11803 output_fp_compare for stack register operands.
11804 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
11805 instructions here. Do not emit stack register pops here. Assert
11806 that FCOMPP pops next to top stack register. Rewrite function.
11808 2017-10-17 Nathan Sidwell <nathan@acm.org>
11810 PR middle-end/82577
11811 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
11812 use DECL_ASSEMBLER_NAME_RAW.
11814 PR middle-end/82546
11815 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
11818 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
11819 Wilco Dijkstra <wilco.dijkstra@arm.com>
11821 * builtins.c (expand_builtin_update_setjmp_buf): Add a
11822 converstion to Pmode from the buf_addr.
11824 2017-10-17 Richard Biener <rguenther@suse.de>
11826 * graphite-dependences.c (scop_get_reads_and_writes): Change
11827 output parameters to references.
11829 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11831 PR 71026/tree-optimization
11832 * fold-const.c (distribute_real_division): Removed.
11833 (fold_binary_loc): Remove calls to distribute_real_divison.
11835 2017-10-17 Richard Biener <rguenther@suse.de>
11837 * graphite-scop-detection.c
11838 (scop_detection::stmt_has_simple_data_refs_p): Always use
11839 the full nest as region.
11840 (try_generate_gimple_bb): Likewise.
11841 * sese.c (scalar_evolution_in_region): Simplify now that
11842 SCEV can handle instantiation in regions.
11843 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11844 in the non-loop part of a function if requested.
11846 2017-10-17 Richard Biener <rguenther@suse.de>
11848 PR tree-optimization/82563
11849 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11851 (graphite_regenerate_ast_isl): Call it.
11852 * graphite-scop-detection.c (build_scops): Remove entry edge split.
11854 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11856 PR tree-optimization/82549
11857 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
11858 Formatting fixes. Instead of calling make_bit_field_ref with negative
11861 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11863 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11864 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
11865 _mm_maskz_reduce_ss): New.
11866 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
11867 __builtin_ia32_reducess_mask): Ditto..
11868 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
11869 * config/i386/sse.md (reduces<mode>): Renamed to ...
11870 (reduces<mode><mask_scalar_name>): ... this.
11871 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
11873 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
11874 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
11876 2017-10-16 David Malcolm <dmalcolm@redhat.com>
11878 * Makefile.in (OBJS): Add unique-ptr-tests.o.
11879 * selftest-run-tests.c (selftest::run_tests): Call
11880 selftest::unique_ptr_tests_cc_tests.
11881 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
11882 * unique-ptr-tests.cc: New file.
11884 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11887 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11889 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11890 (make_hard_regno_born, make_hard_regno_dead): Update
11891 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11893 2017-10-16 Jeff Law <law@redhat.com>
11895 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11897 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11899 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11901 2017-10-16 Olivier Hainque <hainque@adacore.com>
11903 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11904 with_cpu if we were configured for an e500v2 target cpu name.
11906 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11908 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11909 * doc/invoke.texi: Document +nodsp as a valid extension for
11912 2017-10-16 Martin Liska <mliska@suse.cz>
11914 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11915 (test_set_range): Likewise.
11916 (test_range_functions): Rename to ...
11917 (test_bit_in_range): ... this.
11918 (sbitmap_c_tests): Add new test.
11920 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11922 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11924 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11925 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11927 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11929 * config/aarch64/aarch64-builtins.c
11930 (aarch64_types_quadopu_lane_qualifiers): New.
11931 (TYPES_QUADOPU_LANE): New.
11932 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11933 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11934 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11935 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11936 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11937 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11938 (Vdottype, DOTPROD): New.
11939 (sur): Add SDOT and UDOT.
11941 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11943 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11944 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11945 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11946 Add TARGET_DOTPROD.
11947 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11948 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11949 Enable TARGET_DOTPROD.
11950 (cortex-a75.cortex-a55): Likewise.
11951 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11953 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11955 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11956 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11958 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11960 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11961 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11962 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11963 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11964 * config/arm/types.md (neon_dot, neon_dot_q): New.
11965 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11967 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11969 * config/arm/arm.h (TARGET_DOTPROD): New.
11970 * config/arm/arm.c (arm_arch_dotprod): New.
11971 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11972 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11973 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11974 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11975 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11976 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11977 * doc/invoke.texi (armv8.2-a): Document dotprod
11979 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11981 * i386.c (ix86_vec_cost): New function.
11982 (ix86_rtx_costs): Handle vector operations better.
11983 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11984 * x86-tune-costs.h: Add new costs to all tables.
11986 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11988 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11990 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11991 divsd, sqrtss and sqrtsd
11992 * x86-tune-costs.h: Add new entries to all costs.
11993 (znver1_cost): Fix to match real instruction latencies.
11995 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11996 Michael Collison <michael.collison@arm.com>
11998 * compare-elim.c: Include emit-rtl.h.
11999 (can_merge_compare_into_arith): New function.
12000 (try_validate_parallel): Likewise.
12001 (try_merge_compare): Likewise.
12002 (try_eliminate_compare): Call the above when no previous clobber
12004 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
12007 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12009 PR middle-end/62263
12010 PR middle-end/82498
12011 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12012 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12014 PR middle-end/62263
12015 PR middle-end/82498
12016 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12017 to be any operand_equal_p operands. For & (B - 1) require
12018 B to be power of 2. Recognize
12019 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12021 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12024 * optabs.c (expand_memory_blockage): Fix call of
12025 targetm.have_memory_blockage.
12027 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12030 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12031 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12032 objects to extra_objs instead of overwriting it.
12034 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12036 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12037 Use any_fp_register_operand as operand[3] predicate. Simplify
12038 equality test for operands[2] and operands[4] memory location.
12039 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12040 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12041 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12042 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12043 any_fp_register_operand as operand[1] predicate. Simplify
12044 equality test for operands[0] and operands[3] memory location.
12045 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12046 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12047 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12049 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12051 * target-insns.def: Add memory_blockage.
12052 * optabs.c (expand_memory_blockage): New function.
12053 (expand_asm_memory_barrier): Rename ...
12054 (expand_asm_memory_blockage): ... to this.
12055 (expand_mem_thread_fence): Call expand_memory_blockage
12056 instead of expand_asm_memory_barrier.
12057 (expand_mem_singnal_fence): Ditto.
12058 (expand_atomic_load): Ditto.
12059 (expand_atomic_store): Ditto.
12060 * doc/md.texi (Standard Pattern Names For Generation):
12061 Document memory_blockage instruction pattern.
12063 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12065 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12066 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12067 * config/rl78/rl78.md: New define_expand "adddi3".
12069 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12071 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12074 2017-10-13 Jeff Law <law@redhat.com>
12076 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12078 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12081 * internal-fn.c (expand_mul_overflow): If both operands have
12082 the same highpart of -1 or 0 and the topmost bit of lowpart
12083 is different, overflow is if res <= 0 rather than res < 0.
12085 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12087 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12088 TARGET_P9_VECTOR code for unaligned_load case.
12090 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12092 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12094 2017-10-13 Nathan Sidwell <nathan@acm.org>
12096 * tree-core.h (tree_contains_struct): Make bool.
12097 * tree.c (tree_contains_struct): Likewise.
12098 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12099 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12100 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12101 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12102 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12104 2017-10-13 Richard Biener <rguenther@suse.de>
12106 * graphite-isl-ast-to-gimple.c
12107 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12108 parameters and dominance check.
12109 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12110 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12111 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12112 Do not update SSA form here or do intermediate IL verification.
12113 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12114 (graphite_initialize): Remove check on the number of loops in
12115 the function and inline into graphite_transform_loops.
12116 (graphite_finalize): Inline into graphite_transform_loops.
12117 (graphite_transform_loops): Perform SSA update and IL verification
12119 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12121 2017-10-13 Richard Biener <rguenther@suse.de>
12123 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12124 graphite_expression_type_precision): Avoid global constructor
12126 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12127 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12128 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12129 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12130 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12131 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12133 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12136 * config/i386/i386.h (ix86_red_zone_size): New.
12137 * config/i386/i386.md (push peephole2s): Replace
12138 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12140 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12141 Alan Hayward <alan.hayward@arm.com>
12142 David Sherwood <david.sherwood@arm.com>
12144 * combine.c (can_change_dest_mode): Reject changes in
12145 REGMODE_NATURAL_SIZE.
12147 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12148 Alan Hayward <alan.hayward@arm.com>
12149 David Sherwood <david.sherwood@arm.com>
12151 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12152 (expand_debug_source_expr): Likewise.
12153 * combine.c (combine_simplify_rtx): Likewise.
12154 * cse.c (fold_rtx): Likewise.
12155 * fwprop.c (canonicalize_address): Likewise.
12156 * targhooks.c (default_shift_truncation_mask): Likewise.
12158 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12159 Alan Hayward <alan.hayward@arm.com>
12160 David Sherwood <david.sherwood@arm.com>
12162 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12163 (widened_mode): Likewise.
12164 (expand_unop): Likewise.
12165 * ree.c (transform_ifelse): Likewise.
12166 (merge_def_and_ext): Likewise.
12167 (combine_reaching_defs): Likewise.
12168 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12170 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12171 Alan Hayward <alan.hayward@arm.com>
12172 David Sherwood <david.sherwood@arm.com>
12174 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12175 * combine.c (gen_lowpart_for_combine): Likewise.
12176 * dwarf2out.c (rtl_for_decl_location): Likewise.
12177 * final.c (alter_subreg): Likewise.
12178 * rtlhooks.c (gen_lowpart_general): Likewise.
12179 (gen_lowpart_if_possible): Likewise.
12181 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12182 Alan Hayward <alan.hayward@arm.com>
12183 David Sherwood <david.sherwood@arm.com>
12185 * calls.c (expand_call): Use subreg_lowpart_offset.
12186 * cse.c (cse_insn): Likewise.
12187 * regcprop.c (copy_value): Likewise.
12188 (copyprop_hardreg_forward_1): Likewise.
12190 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12193 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12194 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12195 =Q constraints to +Q and into insn condition add check
12196 that operands[0] and operands[1] are equal.
12197 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12198 =Q constraints to +Q and into insn condition add check
12199 that operands[0] is equal to either operands[1] or operands[2].
12202 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12203 instead of handling MINUS_EXPR twice (once for each argument),
12204 canonicalize operand order and handle just once, use rtype where
12205 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12208 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12209 any values of __C while still being pattern recognizable as a simple
12210 rotate instruction.
12212 2017-10-13 Richard Biener <rguenther@suse.de>
12214 PR tree-optimization/82451
12216 2017-10-02 Richard Biener <rguenther@suse.de>
12218 PR tree-optimization/82355
12219 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12220 a mapping for the enclosing loop but avoid generating one for
12221 the loop tree root.
12222 (copy_bb_and_scalar_dependences): Remove premature codegen
12223 error on PHIs in blocks duplicated into multiple places.
12224 * graphite-scop-detection.c
12225 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12226 in the region use it as loop and nest to analyze the DR in.
12227 (try_generate_gimple_bb): Likewise.
12228 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12229 (add_loop_constraints): For blocks in a loop not in the region
12230 create a dimension with a single iteration.
12231 * sese.h (gbb_loop_at_index): Remove assert.
12233 * cfgloop.c (loop_preheader_edge): For the loop tree root
12234 return the single successor of the entry block.
12235 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12236 Reset the SCEV hashtable and niters.
12237 * graphite-scop-detection.c
12238 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12239 assert that we only have POLYNOMIAL_CHREC that vary in loops
12240 contained in the region.
12241 (scop_detection::graphite_can_represent_expr): Adjust.
12242 (scop_detection::stmt_has_simple_data_refs_p): For loops
12243 not in the region set loop to NULL. The nest is now the
12244 entry edge to the region.
12245 (try_generate_gimple_bb): Likewise.
12246 * sese.c (scalar_evolution_in_region): Adjust for
12247 instantiate_scev change.
12248 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12249 Make nest parameter the edge into the region.
12250 (create_data_ref): Likewise.
12251 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12252 entry edge into a region and adjust instantiate_scev calls.
12253 (create_data_ref): Likewise.
12254 (graphite_find_data_references_in_stmt): Likewise.
12255 (find_data_references_in_stmt): Pass the loop preheader edge
12256 from the nest argument.
12257 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12258 parameter the edge into the region.
12259 (instantiate_parameters): Use the loop preheader edge as entry.
12260 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12262 (get_instantiated_value_entry): Make instantiate_below parameter
12263 the edge into the region.
12264 (instantiate_scev_name): Likewise. Adjust dominance checks,
12265 when we cannot use loop-based instantiation instantiate by
12266 walking use-def chains.
12267 (instantiate_scev_poly): Adjust.
12268 (instantiate_scev_binary): Likewise.
12269 (instantiate_scev_convert): Likewise.
12270 (instantiate_scev_not): Likewise.
12271 (instantiate_array_ref): Remove.
12272 (instantiate_scev_3): Likewise.
12273 (instantiate_scev_2): Likewise.
12274 (instantiate_scev_1): Likewise.
12275 (instantiate_scev_r): Do not blindly handle N-operand trees.
12276 Do not instantiate array-refs. Handle all constants and invariants.
12277 (instantiate_scev): Make instantiate_below parameter
12278 the edge into the region.
12279 (resolve_mixers): Use the loop preheader edge for the region
12280 parameter to instantiate_scev_r.
12281 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12283 2017-10-13 Richard Biener <rguenther@suse.de>
12285 PR tree-optimization/82525
12286 * graphite-isl-ast-to-gimple.c
12287 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12289 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12290 Fail code generation when we cannot represent the isl integer.
12291 (binary_op_to_tree): Elide modulo operations that are no-ops
12292 in the type we code generate. Remove now superfluous code
12295 2017-10-13 Richard Biener <rguenther@suse.de>
12297 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12298 (scop_detection::harmful_loop_in_region): Remove premature
12299 IV type restriction.
12300 (scop_detection::graphite_can_represent_scev): We can handle
12301 pointer IVs just fine.
12303 2017-10-13 Alan Modra <amodra@gmail.com>
12305 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12306 "Clobbers and Scratch Registers". Add paragraph on
12307 alternative to clobbers for scratch registers and OpenBLAS
12310 2017-10-13 Alan Modra <amodra@gmail.com>
12312 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12313 example of a memory input for a string of known length. Move
12314 commentary out of table. Add a number of new examples
12315 covering array memory inputs.
12317 2017-10-12 Martin Liska <mliska@suse.cz>
12319 PR tree-optimization/82493
12320 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12321 (test_range_functions): New function.
12322 (sbitmap_c_tests): Likewise.
12323 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12324 * selftest.h (sbitmap_c_tests): New function.
12326 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12328 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12330 * config/rs6000/amo.h: Fix spacing issue.
12332 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12335 * config/i386/i386.md (*ashl<mode>3_mask_1,
12336 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12337 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12340 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12342 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12343 (profile_probability): Set max_probability
12344 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12346 * profile-count.c (profile_probability::differs_from_p): Do not
12347 rely on max_probaiblity == 10000
12349 2017-10-12 Jeff Law <law@redhat.com>
12351 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12354 2017-10-12 Martin Sebor <msebor@redhat.com>
12358 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12359 (handle_alias_pairs): Call it.
12360 * common.opt (-Wattribute-alias): New option.
12361 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12362 * doc/invoke.texi (-Wattribute-alias): Document.
12364 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12367 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12369 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12371 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12372 (make_hard_regno_born, make_hard_regno_dead): Update
12373 bb_killed_pseudos and bb_gen_pseudos.
12375 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12377 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12379 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12381 * config/alpha/alpha.c (alpha_split_conditional_move):
12382 Use std::swap instead of manually swapping.
12383 (alpha_stdarg_optimize_hook): Ditto.
12384 (alpha_canonicalize_comparison): Ditto.
12386 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12388 * tree-loop-distribution.c (struct builtin_info): New struct.
12389 (struct partition): Refactor fields into struct builtin_info.
12390 (partition_free): Free struct builtin_info.
12391 (build_size_arg_loc, build_addr_arg_loc): Delete.
12392 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12393 information from struct builtin_info.
12394 (find_single_drs): New function refactored from classify_partition.
12395 Also moved builtin validity checks to this function.
12396 (compute_access_range, alloc_builtin): New functions.
12397 (classify_builtin_st, classify_builtin_ldst): New functions.
12398 (classify_partition): Refactor code into functions find_single_drs,
12399 classify_builtin_st and classify_builtin_ldst.
12400 (distribute_loop): Don't do runtime alias check when distributing
12402 (find_seed_stmts_for_distribution): New function.
12403 (pass_loop_distribution::execute): Refactor code finding seed
12404 stmts into above function. Support distribution for the innermost
12405 two-level loop nest. Adjust dump information.
12407 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12409 * tree-loop-distribution.c: Adjust the general comment.
12410 (NUM_PARTITION_THRESHOLD): New macro.
12411 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12412 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12413 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12414 in call to build_partition_graph.
12415 (finalize_partitions): New parameter. Make loop distribution more
12416 conservative by fusing more partitions.
12417 (distribute_loop): Don't do runtime alias check in case of loop nest
12419 (find_seed_stmts_for_distribution): New function.
12420 (prepare_perfect_loop_nest): New function.
12421 (pass_loop_distribution::execute): Refactor code finding seed stmts
12422 and loop nest into above functions. Support loop nest distribution.
12423 Adjust dump information accordingly.
12425 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12427 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12428 and set PTYPE_SEQUENTIAL for merged partition.
12430 2017-10-12 Richard Biener <rguenther@suse.de>
12432 PR tree-optimization/69728
12434 2017-09-19 Richard Biener <rguenther@suse.de>
12436 PR tree-optimization/69728
12437 * graphite-sese-to-poly.c (schedule_error): New global.
12438 (add_loop_schedule): Handle empty domain by failing the
12440 (build_original_schedule): Handle schedule_error.
12442 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12443 domain by returning an unchanged schedule.
12445 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12447 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12448 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12450 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12452 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12455 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12458 * expr.c (store_field): Don't optimize away bitsize == 0 store
12459 from CALL_EXPR with addressable return type.
12461 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12463 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12464 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12465 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12466 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12467 TARGET_ISEL instead of TARGET_ISEL<sel>.
12469 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12471 * config/rs6000/rs6000.c
12472 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12474 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12476 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12478 (reg_or_cint_operand): Fix comment.
12479 (reg_or_zero_operand): New predicate.
12480 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12481 * config/rs6000/rs6000.c (output_isel): Delete.
12482 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12483 instead of reg_or_cint_operand. Output instruction directly (not via
12485 (isel_unsigned_<mode>): Ditto.
12486 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12487 gpc_reg_operand. Add an instruction alternative for this. Output
12488 instruction directly.
12489 (*isel_reversed_unsigned_<mode>): Ditto.
12491 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12493 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12494 (TARGET_CANONICALIZE_COMPARISON): Define.
12496 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12499 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12500 Check whether the dest is REG before adding REG_EQUIV note.
12502 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12505 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12507 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12508 (make_hard_regno_born, make_hard_regno_dead): Update
12509 bb_killed_pseudos and bb_gen_pseudos.
12511 2017-10-11 Nathan Sidwell <nathan@acm.org>
12513 * incpath.h (enum incpath_kind): Name enum, prefix values.
12514 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12515 * incpath.c (heads, tails): Use INC_MAX.
12516 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12517 (merge_include_chains, split_quote_chain,
12518 register_include_chains): Update incpath_kind names.
12519 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12520 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12522 (add_framework_path, darwin_register_objc_includes): Likewise.
12523 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12525 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12527 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12528 Do not use float_operator operator predicate.
12529 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12530 * config/i386/predicates.md (float_operator): Remove predicate.
12532 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12534 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12535 (*jccxf_i387): Ditto.
12536 (*jcc<mode>_i387): Ditto.
12537 (*jccu<mode>_i387): Ditto.
12538 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12539 (*jcc_*_i387 splitters): Remove.
12540 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12541 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12542 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12545 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12547 * profile-count.h (slow_safe_scale_64bit): New function.
12548 (safe_scale_64bit): New inline.
12549 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12550 * profile-count.c: Include wide-int.h
12551 (slow_safe_scale_64bit): New.
12553 2017-10-11 Nathan Sidwell <nathan@acm.org>
12555 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12556 HAS_DECL_ASSEMBLER_NAME_P.
12557 * gimple-expr.c (gimple_decl_printable_name: Check
12558 HAS_DECL_ASSEMBLER_NAME_P too.
12559 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12560 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12561 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12562 * passes.c (rest_of_decl_compilation): Check
12563 HAS_DECL_ASSEMBLER_NAME_P too.
12564 * recog.c (verify_changes): Likewise.
12565 * tree-pretty-print.c (dump_decl_name): Likewise.
12566 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12568 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12569 (SET_DECL_ASSEMBLER_NAME): Use it.
12570 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12571 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12572 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12574 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12576 * config.gcc (i386, x86_64): Add extra objects.
12577 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12578 (ix86_min_insn_size): Declare.
12579 (ix86_issue_rate): Declare.
12580 (ix86_adjust_cost): Declare.
12581 (ia32_multipass_dfa_lookahead): Declare.
12582 (ix86_macro_fusion_p): Declare.
12583 (ix86_macro_fusion_pair_p): Declare.
12584 (ix86_bd_has_dispatch): Declare.
12585 (ix86_bd_do_dispatch): Declare.
12586 (ix86_core2i7_init_hooks): Declare.
12587 (ix86_atom_sched_reorder): Declare.
12588 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12589 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12590 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12591 (rip_relative_addr_p): Rename to ...
12592 (ix86_rip_relative_addr_p): ... this one; export.
12593 (memory_address_length): Update.
12594 (ix86_issue_rate): Move to x86-tune-sched.c.
12595 (ix86_flags_dependent): Move to x86-tune-sched.c.
12596 (ix86_agi_dependent): Move to x86-tune-sched.c.
12597 (exact_dependency_1): Move to x86-tune-sched.c.
12598 (exact_store_load_dependency): Move to x86-tune-sched.c.
12599 (ix86_adjust_cost): Move to x86-tune-sched.c.
12600 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12601 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12602 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12603 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12604 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12605 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12606 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12607 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12608 (min_insn_size): Rename to ...
12609 (ix86_min_insn_size): ... this one; export.
12610 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12611 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12612 (core2i7_first_cycle_multipass_backtrack): Move to
12613 x86-tune-sched-core.c.
12614 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12615 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12616 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12617 (ix86_avoid_jump_mispredicts): Update.
12618 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12619 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12620 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12621 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12622 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12623 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12624 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12625 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12626 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12627 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12628 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12629 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12630 (BIG): Move to ix86-tune-sched-bd.c.
12631 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12632 (enum insn_path): Move to ix86-tune-sched-bd.c.
12633 (get_mem_group): Move to ix86-tune-sched-bd.c.
12634 (is_cmp): Move to ix86-tune-sched-bd.c.
12635 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12636 (is_branch): Move to ix86-tune-sched-bd.c.
12637 (is_prefetch): Move to ix86-tune-sched-bd.c.
12638 (init_window): Move to ix86-tune-sched-bd.c.
12639 (allocate_window): Move to ix86-tune-sched-bd.c.
12640 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12641 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12642 (process_end_window): Move to ix86-tune-sched-bd.c.
12643 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12644 (find_constant): Move to ix86-tune-sched-bd.c.
12645 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12646 (has_immediate): Move to ix86-tune-sched-bd.c.
12647 (get_insn_path): Move to ix86-tune-sched-bd.c.
12648 (get_insn_group): Move to ix86-tune-sched-bd.c.
12649 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12650 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12651 (add_insn_window): Move to ix86-tune-sched-bd.c.
12652 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12653 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12654 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12655 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12656 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12657 (do_dispatch): Move to ix86-tune-sched-bd.c.
12658 (has_dispatch): Move to ix86-tune-sched-bd.c.
12659 * config/i386/t-i386: Add new object files.
12660 * config/i386/x86-tune-costs.h: New file.
12661 * config/i386/x86-tune-sched-atom.c: New file.
12662 * config/i386/x86-tune-sched-bd.c: New file.
12663 * config/i386/x86-tune-sched-core.c: New file.
12664 * config/i386/x86-tune-sched.c: New file.
12666 2017-10-11 Liu Hao <lh_mouse@126.com>
12668 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12669 the generic version below instead.
12670 (should_colorize): Recognize Windows consoles as terminals
12672 * pretty-print.c [__MINGW32__] (write_all): New function.
12673 [__MINGW32__] (find_esc_head): Likewise.
12674 [__MINGW32__] (find_esc_terminator): Likewise.
12675 [__MINGW32__] (eat_esc_sequence): Likewise.
12676 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12678 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12681 2017-10-11 Richard Biener <rguenther@suse.de>
12683 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12684 Properly call analyze_scalar_evolution with the loop of the stmt.
12686 2017-10-11 Richard Biener <rguenther@suse.de>
12688 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12689 * tree-core.h (tree_base): Add chrec_var union member.
12690 * tree.h (CHREC_VAR): Remove.
12691 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12692 * tree-chrec.h (build_polynomial_chrec): Adjust.
12693 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12694 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12696 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12698 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12699 * match.pd: ... here.
12700 ((T) X == (T) Y): Relax condition.
12702 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12704 PR tree-optimization/82472
12705 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12707 (break_alias_scc_partitions): Update postorder number.
12709 2017-10-11 Martin Liska <mliska@suse.cz>
12712 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12714 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12716 2017-10-11 Martin Liska <mliska@suse.cz>
12718 * print-rtl.c (print_insn): Move declaration of idbuf
12719 to same scope as name.
12721 2017-10-11 Martin Liska <mliska@suse.cz>
12726 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12727 volatile arguments.
12729 2017-10-11 Martin Liska <mliska@suse.cz>
12732 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12733 volatile arguments.
12735 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12737 * config.gcc (default_gnu_indirect_function): Default to yes for
12738 arm*-*-linux* with glibc.
12740 2017-10-11 Richard Biener <rguenther@suse.de>
12742 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12743 default-defs and types we do not want to analyze.
12744 (interpret_loop_phi): Replace unreachable code with an assert.
12745 (compute_scalar_evolution_in_loop): Remove and inline ...
12746 (analyze_scalar_evolution_1): ... here, replacing condition with
12747 what makes the intent clearer. Remove handling of cases
12748 get_scalar_evolution now handles.
12750 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12752 PR rtl-optimization/81434
12753 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12754 comment for main loop. In sched_group_found if, also add checks for
12755 pass and min_cost_group.
12757 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12759 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12760 (rs6000_insn_cost): New function.
12761 * config/rs6000/rs6000.md (cost): New attribute.
12763 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12764 H.J. Lu <hongjiu.lu@intel.com>
12768 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12769 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12770 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12771 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12772 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12773 OPTION_MASK_ISA_MMX and some other bit set require both
12774 mmx and the ISAs without the mmx bit.
12775 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12776 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12777 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12778 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12779 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12780 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12781 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12782 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12783 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12784 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12785 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12786 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12787 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12788 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12789 Add OPTION_MASK_ISA_MMX.
12791 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12793 * config.gcc (armv7*-*-freebsd*): New target.
12794 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12796 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12798 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
12799 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
12802 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
12804 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
12805 a template parameter.
12806 (WIDE_INT_REF_FOR): Update accordingly.
12807 * tree.h (wi::int_traits <const_tree>): Delete.
12808 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
12809 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
12810 (wi::tree_to_wide_ref): New typedef.
12811 (wi::to_wide): New function.
12812 * calls.c (get_size_range): Use wi::to_wide when operating on
12813 trees as wide_ints.
12814 * cgraph.c (cgraph_node::create_thunk): Likewise.
12815 * config/i386/i386.c (ix86_data_alignment): Likewise.
12816 (ix86_local_alignment): Likewise.
12817 * dbxout.c (stabstr_O): Likewise.
12818 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
12819 * expr.c (const_vector_from_tree): Likewise.
12820 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
12821 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
12822 (fold_negate_expr_1, int_const_binop_1, const_binop)
12823 (fold_convert_const_int_from_real, optimize_bit_field_compare)
12824 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
12825 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
12826 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12827 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12828 (fold_not_const, round_up_loc): Likewise.
12829 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12830 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12831 (alloca_call_type): Likewise.
12832 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12833 * godump.c (go_output_typedef): Likewise.
12834 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12835 * internal-fn.c (get_min_precision): Likewise.
12836 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12837 * ipa-polymorphic-call.c
12838 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12839 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12840 (ipa_modify_call_arguments): Likewise.
12841 * match.pd: Likewise.
12842 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12843 * print-tree.c (print_node_brief, print_node): Likewise.
12844 * stmt.c (expand_case): Likewise.
12845 * stor-layout.c (layout_type): Likewise.
12846 * tree-affine.c (tree_to_aff_combination): Likewise.
12847 * tree-cfg.c (group_case_labels_stmt): Likewise.
12848 * tree-data-ref.c (dr_analyze_indices): Likewise.
12849 (prune_runtime_alias_test_list): Likewise.
12850 * tree-dump.c (dequeue_and_dump): Likewise.
12851 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
12852 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
12853 * tree-pretty-print.c (dump_generic_node): Likewise.
12854 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
12855 (simple_iv_with_niters): Likewise.
12856 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
12857 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
12858 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
12859 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12860 (refine_value_range_using_guard, number_of_iterations_ne_max)
12861 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12862 (get_cst_init_from_scev, record_nonwrapping_iv)
12863 (scev_var_range_cant_overflow): Likewise.
12864 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
12865 * tree-ssa-pre.c (compute_avail): Likewise.
12866 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
12867 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
12868 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
12869 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
12870 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
12871 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
12872 (dump_case_nodes, try_switch_expansion): Likewise.
12873 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
12874 (vect_do_peeling): Likewise.
12875 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
12876 * tree-vect-stmts.c (vectorizable_load): Likewise.
12877 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
12878 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
12879 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
12880 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
12881 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
12882 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
12883 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12884 (evrp_dom_walker::before_dom_children): Likewise.
12885 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12886 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12887 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12888 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12889 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12890 (get_range_pos_neg): Likewise.
12891 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12892 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12893 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12894 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12895 * config/avr/avr.c (avr_fold_builtin): Likewise.
12896 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12897 * config/msp430/msp430.c (msp430_attr): Likewise.
12898 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12899 * config/powerpcspe/powerpcspe-c.c
12900 (altivec_resolve_overloaded_builtin): Likewise.
12901 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12902 (rs6000_expand_ternop_builtin): Likewise.
12903 * config/rs6000/rs6000-c.c
12904 (altivec_resolve_overloaded_builtin): Likewise.
12905 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12906 (rs6000_expand_ternop_builtin): Likewise.
12907 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12909 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12911 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12912 when copying loop nest with only one inner loop.
12914 2017-10-10 Richard Biener <rguenther@suse.de>
12916 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12917 blocks if SCEV is active.
12918 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12920 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12921 (scev_initialize): Assert we are not yet initialized.
12923 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12925 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12926 inner loop's exit stmt by making it always exit the loop, otherwise
12927 we would generate an infinite empty loop.
12929 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12931 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12932 renaming variables in new preheader if it's deleted.
12934 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12936 * tree-loop-distribution.c (struct partition): Remove unused field
12937 loops of the structure.
12938 (partition_alloc, partition_free): Ditto.
12939 (build_rdg_partition_for_vertex): Ditto.
12941 2017-10-09 Jeff Law <law@redhat.com>
12943 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12944 return type to match prototype and documentation.
12946 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12948 * config/rs6000/rs6000.c (processor_costs): Move to ...
12949 * config/rs6000/rs6000.h: ... here.
12950 (rs6000_cost): Declare.
12952 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12954 * except.c (setjmp_fn): New global variable.
12955 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12956 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12957 if DONT_USE_BUILTIN_SETJMP is defined.
12959 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12961 * target.def (insn_cost): New hook.
12962 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12963 * doc/tm.texi: Regenerate.
12964 * rtlanal.c (insn_cost): Use the new hook.
12966 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12968 * combine.c (combine_validate_cost): Compute the new insn_cost,
12969 not just pattern_cost.
12970 (try_combine): Adjust comment.
12972 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12974 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12976 * combine.c (uid_insn_cost): Adjust comment.
12977 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12979 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12980 * dse.c (find_shift_sequence): Ditto.
12981 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12982 (bb_valid_for_noce_process_p): Use pattern_cost.
12983 * rtl.h (insn_rtx_cost): Delete.
12984 (pattern_cost): New prototype.
12985 (insn_cost): New prototype.
12986 * rtlanal.c (insn_rtx_cost): Rename to...
12987 (pattern_cost): ... this.
12990 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12992 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12993 (*jcc<mode>_0_r_i387): Ditto.
12994 (*jccxf_r_i387): Ditto.
12995 (*jcc<mode>_r_i387): Ditto.
12996 (*jccu<mode>_r_i387): Ditto.
12997 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12998 (*jcc): Rename from *jcc_1.
13000 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13002 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
13003 deferred rescans after the lvx/stvx recombination pre-pass.
13005 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13007 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
13008 memory operation instruction support.
13009 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13010 (rs6000-ibm-aix[789]*): Likewise.
13011 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13012 Document new functions.
13014 2017-10-09 Richard Biener <rguenther@suse.de>
13016 PR tree-optimization/82397
13017 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13018 equality only for semantically equal trees.
13020 2017-10-09 Richard Biener <rguenther@suse.de>
13022 PR tree-optimization/82449
13023 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13024 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13025 allow constant addresses.
13026 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13029 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13031 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13034 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13037 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13040 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13043 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13045 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13048 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13049 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13051 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13053 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13054 (WI_BINARY_PREDICATE_RESULT): Likewise.
13055 (wi::binary_traits::operator_result): New type.
13056 (wi::binary_traits::predicate_result): Likewise.
13057 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13058 (generic_wide_int::operator==, generic_wide_int::operator!=)
13059 (generic_wide_int::operator&, generic_wide_int::and_not)
13060 (generic_wide_int::operator|, generic_wide_int::or_not)
13061 (generic_wide_int::operator^, generic_wide_int::operator+
13062 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13064 (operator~, unary operator-, operator==, operator!=, operator&)
13065 (operator|, operator^, operator+, binary operator-, operator*): New
13067 * expr.c (get_inner_reference): Use wi::bit_and_not.
13068 * fold-const.c (fold_binary_loc): Likewise.
13069 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13070 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13071 (bit_value_binop): Likewise.
13072 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13073 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13074 (extract_range_from_binary_expr_1): Likewise.
13075 (masked_increment): Likewise.
13076 (simplify_bit_ops_using_ranges): Likewise.
13078 2017-10-09 Martin Jambor <mjambor@suse.cz>
13081 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13082 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13083 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13085 (hsa_fixup_mov_insn_type): New function.
13086 (hsa_op_with_type::get_in_type): Use it.
13087 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13088 immediates in an assert.
13089 (hsa_op_with_type::extend_int_to_32bit): New method.
13090 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13091 types. Convert to dest type if necessary.
13092 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13093 (reg_for_gimple_ssa): Pass false as min32int to
13094 hsa_type_for_scalar_tree_type.
13095 (gen_hsa_addr): Fixup type when creating addresable temporary.
13096 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13097 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13098 necessary. Call hsa_fixup_mov_insn_type.
13099 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13100 extend operands and convert to dest type if necessary.
13101 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13102 to dest type if necessary.
13103 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13104 if conversion nt necessary and size matches.
13105 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13106 to dest type if necessary.
13107 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13108 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13110 (gen_hsa_clrsb): Likewise.
13111 (gen_hsa_ffs): Likewise.
13112 (gen_hsa_divmod): Extend operands and convert to dest type if
13114 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13116 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13118 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13119 Remove empty default arguments. Use a brace block as output
13121 (conditional return): Ditto.
13123 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13124 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13126 (group_ending_nop): Ditto.
13127 (doloop_end): Ditto.
13128 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13129 (splitters for those): Ditto.
13131 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13133 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13134 a conditional jump (and the compare for it) so that pc_rtx is the
13136 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13137 for the deleted and renamed ctr<mode>_internal[234] patterns.
13138 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13139 Delete second conditional return pattern.
13140 (ctr<mode>_internal2): Delete this second bdnz pattern.
13141 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13142 (ctr<mode>_internal4): Delete this second bdz pattern.
13144 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13146 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13147 (always_initialized_rtx_for_ssa_name_p): New predicate.
13148 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13149 (finish_out_of_ssa): Free new field of SA.
13150 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13151 * tree-ssa-coalesce.c: Include tree-ssa.h.
13152 (get_parm_default_def_partitions): Remove extern keyword.
13153 (get_undefined_value_partitions): New function.
13154 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13155 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13156 uninitialized bits.
13157 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13159 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13161 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13163 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13165 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13166 for targets that preffer 128bit.
13168 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13170 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13172 2017-10-08 Olivier Hainque <hainque@adacore.com>
13174 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13175 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13177 (thumb_set_return_address): Likewise.
13179 2017-10-08 Olivier Hainque <hainque@adacore.com>
13181 * common/config/arm/arm-common.c (arm_except_unwind_info):
13182 Handle DWARF2_UNWIND_INFO.
13184 2017-10-07 Michael Collison <michael.collison@arm.com>
13186 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13189 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13191 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13192 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13193 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13194 defined, force the creation of a new block for a dispatch label.
13196 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13198 * invoke.texi (Wsuggest-attribute=cold): Document.
13199 * common.opt (Wsuggest-attribute=cold): New
13200 * ipa-pure-const.c (warn_function_cold): New function.
13201 * predict.c (compute_function_frequency): Use it.
13202 * predict.h (warn_function_cold): Declare.
13204 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13206 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13209 2017-10-06 Martin Liska <mliska@suse.cz>
13211 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13212 keyword for member functions.
13213 (struct sanopt_tree_couple): New struct.
13214 (struct sanopt_tree_couple_hash): New function.
13215 (struct sanopt_ctx): Add new hash_map.
13216 (has_dominating_ubsan_ptr_check): New function.
13217 (record_ubsan_ptr_check_stmt): Likewise.
13218 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13219 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13220 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13222 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13225 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13226 aarch64_simd_valid_immediate on CONST_VECTORs.
13227 (aarch64_reg_or_bic_imm): Likewise.
13229 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13231 PR rtl-optimization/82396
13232 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13234 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13236 * graphite-dependences.c (scop_get_reads): Move code to...
13237 (scop_get_must_writes): Move code to...
13238 (scop_get_may_writes): Move code to...
13239 (scop_get_reads_and_writes): ... here.
13240 (scop_get_dependences): Call scop_get_reads_and_writes.
13242 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13244 PR tree-optimization/82434
13245 * fold-const.h (can_native_encode_type_p,
13246 can_native_encode_string_p): Remove.
13247 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13248 don't encode anything, just return what would be otherwise returned.
13249 (native_encode_fixed, native_encode_complex, native_encode_vector):
13251 (native_encode_string): Likewise. Inline by hand
13252 can_native_encode_string_p.
13253 (can_native_encode_type_p): Remove.
13254 (can_native_encode_string_p): Remove.
13255 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13256 STRING_CSTs using can_native_encode_string_p, test all
13257 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13258 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13259 argument from native_encode_expr.
13260 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13261 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13264 2017-10-06 Richard Biener <rguenther@suse.de>
13266 PR tree-optimization/82397
13267 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13268 operand_equal_p but rely on data_ref_compare_tree for detecting
13270 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13271 to match up with dr_group_sort_cmp.
13273 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13276 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13278 * config/s390/s390-builtin-types.def: Regenerate.
13280 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13283 * config/s390/s390-builtin-types.def: Regenerate.
13284 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13285 Change flag from B_VXE to B_VX.
13286 (s390_vec_min_dbl): Remove B_VXE flag.
13288 2017-10-06 Richard Biener <rguenther@suse.de>
13290 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13291 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13292 translate_isl_ast_to_gimple::is_valid_rename,
13293 translate_isl_ast_to_gimple::get_rename,
13294 translate_isl_ast_to_gimple::get_def_bb_for_const,
13295 translate_isl_ast_to_gimple::get_new_name,
13296 translate_isl_ast_to_gimple::collect_all_ssa_names,
13297 translate_isl_ast_to_gimple::copy_loop_phi_args,
13298 translate_isl_ast_to_gimple::collect_all_ssa_names,
13299 translate_isl_ast_to_gimple::copy_loop_phi_args,
13300 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13301 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13302 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13303 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13304 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13305 translate_isl_ast_to_gimple::copy_cond_phi_args,
13306 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13307 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13308 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13309 translate_isl_ast_to_gimple::rename_uses,
13310 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13311 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13312 (set_rename_for_each_def): Likewise.
13313 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13314 here. Handle rewriting SCEV analyzable uses here.
13315 (copy_bb_and_scalar_dependences): Generate code for PHI
13317 (graphite_regenerate_ast_isl): Adjust.
13318 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13319 (add_write, add_read): New functions.
13320 (build_cross_bb_scalars_def): Use it and simplify.
13321 (build_cross_bb_scalars_use): Likewise.
13322 (graphite_find_cross_bb_scalar_vars): Inline into...
13323 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13324 simulating out-of-SSA. Compute liveout and add dependencies.
13325 (build_scops): Force an empty entry block.
13326 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13328 (sese_build_liveouts): Declare.
13329 (sese_trivially_empty_bb_p): Likewise.
13330 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13331 compute liveout and debug_liveout.
13332 (sese_bad_liveouts_use): Remove.
13333 (sese_reset_debug_liveouts_bb): Likewise.
13334 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13335 (sese_build_liveouts): Build liveout and debug_liveout and store
13337 (new_sese_info): Adjust.
13338 (free_sese_info): Likewise.
13339 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13340 do not build liveout here.
13341 (move_sese_in_condition): Adjust region entry.
13342 (scev_analyzable_p): Match up with chrec_apply requirements.
13343 (sese_trivially_empty_bb_p): New.
13344 * tree-into-ssa.c (get_reaching_def): Properly support generating
13345 default-defs for incremental rewrite of anonymous names.
13347 2017-10-06 Richard Biener <rguenther@suse.de>
13349 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13350 precision do not perform modulo reduction.
13352 2017-10-06 Richard Biener <rguenther@suse.de>
13354 PR tree-optimization/82436
13355 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13356 conservatively choose the vectorization factor when checking
13357 whether we can perform the required load permutation.
13358 (vect_transform_slp_perm_load): Assert when we may not fail.
13360 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13362 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13363 message for incompatible -msdata=* and -mcall-* options.
13365 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13367 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13368 rate for post-reload scheduling.
13370 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13372 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13374 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13376 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13377 to improve monte carlo in scimark.
13379 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13381 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13382 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13383 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13384 pentium4_cost, nocona_cost): Set reassociation width to 1.
13385 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13386 width to 2 for fp operations and 1 otherwise.
13387 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13389 (atom_cost): Set reassociation width to 2.
13390 (slm_cost, generic_cost): Set fp reassociation width
13391 to 2 and 1 otherwise.
13392 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13393 (core_cost): Set fp reassociation width to 4 and vector to 2.
13394 (ix86_reassociation_width): Rewrite using cost table; special case
13395 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13396 and TARGET_AVX128_OPTIMAL.
13397 * config/i386/i386.h (processor_costs): Add
13398 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13399 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13400 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13401 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13402 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13403 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13405 2017-10-05 Nathan Sidwell <nathan@acm.org>
13407 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13409 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13411 * config/arm/arm.c (arm_test_fpu_data): New.
13412 (arm_run_selftests): Call arm_test_fpu_data.
13414 2017-10-04 Nathan Sidwell <nathan@acm.org>
13416 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13418 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13420 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13422 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13423 check type for aarch64_simd_valid_immediate.
13424 (aarch64_output_simd_mov_immediate): Update prototype.
13425 (aarch64_simd_valid_immediate): Update prototype.
13426 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13427 support for ORR-immediate.
13428 (and<mode>3): modified pattern to add support for BIC-immediate.
13429 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13430 now checks for valid immediate for BIC and ORR based on new enum
13432 (aarch64_output_simd_mov_immediate): Function now used to output
13433 BIC/ORR imm as well based on new enum argument.
13434 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13436 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13437 (aarch64_reg_or_bic_imm): Likewise.
13439 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13441 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13442 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13444 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13448 PR rtl-optimization/82396
13449 * haifa-sched.c (autopref_multipass_init): Simplify
13451 (autopref_rank_data): Simplify sort order.
13452 * sched-int.h (autopref_multipass_data_): Remove
13453 multi_mem_insn_p, min_offset and max_offset.
13455 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13457 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13459 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13461 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13462 vect_doubleint_cvt.
13464 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13466 * doc/sourcebuild.texi: Document vect_long_mult.
13468 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13470 PR tree-optimization/82413
13471 * fold-const.c (build_range_check): Use widest_int when comparing
13472 the maximum ETYPE value with HIGH.
13474 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13476 PR rtl-optimization/82396
13477 * haifa-sched.c (autopref_multipass_init): Simplify
13479 (autopref_rank_data): Simplify sort order.
13480 * sched-int.h (autopref_multipass_data_): Remove
13481 multi_mem_insn_p, min_offset and max_offset.
13483 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13485 PR tree-optimization/82381
13486 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13487 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13490 PR tree-optimization/82374
13491 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13492 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13493 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13494 current_function_decl to the new decl.
13496 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13498 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13499 helper macro for IEEE float128 hardware built-in functions.
13500 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13502 (TRUNCF128_ODD): Likewise.
13503 (ADDF128_ODD): Likewise.
13504 (SUBF128_ODD): Likewise.
13505 (MULF128_ODD): Likewise.
13506 (DIVF128_ODD): Likewise.
13507 (FMAF128_ODD): Likewise.
13508 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13509 UNSPEC_TRUNC_ROUND_TO_ODD.
13510 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13511 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13512 floating point round to odd instructions.
13513 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13514 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13515 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13516 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13517 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13518 (trunc<mode>sf2_hw): Change the truncate with round to odd
13519 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13520 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13521 to odd hardware instructions.
13522 (sub<mode>3_odd): Likewise.
13523 (mul<mode>3_odd): Likewise.
13524 (div<mode>3_odd): Likewise.
13525 (sqrt<mode>2_odd): Likewise.
13526 (fma<mode>4_odd): Likewise.
13527 (fms<mode>4_odd): Likewise.
13528 (nfma<mode>4_odd): Likewise.
13529 (nfms<mode>4_odd): Likewise.
13530 (trunc<mode>df2_odd): Change the truncate with round to odd
13531 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13533 * doc/extend.texi (PowerPC built-in functions): Update documentation
13534 for existing IEEE float128-bit built-in functions. Add built-in
13535 functions that generate the IEEE 128-bit floating point round to
13538 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13540 PR rtl-optimization/77729
13541 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13542 to (X&(C1&~C2))|C2 transformations.
13544 2017-10-03 Martin Jambor <mjambor@suse.cz>
13546 PR tree-optimization/82363
13547 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13548 mismatch, mark lacc written regardless of racc.
13550 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13552 PR tree-optimization/82381
13553 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13554 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13555 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13556 fallthrough into reassoc_stmt_dominates_stmt_p.
13559 * combine.c (combine_instructions): Don't combine in unreachable
13562 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13565 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13566 function to not use the have_cpu variable. Do not set cpu_index,
13567 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13568 or the default cpu.
13569 (rs6000_valid_attribute_p): Remove duplicate initializations of
13570 old_optimize and func_optimize.
13571 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13572 (rs6000_activate_target_options): Make global.
13573 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13576 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13578 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13579 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13580 Set *poffset to 0 and *psize and *pmax_size to -1 if
13581 *poffset + *psize overflows in HOST_WIDE_INT.
13583 PR tree-optimization/82387
13584 PR tree-optimization/82388
13585 PR tree-optimization/82389
13586 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13587 instead of live_bytes non-NULL.
13589 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13592 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13595 2017-10-02 Richard Biener <rguenther@suse.de>
13597 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13598 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13599 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13601 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13603 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13604 requested precision matches the type's.
13605 * calls.c (alloc_max_size): Calculate the new candidate size as
13606 a widest_int and use wi::to_widest when comparing it with the
13607 current candidate size.
13608 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13609 zero rather than integer_zero_node.
13610 * match.pd: Check for a no-op conversion before using wi::add
13611 rather than after. Use tree_to_uhwi when summing small shift
13612 counts into an unsigned int.
13614 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13615 Alan Hayward <alan.hayward@arm.com>
13616 David Sherwood <david.sherwood@arm.com>
13619 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13620 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13621 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13622 POINTER_AND_FP_REGS.
13624 2017-10-02 Richard Biener <rguenther@suse.de>
13626 PR tree-optimization/82355
13627 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13628 a mapping for the enclosing loop but avoid generating one for
13629 the loop tree root.
13630 (copy_bb_and_scalar_dependences): Remove premature codegen
13631 error on PHIs in blocks duplicated into multiple places.
13632 * graphite-scop-detection.c
13633 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13634 in the region use it as loop and nest to analyze the DR in.
13635 (try_generate_gimple_bb): Likewise.
13636 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13637 (add_loop_constraints): For blocks in a loop not in the region
13638 create a dimension with a single iteration.
13639 * sese.h (gbb_loop_at_index): Remove assert.
13641 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13643 * omp-expand.c (adjust_context_scope): New function.
13644 (expand_parallel_call): Call adjust_context_scope.
13646 2017-10-01 Jeff Law <law@redhat.com>
13648 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13649 dom_opt_dom_walker class with direct access to private members.
13650 Add comments. Call test_for_singularity.
13651 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13652 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13653 m_dummy_cond anymore.
13654 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13656 (pass_dominator:execute): Build the dummy_cond here and pass it
13657 to the dom_opt_dom_walker ctor.
13658 (test_for_singularity): New function.
13660 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13661 Maya Rashish <coypu@sdf.org>
13663 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13664 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13665 (alpha*-*-netbsd*) Use nbsd_tm_file.
13666 (arm*-*-netbsdelf*) Likewise.
13667 (i[34567]86-*-netbsdelf*) Likewise.
13668 (x86_64-*-netbsd*) Likewise.
13669 (mips*-*-netbsd*) Likewise.
13670 (powerpc-*-netbsd*) Likewise.
13671 (sh*-*-netbsd*) Likewise.
13672 (sparc-*-netbsdelf*) Likewise.
13673 (sparc64-*-netbsd*) Likewise.
13674 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13676 (vax-*-netbsdelf*) Likewise.
13677 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13678 (UINT_FAST8_TYPE) Likewise.
13679 (INT_FAST16_TYPE) Check CHAR_FAST16.
13680 (UINT_FAST16_TYPE) Likewise.
13682 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13685 * config/i386/i386.md
13686 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13687 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13688 *divmodsi4_zext_2): New define_insn_and_split.
13689 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13690 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13691 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13692 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13693 New define_insn_and_split.
13694 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13695 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13696 operands[1] having DImode when mode is SImode.
13698 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13699 always SImode for DIV and MOD in REG_EQUAL notes.
13701 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13703 PR middle-end/82319
13704 * match.pd: Fix handling of NaNs in pattern.
13706 2017-09-29 Jeff Law <law@redhat.com>
13708 * sbitmap.c (bitmap_bit_in_range_p): New function.
13709 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13710 * tree-ssa-dse.c (live_bytes_read): New function.
13711 (dse_classify_store): Ignore reads of dead bytes.
13713 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13714 typos and whitespace errors.
13715 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13716 * config/s390/s390.c (s390_emit_prologue): Likewise.
13718 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13721 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13722 with a symbol for LRA.
13724 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13726 PR rtl-optimization/82338
13727 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13729 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13731 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13732 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13733 (qsort_chk): Declare.
13734 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13735 (qsort_chk): New function.
13737 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13739 PR tree-optimization/82337
13740 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13741 phi definition if the PHI result appears in an abnormal PHI.
13742 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13743 basis appears in an abnormal PHI.
13745 2017-09-29 Richard Biener <rguenther@suse.de>
13747 * graphite-isl-ast-to-gimple.c
13748 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13749 (binary_op_to_tree): Use it.
13750 (get_rename_from_scev): Likewise.
13751 (copy_loop_phi_nodes): Likewise.
13752 (copy_bb_and_scalar_dependences): Likewise.
13753 (translate_pending_phi_nodes): Likewise.
13755 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13758 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13759 for movabsq $(i32 << shift), r64.
13761 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13763 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13764 index when encoding %esp as %rsp to avoid 0x67 prefix.
13766 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13768 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13769 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13771 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13773 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13774 Extensions with more than 16 double VFP registers.
13775 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13776 to_clear_mask and all code related to it. Replace the remaining
13777 entry by a sbitmap and adapt code accordingly.
13779 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13781 * brig-builtins.def: Change pure attributes to const.
13783 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13785 * config.gcc (default_gnu_indirect_function): Default to yes for
13786 sparc*-*-linux* with glibc.
13788 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13790 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13791 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13792 when creating .init_array and .fini_array sections with priority
13795 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13798 * config/aarch64/aarch64.c
13799 (aarch64_builtin_support_vector_misalignment): Always return false
13800 when misalignment is unknown.
13802 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13804 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
13805 this function to return false if the definition used by the swap
13806 instruction is artificial, or if the memory address from which the
13807 constant value is loaded is not represented by a base address held
13808 in a register or if the base address register is a frame or stack
13809 pointer. Additionally, return false if the base address of the
13810 loaded constant is a SYMBOL_REF but is not considered to be a
13812 (replace_swapped_load_constant): New function.
13813 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
13814 loaded constant vector with a load of a swapped constant vector.
13816 2017-09-27 Carl Love <cel@us.ibm.com>
13818 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
13819 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
13820 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
13823 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
13825 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
13826 first, always call autopref_rank_data otherwise.
13828 2017-09-27 Richard Biener <rguenther@suse.de>
13830 * graphite-scop-detection.c (find_scop_parameters): Move
13831 loop bound handling ...
13832 (gather_bbs::before_dom_children): ... here, avoiding the need
13833 to build scop_info->loop_nest.
13834 (record_loop_in_sese): Remove.
13835 * sese.h (sese_info_t::loop_nest): Remove.
13836 * sese.c (new_sese_info): Do not allocate loop_nest.
13837 (free_sese_info): Do not free loop_nest.
13839 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13842 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13843 lhs from calls if the lhs has addressable type.
13845 2017-09-27 Richard Biener <rguenther@suse.de>
13847 * graphite.h (scop::max_alias_set): New member.
13848 * graphite-scop-detection.c: Remove references to non-existing
13849 --param in comments.
13850 (build_alias_sets): Record the maximum alias set used for drs.
13851 (build_scops): Support zero as unlimited for
13852 --param graphite-max-arrays-per-scop.
13853 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
13854 and inline into ...
13855 (build_poly_sr_1): ... here. Compute alias set based on the
13856 maximum alias set used for drs rather than
13857 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
13859 2017-09-27 Richard Biener <rguenther@suse.de>
13861 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13862 --param loop-block-tile-size=0 to disable tiling.
13864 2017-09-27 Richard Biener <rguenther@suse.de>
13866 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
13867 (graphite-max-nb-scop-params): Document special value zero.
13868 * domwalk.h (dom_walker::STOP): New symbolical constant.
13869 (dom_walker::dom_walker): Add optional parameter for bb to
13871 (dom_walker::~dom_walker): Declare.
13872 (dom_walker::before_dom_children): Document STOP return value.
13873 (dom_walker::m_user_bb_to_rpo): New member.
13874 (dom_walker::m_bb_to_rpo): Likewise.
13875 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
13876 mapping here if not provided by the user.
13877 (dom_walker::~dom_walker): Free bb to RPO mapping if not
13878 provided by the user.
13879 (dom_walker::STOP): Define.
13880 (dom_walker::walk): Do not compute bb to RPO mapping here.
13881 Support STOP return value from before_dom_children to stop
13883 * graphite-optimize-isl.c (optimize_isl): If the schedule
13884 is the same still generate code if -fgraphite-identity
13885 or -floop-parallelize-all are given.
13886 * graphite-scop-detection.c: Include cfganal.h.
13887 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13889 (gather_bbs::before_dom_children): Return STOP for BBs
13891 (build_scops): Compute bb to RPO mapping and pass it to
13892 the domwalk. Treat --param graphite-max-nb-scop-params=0
13893 as not limiting the number of params.
13894 * graphite.c (graphite_initialize): Remove limit on the
13895 number of basic-blocks in a function.
13896 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13897 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13898 default value of 10.
13900 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13902 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13903 Adjust code to eliminate needing to do the shift right 32-bits
13904 operation after XSCVDPSPN.
13906 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13908 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13909 ((X / Y) != 0 -> X >= Y): Likewise.
13911 2017-09-26 Carl Love <cel@us.ibm.com>
13913 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13914 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13915 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13916 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13917 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13918 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13919 definitions and overloading.
13920 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13921 statement for P9V_BUILTIN_XST_LEN_R.
13922 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13923 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13924 define_expand and define_insn for the instructions and builtins.
13925 * doc/extend.texi: Update the built-in documentation file for the new
13926 built-in functions.
13927 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13928 define_insn for the instructions
13930 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13933 * gcc/config/netbsd-protos.h: New file.
13934 * gcc/config/netbsd.c: New file.
13935 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13936 * gcc/config/t-netbsd: New file.
13937 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13938 (tmake_file) Add t-netbsd.
13939 (extra_objs) Add netbsd.o.
13941 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13945 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13947 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13949 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13950 sign extension from a vector register to a GPR by doing a 32-bit
13951 direct move and then an EXTSW.
13952 (extendsi<mode>2 splitter): Likewise.
13953 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13954 right or vector extract after doing XSCVDPSPN. Use
13955 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13957 (movdi_from_sf_zero_ext): Likewise.
13958 (reload_gpr_from_vsxsf): Likewise.
13959 (p8_mfvsrd_4_disf): Delete, no longer used.
13960 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13961 then needing to move the SFmode to a GPR to use the XSCVDPSP
13962 instruction instead of FRSP and XSCVDPSPN.
13963 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13964 it is adjacent to the other XSCVSPDP insns.
13965 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13966 SFmode to be in traditional Altivec registers.
13967 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13968 (vsx_xscvspdpn): Likewise.
13969 (vsx_xscvspdpn_scalar): Likewise.
13971 2017-09-26 Martin Jambor <mjambor@suse.cz>
13973 * tree-sra.c (compare_access_positions): Put integral types first,
13974 stabilize sorting of integral types, remove conditions putting
13975 non-full-precision integers last.
13976 (sort_and_splice_var_accesses): Disable scalarization if a
13977 non-integert would be represented by a non-full-precision integer.
13979 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13981 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13982 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13983 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13984 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13985 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13986 conditionals inside the function instead of around it. Call
13987 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13988 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13990 2017-09-26 Richard Biener <rguenther@suse.de>
13992 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13994 (scop_detection::build_scop_breadth): ... this. Removed.
13995 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13996 (scop_detection::harmful_stmt_in_bb): Likewise.
13997 (scop_detection::graphite_can_represent_stmt): Likewise.
13998 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13999 (scop_detection::can_represent_loop): Remove recursion, fold in ...
14000 (scop_detection::can_represent_loop_1): ... this. Removed.
14001 (scop_detection::harmful_loop_in_region): Simplify after inlining
14002 the above and remove more quadraticness.
14003 (build_scops): Adjust.
14004 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
14007 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14010 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14011 REGNO (base) == SP_REG if base is a REG.
14013 PR middle-end/35691
14014 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14015 if it is different SSA_NAME.
14016 (optimize_range_tests_cmp_bitwise): New function.
14017 (optimize_range_tests): Call it.
14019 2017-09-26 Richard Biener <rguenther@suse.de>
14021 PR tree-optimization/82321
14022 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14023 for the def being inside the loop.
14025 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14027 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14029 * config/s390/s390-builtins.def: Fix constraint on op4.
14031 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14033 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14034 independent expanders.
14035 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14036 ("vec_ordered", "vec_unordered"): New expanders.
14038 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14040 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14043 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14045 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14046 vec_unpacks_lo_v16qi.
14047 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14049 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14051 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14052 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14053 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14055 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14057 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14059 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14061 ("*vec_slb<mode>"): New insn pattern.
14062 ("vec_shr_<mode>"): New expander.
14063 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14064 and force the shift count operand to V16QImode.
14065 ("vec_srb<mode>"): Set shift count mode to V16QI.
14067 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14069 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14070 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14071 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14073 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14076 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14078 2017-09-26 Richard Biener <rguenther@suse.de>
14080 PR tree-optimization/82320
14081 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14084 2017-09-25 Jeff Law <law@redhat.com>
14086 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14087 prototype for new argument.
14088 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14089 mostly extracted from rs6000_emit_allocate_stack.
14090 (rs6000_emit_probe_stack_range_stack_clash): New function.
14091 (rs6000_emit_allocate_stack): Call
14092 rs6000_emit_probe_stack_range_stack_clash as needed.
14093 (rs6000_emit_probe_stack_range): Add additional argument
14094 to call to gen_probe_stack_range{si,di}.
14095 (output_probe_stack_range): New.
14096 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14097 (output_probe_stack_range_stack_clash): New.
14098 (rs6000_emit_prologue): Emit notes into dump file as requested.
14099 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14100 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14101 Add additional operand and pass it to output_probe_stack_range.
14103 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14105 PR tree-optimization/82163
14106 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14107 (checking_verify_loop_closed_ssa): New parameter.
14108 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14109 (check_loop_closed_ssa_stmt): Delete.
14110 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14111 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14112 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14115 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14117 * brig-builtins.def: Treat HSAIL barrier builtins as
14118 setjmp/longjump style functions.
14120 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14122 * target.def (constant_alignment): New hook.
14123 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14124 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14125 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14126 * doc/tm.texi: Regenerate.
14127 * targhooks.h (default_constant_alignment): Declare.
14128 (constant_alignment_word_strings): Likewise.
14129 * targhooks.c (default_constant_alignment): New function.
14130 (constant_alignment_word_strings): Likewise.
14131 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14132 instead of CONSTANT_ALIGNMENT.
14133 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14134 (force_const_mem): Likewise.
14135 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14136 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14137 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14138 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14139 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14141 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14142 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14143 constant_alignment_word_strings.
14144 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14145 (CONSTANT_ALIGNMENT): Likewise.
14146 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14147 (arm_constant_alignment): New function.
14148 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14149 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14150 constant_alignment_word_strings.
14151 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14152 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14153 constant_alignment_word_strings.
14154 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14155 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14156 (cris_constant_alignment): New function.
14157 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14158 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14159 (epiphany_constant_alignment): New function.
14160 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14161 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14162 constant_alignment_word_strings.
14163 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14164 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14165 constant_alignment_word_strings.
14166 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14167 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14168 constant_alignment_word_strings.
14169 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14170 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14171 * config/i386/i386.c (ix86_constant_alignment): Make static.
14172 Use the same interface as the target hook.
14173 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14174 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14175 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14176 constant_alignment_word_strings.
14177 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14178 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14179 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14180 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14181 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14182 constant_alignment_word_strings.
14183 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14184 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14185 constant_alignment_word_strings.
14186 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14187 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14188 constant_alignment_word_strings.
14189 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14190 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14192 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14193 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14194 * config/mips/mips.c (mips_constant_alignment): New function.
14195 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14196 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14197 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14198 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14199 (mmix_constant_alignment): Make static. Use the same interface
14200 as the target hook.
14201 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14202 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14203 constant_alignment_word_strings.
14204 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14205 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14206 constant_alignment_word_strings.
14207 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14208 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14209 constant_alignment_word_strings.
14210 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14211 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14212 (rs6000_constant_alignment): New function.
14213 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14214 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14215 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14216 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14217 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14218 (rs6000_constant_alignment): New function.
14219 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14220 * config/s390/s390.c (s390_constant_alignment): New function.
14221 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14222 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14223 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14224 constant_alignment_word_strings.
14225 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14226 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14227 (sparc_constant_alignment): New function.
14228 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14229 * config/spu/spu.c (spu_constant_alignment): New function.
14230 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14231 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14232 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14233 constant_alignment_word_strings.
14234 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14235 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14236 constant_alignment_word_strings.
14237 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14238 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14239 constant_alignment_word_strings.
14240 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14241 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14242 (visium_constant_alignment): New function.
14243 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14244 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14245 (xtensa_constant_alignment): New function.
14246 * system.h (CONSTANT_ALIGNMENT): Poison.
14248 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14250 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14251 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14252 (rs6000_builtin_valid_without_lhs): New helper function.
14253 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14254 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14256 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14258 * target.h (vec_perm_indices): Use unsigned short rather than
14260 (auto_vec_perm_indices): Likewise.
14261 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14262 Use unsigned int rather than unsigned char.
14263 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14265 2017-09-25 Richard Biener <rguenther@suse.de>
14267 * cfgloop.h (sort_sibling_loops): Declare.
14268 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14269 (sort_sibling_loops): New function sorting the sibling loop list
14271 * graphite.c (graphite_transform_loops): Sort sibling loops.
14273 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14275 * target.def (vec_perm_const_ok): Change sel parameter to
14277 * optabs-query.c (can_vec_perm_p): Update accordingly.
14278 * doc/tm.texi: Regenerate.
14279 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14280 auto_vec_perm_indices and remove separate nelt field.
14281 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14282 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14283 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14284 (aarch64_expand_vec_perm_const): Update accordingly.
14285 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14286 to vec_perm_indices.
14287 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14288 auto_vec_perm_indices and remove separate nelt field.
14289 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14290 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14291 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14293 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14294 to vec_perm_indices.
14295 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14296 sel to vec_perm_indices.
14297 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14298 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14299 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14301 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14304 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14307 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14308 on the FUNCTION_DECL function context if it has a DIE that is a
14311 2017-09-25 Richard Biener <rguenther@suse.de>
14313 PR tree-optimization/82285
14314 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14317 2017-09-25 Tom de Vries <tom@codesourcery.com>
14321 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14324 2017-09-25 Richard Biener <rguenther@suse.de>
14326 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14327 ISL errors other than isl_error_quota happen. Dump if the
14328 schedule is the same.
14329 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14330 errors instead of aborting inside ISL.
14332 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14335 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14336 of libgcc_eh for m64.
14337 * config/i386/darwin64.h: Likewise.
14339 2017-09-25 Richard Biener <rguenther@suse.de>
14341 PR middle-end/82144
14342 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14343 attribute for incomplete types nor twice for complete ones.
14345 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14348 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14349 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14352 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14355 * config/i386/i386.opt (mprefer-avx256): Use
14356 ix86_target_flags variable.
14357 * config/i386/i386.c (ix86_target_string): Move
14358 -mprefer-avx256 to flag2_opts.
14360 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14362 PR middle-end/35691
14363 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14364 and x != -1 | y != -1 into (x & y) != -1.
14366 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14368 * config.gcc: Add new case statement to set
14369 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14370 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14371 s390x-*-linux* case statements. Added aarch64 to the list of
14372 supported architectures.
14374 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14376 PR tree-optimization/82289
14377 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14378 STMT_VINFO_RELEVANT_P.
14380 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14381 Alan Hayward <alan.hayward@arm.com>
14382 David Sherwood <david.sherwood@arm.com>
14384 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14385 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14386 (extract_range_from_binary_expr_1): Don't call
14387 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14389 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14390 Alan Hayward <alan.hayward@arm.com>
14391 David Sherwood <david.sherwood@arm.com>
14393 * target.def (preferred_vector_alignment): New hook.
14394 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14396 * doc/tm.texi: Regenerate.
14397 * targhooks.h (default_preferred_vector_alignment): Declare.
14398 * targhooks.c (default_preferred_vector_alignment): New function.
14399 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14401 (DR_TARGET_ALIGNMENT): New macro.
14402 (aligned_access_p): Update commentary.
14403 (vect_known_alignment_in_bytes): New function.
14404 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14406 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14407 Calculate the misalignment based on the target alignment rather than
14409 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14410 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14411 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14412 the target alignment, rather than masking the element misalignment
14413 with the number of elements in a vector. Also use the target
14414 alignment when calculating the maximum number of peels.
14415 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14416 instead of TYPE_ALIGN_UNIT.
14417 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14418 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14419 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14420 (vect_create_data_ref_ptr): Likewise.
14421 (vect_setup_realignment): Realign by ANDing with
14422 -DR_TARGET_MISALIGNMENT.
14423 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14424 the number of peels based on DR_TARGET_ALIGNMENT.
14425 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14426 with the guaranteed alignment boundary when deciding whether
14428 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14429 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14430 (ensure_base_align): Remove stmt_info parameter. Get the
14431 target base alignment from DR_TARGET_ALIGNMENT.
14432 (vectorizable_store): Update call accordingly. Interpret
14433 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14435 (vectorizable_load): Likewise.
14437 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14438 Alan Hayward <alan.hayward@arm.com>
14439 David Sherwood <david.sherwood@arm.com>
14441 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14442 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14443 (vect_enhance_data_refs_alignment): Likewise.
14445 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14447 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14448 error. Only quit immediately if parsing is complete.
14449 (BEGIN): Initialize fatal_err and parse_done.
14450 (begin fpu, end fpu): Check number of arguments.
14451 (begin arch, end arch): Likewise.
14452 (begin cpu, end cpu): Likewise.
14453 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14454 (optalias): Likewise.
14456 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14458 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14459 * config/arm/arm-isa.h: Delete. Move definitions to ...
14460 * arm-cpus.in: ... here. Use new feature and fgroup values.
14461 * config/arm/arm.c (arm_option_override): Use lower case for feature
14463 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14464 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14465 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14467 (print_isa_bits_for): New function.
14468 (gen_isa): New function.
14469 (gen_comm_data): Use print_isa_bits_for.
14470 (define feature): New keyword.
14471 (define fgroup): New keyword.
14472 * config/arm/t-arm (TM_H): Remove.
14473 (GTM_H): Add arm-isa.h.
14474 (arm-isa.h): Add rule to generate file.
14475 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14476 case for feature bit names.
14478 2017-09-22 Richard Biener <rguenther@suse.de>
14480 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14482 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14483 print of no dependency loops ...
14484 * graphite.c (graphite_transform_loops): ... here.
14485 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14487 (same_close_phi_node, remove_duplicate_close_phi,
14488 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14489 (canonicalize_loop_closed_ssa): ... here and simplify.
14490 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14491 (optimize_isl): Use dump_printf_loc to tell when we stopped
14492 optimizing because of an ISL timeout.
14494 2017-09-22 Richard Biener <rguenther@suse.de>
14496 PR tree-optimization/82291
14497 * tree-if-conv.c (predicate_mem_writes): Make sure to
14498 remove writes in blocks predicated with false.
14500 2017-09-22 Richard Biener <rguenther@suse.de>
14502 * sese.c: Include cfganal.h.
14503 (if_region_set_false_region): Remove.
14504 (create_if_region_on_edge): Likewise.
14505 (move_sese_in_condition): Re-implement without destroying
14508 2017-09-22 Richard Biener <rguenther@suse.de>
14510 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14511 Verify both BBs contain loop PHI nodes before dispatching to
14512 copy_loop_phi_args.
14513 (graphite_regenerate_ast_isl): Do not recompute dominators,
14514 do not verify three times. Restructure for clarity.
14515 * graphite-scop-detection.c (same_close_phi_node,
14516 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14517 defined_in_loop_p, canonicalize_loop_closed_ssa,
14518 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14519 checking and SSA rewrite, move to ...
14520 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14521 (graphite_initialize): Do not pass in ctx, do not reset the
14522 SCEV cache, compute only dominators.
14523 (graphite_transform_loops): Allocate ISL ctx after
14524 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14525 Maintain post-dominators only around build_scops.
14526 * sese.c (if_region_set_false_region): Make static. Free
14527 and recompute dominators.
14528 (move_sese_in_condition): Assert we don't get called with
14529 post-dominators computed.
14530 * sese.h (if_region_set_false_region): Remove.
14532 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14534 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14535 mode attribute for TARGET_AVX512VL.
14537 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14539 * config/i386/i386.opt (mprefer-avx256): New option.
14540 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14542 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14543 for TARGET_PREFER_AVX256.
14544 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14546 2017-09-21 Jeff Law <law@redhat.com>
14548 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14549 Fix dump output if the only stack space is for pushed registers.
14551 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14553 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14556 2017-09-21 Martin Sebor <msebor@redhat.com>
14559 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14560 code (in C++) or code that triggers warnings.
14562 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14564 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14566 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14568 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14569 * sched-rgn.c: Ditto.
14570 * sel-sched-ir.c: Ditto.
14572 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14574 * toplev.h (set_random_seed): Adjust return type.
14575 * toplev.c (init_local_tick): Move eager initialization of random_seed
14576 to get_random_seed. Adjust comment.
14577 (init_random_seed): Inline to get_random_seed, delete.
14578 (get_random_seed): Initialize random_seed lazily.
14579 (set_random_seed): Do not return previous value.
14580 (print_switch_value): Do not call get_random_seed.
14582 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14584 * cgraph.c (delete_function_version): New, broken out from...
14585 (cgraph_node::delete_function_version): ...here. Rename to
14586 cgraph_node::delete_function_version_by_decl. Update all uses.
14587 (cgraph_node::remove): Call delete_function_version.
14589 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14592 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14593 VAR_DECLs to which addressable non-volatile parameters are mapped
14594 and for id->retvar after the return value assignment. Clear
14595 id->retval and id->retbnd after inlining.
14597 2017-09-21 Richard Biener <rguenther@suse.de>
14599 PR tree-optimization/82276
14600 PR tree-optimization/82244
14601 * tree-vrp.c (build_assert_expr_for): Set
14602 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14604 (remove_range_assertions): Revert earlier change.
14606 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14609 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14611 2017-09-21 Richard Biener <rguenther@suse.de>
14613 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14614 Restore valid IL after code generation errors.
14615 * graphite.c (graphite_transform_loops): Diagnose code
14616 generation issues as MSG_MISSED_OPTIMIZATION and continue
14617 with processing SCOPs.
14619 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14620 Alan Hayward <alan.hayward@arm.com>
14621 David Sherwood <david.sherwood@arm.com>
14623 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14624 rather than choosing between plus_constant and gen_rtx_<CODE>.
14625 * expr.c (emit_push_insn): Likewise.
14626 (expand_expr_real_2): Likewise.
14628 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14629 Alan Hayward <alan.hayward@arm.com>
14630 David Sherwood <david.sherwood@arm.com>
14632 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14634 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14635 Alan Hayward <alan.hayward@arm.com>
14636 David Sherwood <david.sherwood@arm.com>
14638 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14639 calling tree_to_uhwi.
14641 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14642 Alan Hayward <alan.hayward@arm.com>
14643 David Sherwood <david.sherwood@arm.com>
14645 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14646 INTEGER_CST rather than a negative test for ADDR_EXPR.
14648 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14649 Alan Hayward <alan.hayward@arm.com>
14650 David Sherwood <david.sherwood@arm.com>
14652 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14653 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14655 2017-09-21 Richard Biener <rguenther@suse.de>
14657 PR tree-optimization/71351
14658 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14659 graphite_create_new_loop_guard): Remove, fold remaining parts
14661 (translate_isl_ast_node_for): ... here and simplify.
14663 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14666 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14667 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14668 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14669 alternative always use QI mode, for -Os imov (=R,R) alternative
14670 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14673 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14674 Jeff Law <law@redhat.com>
14676 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14677 (allocate_stack_space): New function, partially extracted from
14678 s390_emit_prologue.
14679 (s390_emit_prologue): Track offset to most recent stack probe.
14680 Code to allocate space moved into allocate_stack_space.
14681 Dump actions when no stack is allocated.
14682 (s390_prologue_plus_offset): New function.
14683 (s390_emit_stack_probe): Likewise.
14685 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14687 * common.opt (Wa, Wl, Wp, g, gz=): Add
14689 (gno-column-info): Remove.
14690 (gcolumn-info): Drop RejectNegative.
14691 (gno-): New prefix.
14692 (gno-record-gcc-switches): Remove.
14693 (grecord-gcc-switches): Drop RejectNegative.
14694 (gno-split-dwarf): Remove.
14695 (gsplit-dwarf): Drop RejectNegative.
14696 (gno-strict-dwarf): Remove.
14697 (gstrict-dwarf): Drop RejectNegative.
14698 * config/darwin.opt (gfull, gused): Add RejectNegative.
14699 * dwarf2out.c (gen_producer_string): Drop
14700 gno-record-gcc-switches handler.
14701 * optc-gen.awk: Add g to prefixes with negative forms.
14702 * opts-common.c (remapping_prefix_p): New.
14703 (find_opt): Check it.
14704 (generate_canonical_option): Test g prefix.
14705 (option_map): Add -gno- mapping.
14706 (add_misspelling_candidates): Check remapping_prefix_p.
14708 2017-09-20 Jeff Law <law@redhat.com>
14710 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14711 thinko in stack clash protection support.
14713 * explow.c (compute_stack_clash_protection_loop_data): Use
14714 CONST_INT_P instead of explicit test. Verify object is a
14715 CONST_INT_P before looking at INTVAL.
14716 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14717 instead of explicit test.
14719 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14722 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14723 address instead of to r1 and r11.
14725 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14727 * config.gcc: Support "knm".
14728 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14729 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14731 * config/i386/i386.c (m_KNM): Define.
14732 (processor_target_table): Add "knm".
14734 (ix86_option_override_internal): Add "knm".
14735 (ix86_issue_rate): Add PROCESSOR_KNM.
14736 (ix86_adjust_cost): Ditto.
14737 (ia32_multipass_dfa_lookahead): Ditto.
14738 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14739 (fold_builtin_cpu): Add M_INTEL_KNM.
14740 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14741 (processor_type): Add PROCESSOR_KNM.
14742 * config/i386/x86-tune.def: Add m_KNM.
14743 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14745 2017-09-20 Richard Biener <rguenther@suse.de>
14747 PR tree-optimization/80213
14748 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14749 are allowed in empty BBs as well.
14750 (canonicalize_loop_closed_ssa): Also look for other complex
14752 (scop_detection::get_sese): Include the loop-closed PHI block
14754 (scop_detection::merge_sese): Remove code adding extra blocks.
14755 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14756 (build_scops): Assert the final returned scop is invalid.
14758 2017-09-20 Richard Biener <rguenther@suse.de>
14760 PR tree-optimization/82264
14761 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14763 (vn_phi_lookup): Likewise.
14764 (vn_phi_insert): Likewise.
14766 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14768 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14769 that fits into uhwi or shwi, add DW_AT_const_value regardless
14770 of early_dwarf without going through RTL, using add_AT_unsigned
14773 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14774 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14775 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14776 (DEBUG_MACRO_SECTION): Likewise.
14777 (DEBUG_LTO_MACRO_SECTION): Likewise.
14778 (DEBUG_STR_DWO_SECTION): Likewise.
14779 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14780 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14781 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14782 (DEBUG_LTO_LINE_STR_SECTION): Define.
14783 (init_sections_and_labels): Initialize debug_line_str_section
14784 variable. Initialize debug_loc_section for -gdwarf-5 to
14785 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14787 2017-09-20 Richard Biener <rguenther@suse.de>
14789 * graphite-sese-to-poly.c (extract_affine): Properly handle
14790 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14792 2017-09-20 Richard Biener <rguenther@suse.de>
14794 PR tree-optimization/81373
14795 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14796 Force SESE live-out defs to be handled even if they are
14799 2017-09-19 Jeff Law <law@redhat.com>
14801 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
14802 nothing for stack adjustments with REG_STACK_CHECK.
14803 * sched-deps.c (parse_add_or_inc): Reject insns with
14804 REG_STACK_CHECK from dependency breaking.
14805 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
14806 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
14807 * reg-notes.def (STACK_CHECK): New note.
14809 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
14810 (ix86_expand_prologue): Dump stack clash info as needed.
14811 Call ix86_adjust_stack_and_probe_stack_clash as needed.
14813 * function.c (dump_stack_clash_frame_info): New function.
14814 * function.h (dump_stack_clash_frame_info): Prototype.
14815 (enum stack_clash_probes): New enum.
14817 * config/alpha/alpha.c (alpha_expand_prologue): Also check
14818 flag_stack_clash_protection.
14819 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14820 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
14821 (arm_frame_pointer_required): Likewise.
14822 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
14823 (ia64_expand_prologue): Likewise.
14824 * config/mips/mips.c (mips_expand_prologue): Likewise.
14825 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
14826 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14827 (sparc_flat_expand_prologue): Likewise.
14828 * config/spu/spu.c (spu_expand_prologue): Likewise.
14830 * explow.c: Include "params.h".
14831 (anti_adjust_stack_and_probe_stack_clash): New function.
14832 (get_stack_check_protect): Likewise.
14833 (compute_stack_clash_protection_loop_data): Likewise.
14834 (emit_stack_clash_protection_loop_start): Likewise.
14835 (emit_stack_clash_protection_loop_end): Likewise.
14836 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14837 Use anti_adjust_stack_and_probe_stack_clash.
14838 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14839 (emit_stack_clash_protection_loop_start): Likewise.
14840 (emit_stack_clash_protection_loop_end): Likewise.
14841 * rtl.h (get_stack_check_protect): Prototype.
14842 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14843 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14844 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14846 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14848 * doc/tm.texi: Rebuilt.
14849 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14850 get_stack_check_protect.
14851 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
14852 * config/arm/arm.c (arm_expand_prologue): Likewise.
14853 (arm_frame_pointer_required): Likewise.
14854 * config/i386/i386.c (ix86_expand_prologue): Likewise.
14855 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
14856 * config/mips/mips.c (mips_expand_prologue): Likewise.
14857 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
14858 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
14859 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14860 (sparc_flat_expand_prologue): Likewise.
14862 * common.opt (-fstack-clash-protection): New option.
14863 * flag-types.h (enum stack_check_type): Note difference between
14864 -fstack-check= and -fstack-clash-protection.
14865 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
14866 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
14867 * toplev.c (process_options): Issue warnings/errors for cases
14868 not handled with -fstack-clash-protection.
14869 * doc/invoke.texi (-fstack-clash-protection): Document new option.
14870 (-fstack-check): Note additional problem with -fstack-check=generic.
14871 Note that -fstack-check is primarily for Ada and refer users
14872 to -fstack-clash-protection for stack-clash-protection.
14873 Document new params for stack clash protection.
14875 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
14877 * config/i386/i386.c (ix86_split_long_move): Do not handle
14878 address used for LEA in a special way.
14880 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
14882 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14884 2017-09-19 Martin Sebor <msebor@redhat.com>
14887 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14888 of incompatible types.
14890 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14892 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14893 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14894 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14895 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14897 2017-09-19 Richard Biener <rguenther@suse.de>
14899 PR tree-optimization/82244
14900 * tree-vrp.c (remove_range_assertions): Do not propagate
14901 a constant to abnormals but replace the assert with a copy.
14903 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14905 PR rtl-optimization/57878
14906 PR rtl-optimization/68988
14907 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14908 avoidance test involving non_reload_pseudos. Move frequency test
14909 below the general fragmentation avoidance test.
14911 2017-09-19 Richard Biener <rguenther@suse.de>
14913 PR tree-optimization/69728
14914 * graphite-sese-to-poly.c (schedule_error): New global.
14915 (add_loop_schedule): Handle empty domain by failing the
14917 (build_original_schedule): Handle schedule_error.
14919 2017-09-19 Richard Biener <rguenther@suse.de>
14921 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14922 Do not iterate to sibling loops but only to siblings of inner
14925 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14928 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14929 that the registers are different.
14931 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14933 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14934 to processor_model and "amdfam17h" to arch_names_table.
14935 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14937 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14940 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14942 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14943 Alan Hayward <alan.hayward@arm.com>
14944 David Sherwood <david.sherwood@arm.com>
14946 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14948 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14950 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14951 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14952 vect_schedule_slp_instance.
14953 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14954 Update call to vect_slp_analyze_node_operations. Simplify return
14956 (vect_slp_analyze_bb_1): Update call accordingly.
14957 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14958 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14959 (vect_schedule_slp): Update call accordingly.
14961 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14962 Alan Hayward <alan.hayward@arm.com>
14963 David Sherwood <david.sherwood@arm.com>
14965 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14966 with types that aren't in fact scalar.
14968 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14970 * tree-vect-slp.c (vect_record_max_nunits): New function,
14972 (vect_build_slp_tree_1): ...here.
14973 (vect_build_slp_tree_2): Call it for phis too.
14975 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14977 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14978 to vect_get_vec_def_for_operand when getting the mask operand.
14980 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14981 Alan Hayward <alan.hayward@arm.com>
14982 David Sherwood <david.sherwood@arm.com>
14984 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14987 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14988 Alan Hayward <alan.hayward@arm.com>
14989 David Sherwood <david.sherwood@arm.com>
14991 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14992 calculation for vector booleans.
14994 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14995 Alan Hayward <alan.hayward@arm.com>
14996 David Sherwood <david.sherwood@arm.com>
14998 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
15000 (vect_transform_stmt): ...here.
15001 (vect_analyze_stmt): Use it instead of calling
15002 vectorizable_live_operation directly.
15004 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
15006 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
15007 non-SIMT targets in acc vector loops.
15009 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15011 * configure.ac: Add arc and check if assembler supports gdwarf2.
15012 * configure: Regenerate.
15014 2017-09-18 Richard Biener <rguenther@suse.de>
15016 PR tree-optimization/82220
15017 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15018 epilogue niters from the min_profitable_iters compute.
15020 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15023 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15024 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15025 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15027 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15030 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15031 switching to a new text section.
15033 2017-09-18 Richard Biener <rguenther@suse.de>
15035 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15037 (build_alias_set): Reject aliases with no access function.
15039 2017-09-18 Richard Biener <rguenther@suse.de>
15041 PR tree-optimization/79622
15042 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15044 (build_cross_bb_scalars_use): Likewise.
15046 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15048 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15050 2017-09-18 Alan Modra <amodra@gmail.com>
15053 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15054 stack_pointer_rtx for count 0. Update comments. Break up
15055 large rtl expression.
15057 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15059 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15060 Increase to 20 bytes.
15061 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15062 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15063 or avx version of the stub.
15065 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15068 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15069 compute the minimum stack alignment. Also update preferred stack
15070 boundary for leaf functions.
15072 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15074 PR tree-optimization/82228
15075 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15078 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15080 * common/config/nds32/nds32-common.c
15081 (nds32_option_optimization_table): Refine formatting.
15082 (nds32_option_optimization_table): Use -fsched-pressure and
15083 -fomit-frame-pointer for specific optimization level.
15085 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15087 * config/nds32/nds32.c: Refine formatting and comments.
15088 * config/nds32/nds32.h: Likewise.
15089 * config/nds32/nds32.md: Likewise.
15090 * config/nds32/nds32-cost.c: Likewise.
15091 * config/nds32/nds32-isr.c: Likewise.
15092 * config/nds32/nds32-md-auxiliary.c: Likewise.
15093 * config/nds32/nds32-multiple.md: Likewise.
15094 * config/nds32/nds32-predicates.c: Likewise.
15096 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15097 Jakub Jelinek <jakub@redhat.com>
15099 Add support for -std=c++2a.
15100 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15102 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15104 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15107 * doc/extend.texi (Common Function Attributes): Add
15108 references to ARM, AArch64, and S/390 specific attributes.
15109 (Function Specific Option Pragmas): Add AArch64 and S/390
15110 to list of back ends that support the target pragma.
15112 2017-09-15 Nathan Sidwell <nathan@acm.org>
15114 * doc/standards.texi: Fix C++17 description. Update URLs for
15117 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15119 * common.opt (Wcast-align=strict): New warning option.
15120 * doc/invoke.texi: Document -Wcast-align=strict.
15122 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15124 * cgraph.h (cgraph_thunk_info): Add comments.
15125 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15126 assert for VIRTUAL_* arguments stricter.
15128 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15130 PR tree-optimization/71026
15131 * match.pd: Move RDIV patterns from fold-const.c
15132 * fold-const.c (distribute_real_division): Removed.
15133 (fold_binary_loc): Remove calls to distribute_real_divison.
15135 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15137 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15138 c++1z and gnu++1z as deprecated. Change other references to
15139 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15140 Change -Wc++1z-compat to -Wc++17-compat.
15141 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15142 * dwarf2out.c (highest_c_language): Handle C++17.
15143 (gen_compile_unit_die): Likewise.
15145 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15147 PR rtl-optimization/82192
15148 * combine.c (make_extraction): Don't look through non-paradoxical
15149 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15152 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15153 Alan Hayward <alan.hayward@arm.com>
15154 David Sherwood <david.sherwood@arm.com>
15156 * target.def (function_arg_offset): New hook.
15157 * targhooks.h (default_function_arg_offset): Declare.
15158 * targhooks.c (default_function_arg_offset): New function.
15159 * function.c (locate_and_pad_parm): Use
15160 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15161 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15162 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15163 * doc/tm.texi: Regenerate.
15164 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15165 * config/spu/spu.c (spu_function_arg_offset): New function.
15166 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15167 * system.h (FUNCTION_ARG_OFFSET): Poison.
15169 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15170 Alan Hayard <alan.hayward@arm.com>
15171 David Sherwood <david.sherwood@arm.com>
15173 * target.def (truly_noop_truncation): New hook.
15174 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15175 than TRULY_NOOP_TRUNCATION.
15176 * hooks.h (hook_bool_uint_uint_true): Declare.
15177 * hooks.c (hook_bool_uint_uint_true): New function.
15178 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15179 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15180 * doc/tm.texi: Regenerate.
15181 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15182 rather than TRULY_NOOP_TRUNCATION in comments.
15183 (simplify_comparison): Likewise.
15184 (record_truncated_value): Likewise.
15185 * expmed.c (extract_bit_field_1): Likewise.
15186 (extract_split_bit_field): Likewise.
15187 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15188 instead of TRULY_NOOP_TRUNCATION.
15189 * function.c (assign_parm_setup_block): Likewise.
15190 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15191 * rtlhooks.c: Include target.h.
15192 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15193 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15194 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15195 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15196 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15197 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15198 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15199 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15200 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15201 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15202 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15203 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15204 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15205 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15206 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15207 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15208 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15209 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15210 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15211 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15212 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15213 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15214 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15215 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15216 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15217 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15218 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15219 rather than TRULY_NOOP_TRUNCATION in comments.
15220 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15221 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15222 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15223 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15224 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15225 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15226 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15227 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15228 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15229 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15230 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15231 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15232 rather than TRULY_NOOP_TRUNCATION in comments.
15233 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15234 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15235 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15236 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15237 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15238 TRULY_NOOP_TRUNCATION condition.
15239 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15240 (TRULY_NOOP_TRUNCATION): Delete.
15241 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15242 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15243 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15244 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15245 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15246 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15247 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15248 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15249 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15250 rather than TRULY_NOOP_TRUNCATION in comments.
15251 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15252 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15253 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15254 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15255 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15256 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15258 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15261 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15262 (*cmp_ior): Likewise.
15263 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15264 (*ior_scc_scc_cmp): Likewise.
15265 (*and_scc_scc): Likewise.
15266 (*and_scc_scc_cmp): Likewise.
15268 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15269 Alan Hayard <alan.hayward@arm.com>
15270 David Sherwood <david.sherwood@arm.com>
15272 * target.def (can_change_mode_class): New hook.
15273 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15274 (hard_regno_nregs): Likewise.
15275 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15276 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15277 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15278 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15279 (LOAD_EXTEND_OP): Update accordingly.
15280 * doc/tm.texi: Regenerate.
15281 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15282 CANNOT_CHANGE_MODE_CLASS.
15283 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15284 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15285 * combine.c (simplify_set): Update accordingly.
15286 * emit-rtl.c (validate_subreg): Likewise.
15287 * recog.c (general_operand): Likewise.
15288 * regcprop.c (mode_change_ok): Likewise.
15289 * reload1.c (choose_reload_regs): Likewise.
15290 (inherit_piecemeal_p): Likewise.
15291 * rtlanal.c (simplify_subreg_regno): Likewise.
15292 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15293 instead of CANNOT_CHANGE_MODE_CLASS.
15294 (reload_cse_simplify_operands): Likewise.
15295 * reload.c (push_reload): Use targetm.can_change_mode_class
15296 instead of CANNOT_CHANGE_MODE_CLASS.
15297 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15298 REG_CANNOT_CHANGE_MODE_P.
15299 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15300 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15301 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15302 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15303 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15304 (arm_can_change_mode_class): New function.
15305 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15306 than CANNOT_CHANGE_MODE_CLASS in comments.
15307 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15308 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15309 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15310 (ix86_can_change_mode_class): ...this new function, inverting the
15311 sense of the return value.
15312 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15313 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15314 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15315 (ia64_can_change_mode_class): New function.
15316 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15317 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15318 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15319 (m32c_can_change_mode_class): ...this new function, inverting the
15320 sense of the return value.
15321 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15322 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15323 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15324 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15325 (mips_can_change_mode_class): ...this new function, inverting the
15326 sense of the return value.
15327 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15328 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15329 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15330 (msp430_can_change_mode_class): New function.
15331 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15332 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15333 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15334 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15335 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15336 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15337 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15338 (pa_cannot_change_mode_class): Replace with...
15339 (pa_can_change_mode_class): ...this new function, inverting the
15340 sense of the return value.
15341 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15342 than CANNOT_CHANGE_MODE_CLASS in comments.
15343 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15344 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15345 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15346 (pdp11_cannot_change_mode_class): Replace with...
15347 (pdp11_can_change_mode_class): ...this new function, inverting the
15348 sense of the return value.
15349 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15350 * config/powerpcspe/powerpcspe-protos.h
15351 (rs6000_cannot_change_mode_class_ptr): Delete.
15352 * config/powerpcspe/powerpcspe.c
15353 (rs6000_cannot_change_mode_class_ptr): Delete.
15354 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15355 (rs6000_option_override_internal): Assign to
15356 targetm.can_change_mode_class instead of
15357 rs6000_cannot_change_mode_class_ptr.
15358 (rs6000_cannot_change_mode_class): Replace with...
15359 (rs6000_can_change_mode_class): ...this new function, inverting the
15360 sense of the return value.
15361 (rs6000_debug_cannot_change_mode_class): Replace with...
15362 (rs6000_debug_can_change_mode_class): ...this new function.
15363 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15364 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15365 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15366 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15367 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15369 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15370 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15371 (rs6000_option_override_internal): Assign to
15372 targetm.can_change_mode_class instead of
15373 rs6000_cannot_change_mode_class_ptr.
15374 (rs6000_cannot_change_mode_class): Replace with...
15375 (rs6000_can_change_mode_class): ...this new function, inverting the
15376 sense of the return value.
15377 (rs6000_debug_cannot_change_mode_class): Replace with...
15378 (rs6000_debug_can_change_mode_class): ...this new function.
15379 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15380 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15381 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15382 (s390_can_change_mode_class): ...this new function, inverting the
15383 sense of the return value.
15384 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15385 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15386 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15387 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15388 (sh_cannot_change_mode_class): Replace with...
15389 (sh_can_change_mode_class): ...this new function, inverting the
15390 sense of the return value.
15391 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15392 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15393 (sparc_can_change_mode_class): New function.
15394 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15395 * config/spu/spu.c (spu_can_change_mode_class): New function.
15396 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15397 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15398 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15399 (visium_can_change_mode_class): New function.
15400 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15402 2017-09-15 Richard Biener <rguenther@suse.de>
15404 PR tree-optimization/82217
15405 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15406 but not undefined case.
15408 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15411 * postreload.c (reload_cse_simplify_operands): Skip
15412 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15414 2017-09-15 Richard Biener <rguenther@suse.de>
15416 PR tree-optimization/68823
15417 * graphite-scop-detection.c (build_alias_set): If we have a
15418 possible dependence check whether we can handle them by just
15419 looking at the DRs DR_ACCESS_FNs.
15420 (build_scops): If build_alias_set fails, fail the SCOP.
15422 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15424 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15425 to support float128 built-in functions that require the ISA 3.0
15427 (BU_FLOAT128_3_HW): Likewise.
15428 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15429 built-in functions.
15430 (FMAF128): Likewise.
15432 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15433 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15434 floating point instructions.
15435 (rs6000_invalid_builtin): Likewise.
15436 (rs6000_builtin_mask_names): Likewise.
15437 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15438 (RS6000_BTM_FLOAT128_HW): Likewise.
15439 (RS6000_BTM_COMMON): Likewise.
15440 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15442 * doc/extend.texi (RS/6000 built-in functions): Document the
15443 IEEE 128-bit floating point square root and fused multiply-add
15444 built-in functions.
15446 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15448 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15449 reg (r2) isn't in the set of registers defined in the prologue.
15451 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15452 Alan Hayward <alan.hayward@arm.com>
15453 David Sherwood <david.sherwood@arm.com>
15455 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15456 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15457 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15458 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15459 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15461 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15462 max_vectorization_factor.
15463 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15465 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15466 Alan Hayward <alan.hayward@arm.com>
15467 David Sherwood <david.sherwood@arm.com>
15469 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15470 (vect_worthwhile_without_simd_p): Declare.
15471 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15472 (vectorizable_reduction): Use it.
15473 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15474 (vectorizable_operation): Likewise.
15476 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15477 Alan Hayward <alan.hayward@arm.com>
15478 David Sherwood <david.sherwood@arm.com>
15480 * tree-vectorizer.h (vect_get_num_copies): New function.
15481 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15482 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15483 (vectorizable_induction): Likewise.
15484 (vectorizable_live_operation): Likewise.
15485 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15486 (vectorizable_bswap): Likewise.
15487 (vectorizable_call): Likewise.
15488 (vectorizable_conversion): Likewise.
15489 (vectorizable_assignment): Likewise.
15490 (vectorizable_shift): Likewise.
15491 (vectorizable_operation): Likewise.
15492 (vectorizable_store): Likewise.
15493 (vectorizable_load): Likewise.
15494 (vectorizable_condition): Likewise.
15495 (vectorizable_comparison): Likewise.
15496 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15498 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15499 Alan Hayward <alan.hayward@arm.com>
15500 David Sherwood <david.sherwood@arm.com>
15502 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15503 of vect_init_vector.
15505 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15506 Alan Hayward <alan.hayward@arm.com>
15507 David Sherwood <david.sherwood@arm.com>
15509 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15510 an inline wrapper that provides a location.
15511 (gimple_build_vector): Likewise.
15512 * gimple-fold.c (gimple_build_vector_from_val): New function.
15513 (gimple_build_vector): Likewise.
15514 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15515 functions to build the initial value. Always return a gimple value.
15516 (get_initial_defs_for_reduction): Likewise. Only compute
15518 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15519 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15520 (vectorizable_induction): Use gimple_build_vector rather than
15523 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15524 Alan Hayward <alan.hayward@arm.com>
15525 David Sherwood <david.sherwood@arm.com>
15527 * target.h (vec_perm_indices): New typedef.
15528 (auto_vec_perm_indices): Likewise.
15529 * optabs-query.h: Include target.h
15530 (can_vec_perm_p): Take a vec_perm_indices *.
15531 * optabs-query.c (can_vec_perm_p): Likewise.
15532 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15533 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15534 * tree-vect-generic.c (lower_vec_perm): Likewise.
15535 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15536 (vect_grouped_load_supported): Likewise.
15537 (vect_shift_permute_load_chain): Likewise.
15538 (vect_permute_store_chain): Use auto_vec_perm_indices.
15539 (vect_permute_load_chain): Likewise.
15540 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15541 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15542 Update uses of can_vec_perm_p.
15543 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15544 mode with a number of elements. Take a vec_perm_indices *.
15545 (vect_create_epilog_for_reduction): Update accordingly.
15546 Use auto_vec_perm_indices.
15547 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15548 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15549 (vect_transform_slp_perm_load): Likewise.
15550 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15551 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15552 (vect_gen_perm_mask_checked): Likewise.
15553 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15554 (vect_gen_perm_mask_checked): Likewise.
15555 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15556 (vectorizable_store): Likewise.
15557 (vectorizable_load): Likewise.
15558 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15559 (vectorizable_bswap): Likewise.
15561 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15562 Alan Hayward <alan.hayward@arm.com>
15563 David Sherwood <david.sherwood@arm.com>
15565 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15566 * tree.c (build_vector): Likewise.
15567 (build_vector_from_ctor): Update accordingly.
15568 (build_vector_from_val): Likewise.
15569 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15570 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15571 * tree-vect-generic.c (add_rshift): Likewise.
15572 (expand_vector_divmod): Likewise.
15573 (optimize_vector_constructor): Likewise.
15574 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15575 (vect_transform_slp_perm_load): Likewise.
15576 (vect_schedule_slp_instance): Likewise.
15577 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15578 (vectorizable_call): Likewise.
15579 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15580 * expmed.c (make_tree): Likewise.
15581 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15582 a vector passed to build_vector.
15583 (fold_convert_const): Likewise.
15584 (exact_inverse): Likewise.
15585 (fold_ternary_loc): Likewise.
15586 (fold_relational_const): Likewise.
15587 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15588 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15589 (const_unop): Likewise. Store the reduction accumulator in a
15590 variable rather than an array.
15591 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15592 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15593 the new vector, rather than constructing it after the input arrays.
15594 (native_interpret_vector): Use auto_vec<tree> when building
15595 a vector passed to build_vector. Add elements in order.
15596 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15597 auto_vec<tree> when building a vector passed to build_vector.
15598 (vect_create_epilog_for_reduction): Likewise.
15599 (vectorizable_induction): Likewise.
15600 (get_initial_def_for_reduction): Likewise. Fix indentation of
15602 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15604 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15605 passed to build_vector.
15607 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15608 Alan Hayward <alan.hayward@arm.com>
15609 David Sherwood <david.sherwood@arm.com>
15611 * tree-core.h (tree_base::u): Add an "nelts" field.
15612 (tree_vector): Use VECTOR_CST_NELTS as the length.
15613 * tree.c (tree_size): Likewise.
15614 (make_vector): Initialize VECTOR_CST_NELTS.
15615 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15616 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15617 TYPE_VECTOR_SUBPARTS.
15618 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15619 as the number of units, setting it from VECTOR_CST_NELTS.
15620 (const_vector_from_tree): Likewise.
15621 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15622 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15623 (fold_negate_expr_1): Likewise.
15624 (fold_convert_const): Likewise.
15625 (const_binop): Likewise. Differentiate the number of output and
15627 (const_unop): Likewise.
15628 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15629 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15630 in cases that did the opposite.
15632 2017-09-14 Richard Biener <rguenther@suse.de>
15634 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15637 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15639 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15641 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15644 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15645 if and where to split a bb, except for splitting before debug insn
15646 sequences followed by non-label real insn. Delete debug insns
15647 in between basic blocks.
15649 * combine.c (make_compound_operation_int): Formatting fixes.
15651 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15652 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15653 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15654 * config/sol2.h (LINK_EH_SPEC): Likewise.
15655 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15656 * config/s390/linux.h (LINK_SPEC): Likewise.
15657 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15658 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15659 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15660 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15661 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15662 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15663 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15664 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15666 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15668 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15670 (ENDFILE_LINUX_SPEC): Likewise.
15671 (LINK_EH_SPEC): Likewise.
15672 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15673 (LINK_OS_LINUX_SPEC32): Likewise.
15674 (LINK_OS_LINUX_SPEC64): Likewise.
15675 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15676 (LINK_OS_LINUX_SPEC): Likewise.
15678 2017-09-13 Martin Liska <mliska@suse.cz>
15680 PR middle-end/82154
15681 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15682 CASE_HIGH is NULL_TREE.
15684 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15685 Alan Hayward <alan.hayward@arm.com>
15686 David Sherwood <david.sherwood@arm.com>
15688 * target.def (secondary_memory_needed): New hook.
15689 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15690 instead of SECONDARY_MEMORY_NEEDED.
15691 (secondary_memory_needed_mode): Likewise.
15692 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15693 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15694 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15695 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15696 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15697 * doc/tm.texi: Regenerate.
15698 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15699 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15700 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15701 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15702 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15703 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15704 mode argument first and change the reg_class arguments to reg_class_t.
15705 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15706 Make static. Update the call to inline_secondary_memory_needed.
15707 (ix86_register_move_cost): Update the call to
15708 inline_secondary_memory_needed.
15709 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15710 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15712 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15713 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15715 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15716 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15717 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15718 and match hook interface. Add comment from mips.h.
15719 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15720 * config/mmix/mmix.md (truncdfsf2): Refer to
15721 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15723 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15724 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15725 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15726 (pa_secondary_memory_needed): New function.
15727 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15728 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15729 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15730 (pdp11_secondary_memory_needed): Make static and match hook interface.
15731 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15732 * config/powerpcspe/powerpcspe-protos.h
15733 (rs6000_secondary_memory_needed_ptr): Delete.
15734 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15736 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15737 (rs6000_option_override_internal): Assign to
15738 targetm.secondary_memory_needed rather than
15739 rs6000_secondary_memory_needed_ptr.
15740 (rs6000_secondary_memory_needed): Match hook interface.
15741 (rs6000_debug_secondary_memory_needed): Likewise.
15742 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15743 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15744 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15745 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15746 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15747 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15749 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15750 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15751 (rs6000_option_override_internal): Assign to
15752 targetm.secondary_memory_needed rather than
15753 rs6000_secondary_memory_needed_ptr.
15754 (rs6000_secondary_memory_needed): Match hook interface.
15755 (rs6000_debug_secondary_memory_needed): Likewise.
15756 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15757 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15758 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15759 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15760 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15761 (sparc_secondary_memory_needed): New function.
15762 * lra-constraints.c (check_and_process_move): Refer to
15763 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15765 (curr_insn_transform): Likewise.
15766 (process_alt_operands): Use targetm.secondary_memory_needed
15767 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15768 (check_secondary_memory_needed_p): Likewise.
15769 (choose_split_class): Likewise.
15770 * reload.c: Unconditionally include code that was previously
15771 conditional on SECONDARY_MEMORY_NEEDED.
15772 (push_secondary_reload): Use targetm.secondary_memory_needed
15773 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15774 (push_reload): Likewise.
15775 * reload1.c: Unconditionally include code that was previously
15776 conditional on SECONDARY_MEMORY_NEEDED.
15777 (choose_reload_regs): Use targetm.secondary_memory_needed
15778 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15779 (gen_reload): Likewise.
15780 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15782 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15783 Alan Hayward <alan.hayward@arm.com>
15784 David Sherwood <david.sherwood@arm.com>
15786 * target.def (secondary_memory_needed_mode): New hook:
15787 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15788 * targhooks.h (default_secondary_memory_needed_mode): New function.
15789 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15790 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15791 * doc/tm.texi: Regenerate.
15792 * lra-constraints.c (check_and_process_move): Use
15793 targetm.secondary_memory_needed_mode instead of
15794 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15795 (curr_insn_transform): Likewise.
15796 * reload.c (get_secondary_mem): Likewise.
15797 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15798 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
15800 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15801 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15802 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
15803 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15804 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
15806 * config/powerpcspe/powerpcspe-protos.h
15807 (rs6000_secondary_memory_needed_mode): Delete.
15808 * config/powerpcspe/powerpcspe.c
15809 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15810 (rs6000_secondary_memory_needed_mode): Make static.
15811 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15812 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
15814 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15816 (rs6000_secondary_memory_needed_mode): Make static.
15817 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15818 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
15819 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15820 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15821 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15823 (sparc_secondary_memory_needed_mode): New function.
15824 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
15826 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15828 * config/aarch64/constraints.md (Umq): New constraint.
15829 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15831 (mov<mode>): Update condition.
15833 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15835 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15836 when bitposition is the same.
15838 2017-09-13 Richard Biener <rguenther@suse.de>
15840 * dwarf2out.c (output_die_symbol): Remove.
15841 (output_die): Do not output a DIEs symbol.
15843 2017-09-13 Richard Biener <rguenther@suse.de>
15845 PR middle-end/82128
15846 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15847 default-def to avoid breaking iterator update with the weird
15848 interaction with cgraph_update_edges_for_call_stmt_node.
15850 2017-09-13 Richard Biener <rguenther@suse.de>
15852 * tree-cfg.c (verify_gimple_assign_binary): Add verification
15853 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
15854 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
15855 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
15857 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15859 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15860 Disable pc relative literal load irrespective of
15861 TARGET_FIX_ERR_A53_84341 for default.
15863 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
15865 * config/sparc/sparc.c (output_return): Output the source location of
15866 the insn in the delay slot, if any.
15867 (output_sibcall): Likewise.
15869 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
15872 * common.opt (-static-pie): New alias.
15873 (shared): Negate static-pie.
15874 (-no-pie): Update help text.
15876 (static-pie): New option.
15877 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
15878 -static-pie support.
15879 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
15880 (LINK_EH_SPEC): Likewise.
15881 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15882 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15883 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15884 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15885 (init_gcc_specs): Likewise.
15886 (init_spec): Likewise.
15887 (display_help): Update help message for -pie.
15888 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15891 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15893 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15894 (movdi_aarch64): Likewise.
15895 (movti_aarch64): Likewise.
15897 2017-09-12 Simon Wright <simon@pushface.org>
15900 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15901 calculation of the minor version, always output as 0.
15903 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15906 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15907 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15908 on it early, rather than manual conversion late. For
15909 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15910 instead of performing manual conversion.
15912 2017-09-12 Carl Love <cel@us.ibm.com>
15914 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15915 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15917 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15918 VMULOSW): Add definitions.
15919 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15920 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15921 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15922 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15923 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15925 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15927 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15929 (movti_aarch64): Likewise.
15930 (movdf_aarch64): Likewise.
15931 (movtf_aarch64): Likewise.
15932 (load_pairdi): Likewise.
15933 (store_pairdi): Likewise.
15934 (load_pairdf): Likewise.
15935 (store_pairdf): Likewise.
15936 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15937 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15938 (ldr_got_small_<mode>): Likewise.
15939 (ldr_got_small_28k_<mode>): Likewise.
15940 (ldr_got_tiny): Likewise.
15941 * config/aarch64/iterators.md (ldst_sz): New.
15942 (ldpstp_sz): Likewise.
15943 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15945 (thunderx_load): Split load_8 to load_16.
15946 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15948 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15949 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15950 (xgene1_store_pair): Split store_8 to store_16.
15951 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15952 (falkor_st_0_st_sd): Split store_8 to store_16.
15954 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15956 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15957 and store1/2/3/4 to store_4/8/12/16.
15958 * config/aarch64/aarch64.md: Update for rename.
15959 * config/arm/arm.md: Likewise.: Likewise.
15960 * config/arm/arm.c: Likewise.
15961 * config/arm/thumb1.md: Likewise.
15962 * config/arm/thumb2.md: Likewise.
15963 * config/arm/vfp.md: Likewise.
15964 * config/arm/arm-generic.md: Likewise.
15965 * config/arm/arm1020e.md: Likewise.
15966 * config/arm/arm1026ejs.md: Likewise.
15967 * config/arm/arm1136jfs.md: Likewise.
15968 * config/arm/arm926ejs.md: Likewise.
15969 * config/arm/cortex-a15.md: Likewise.
15970 * config/arm/cortex-a17.md: Likewise.
15971 * config/arm/cortex-a5.md: Likewise.
15972 * config/arm/cortex-a53.md: Likewise.
15973 * config/arm/cortex-a57.md: Likewise.
15974 * config/arm/cortex-a7.md: Likewise.
15975 * config/arm/cortex-a8.md: Likewise.
15976 * config/arm/cortex-a9.md: Likewise.
15977 * config/arm/cortex-m4.md: Likewise.
15978 * config/arm/cortex-m7.md: Likewise.
15979 * config/arm/cortex-r4.md: Likewise.
15980 * config/arm/exynos-m1.md: Likewise.
15981 * config/arm/fa526.md: Likewise.
15982 * config/arm/fa606te.md: Likewise.
15983 * config/arm/fa626te.md: Likewise.
15984 * config/arm/fa726te.md: Likewise.
15985 * config/arm/fmp626.md: Likewise.
15986 * config/arm/iwmmxt.md: Likewise.
15987 * config/arm/ldmstm.md: Likewise.
15988 * config/arm/marvell-pj4.md: Likewise.
15989 * config/arm/xgene1.md: Likewise.
15990 * config/aarch64/thunderx.md: Likewise.
15991 * config/aarch64/thunderx2t99.md: Likewise.
15992 * config/aarch64/falkor.md: Likewise.
15994 2017-09-12 Martin Liska <mliska@suse.cz>
15996 * attribs.c (private_lookup_attribute): New function.
15997 * attribs.h (private_lookup_attribute): Declared here.
15998 (lookup_attribute): Called from this place.
16000 2017-09-12 Richard Biener <rguenther@suse.de>
16002 PR tree-optimization/82157
16003 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
16004 stmts with side-effects.
16006 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16007 Alan Hayward <alan.hayward@arm.com>
16008 David Sherwood <david.sherwood@arm.com>
16010 * target.def (hard_regno_nregs): New hook.
16011 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16012 * targhooks.h (default_hard_regno_nregs): Declare.
16013 * targhooks.c (default_hard_regno_nregs): New function.
16014 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16015 (TARGET_HARD_REGNO_NREGS): ...this hook.
16016 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16017 (CLASS_MAX_NREGS): Likewise.
16018 * doc/tm.texi: Regenerate.
16019 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16020 instead of HARD_REGNO_NREGS.
16021 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16022 HARD_REGNO_NREGS in the comment.
16023 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16024 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16025 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16026 Return an unsigned int.
16027 (TARGET_HARD_REGNO_NREGS): Redefine.
16028 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16029 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16030 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16031 (arc_hard_regno_nregs): New function.
16032 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16033 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16034 (arm_hard_regno_nregs): New function.
16035 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16036 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16037 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16038 (TARGET_HARD_REGNO_NREGS): Redefine.
16039 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16040 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16041 (HARD_REGNO_NREGS): Delete.
16042 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16043 (cr16_hard_regno_nregs): New function.
16044 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16045 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16046 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16047 (cris_hard_regno_nregs): New function.
16048 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16049 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16050 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16051 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16052 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16053 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16054 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16055 (frv_hard_regno_nregs): Make static. Take and return an
16057 (frv_class_max_nregs): Remove outdated copy of documentation.
16058 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16059 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16060 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16061 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16062 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16063 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16064 (TARGET_HARD_REGNO_NREGS): Redefine.
16065 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16066 (CLASS_MAX_NREGS): Update comment.
16067 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16068 (ia64_hard_regno_nregs): New function.
16069 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16070 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16071 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16072 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16073 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16075 (m32c_hard_regno_nregs): Likewise. Make static.
16076 (TARGET_HARD_REGNO_NREGS): Redefine.
16077 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16078 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16079 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16080 (m68k_hard_regno_nregs): New function.
16081 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16082 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16083 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16084 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16085 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16086 Take and return an unsigned int.
16087 (TARGET_HARD_REGNO_NREGS): Redefine.
16088 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16089 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16090 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16091 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16092 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16093 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16094 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16095 (msp430_hard_regno_nregs): Make static. Take and return an
16097 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16098 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16099 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16100 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16101 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16102 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16103 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16104 (TARGET_HARD_REGNO_NREGS): Redefine.
16105 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16106 (PA_HARD_REGNO_NREGS): ...this.
16107 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16108 (PA_HARD_REGNO_NREGS): ...this.
16109 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16110 (pa_hard_regno_nregs): New function.
16111 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16112 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16113 (pdp11_hard_regno_nregs): New function.
16114 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16115 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16116 (rs6000_hard_regno_nregs_hook): New function.
16117 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16118 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16119 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16120 Take and return an unsigned int. Move earlier in file.
16121 (TARGET_HARD_REGNO_NREGS): Redefine.
16122 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16123 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16124 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16125 (rl78_hard_regno_nregs): Make static. Take and return an
16127 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16128 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16129 (rs6000_hard_regno_nregs_hook): New function.
16130 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16131 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16132 (TARGET_HARD_REGNO_NREGS): Redefine.
16133 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16134 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16135 instead of HARD_REGNO_NREGS.
16136 (s390_hard_regno_nregs): New function.
16137 (s390_hard_regno_mode_ok): Add comment from s390.h.
16138 (TARGET_HARD_REGNO_NREGS): Redefine.
16139 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16140 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16141 (sh_hard_regno_nregs): New function.
16142 (sh_pass_in_reg_p): Use it.
16143 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16144 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16145 (sparc_hard_regno_nregs): New function.
16146 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16147 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16148 (spu_function_arg_advance): Use it, supplying a valid register number.
16149 (TARGET_HARD_REGNO_NREGS): Redefine.
16150 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16151 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16152 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16153 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16154 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16155 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16156 (CLASS_MAX_NREGS): Remove copy of old documentation.
16157 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16158 (visium_hard_regno_nregs): New function.
16159 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16160 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16161 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16162 xtensa_hard_regno_nregs): New function.
16163 * system.h (HARD_REGNO_NREGS): Poison.
16165 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16167 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16168 hard_regno_nregs instead of HARD_REGNO_NREGS.
16169 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16170 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16171 (c6x_expand_epilogue): Likewise.
16172 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16173 (frv_read_iacc_argument): Likewise.
16174 * config/sh/sh.c: Include regs.h.
16175 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16176 (regs_used): Likewise.
16177 (output_stack_adjust): Likewise.
16178 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16179 * expmed.c: Include regs.h.
16180 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16181 * ree.c: Include regs.h.
16182 (combine_reaching_defs): Use hard_regno_nregs instead of
16184 (add_removable_extension): Likewise.
16186 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16188 * regs.h (hard_regno_nregs): Turn into a function.
16189 (end_hard_regno): Update accordingly.
16190 * caller-save.c (setup_save_areas): Likewise.
16191 (save_call_clobbered_regs): Likewise.
16192 (replace_reg_with_saved_mem): Likewise.
16193 (insert_restore): Likewise.
16194 (insert_save): Likewise.
16195 * combine.c (can_change_dest_mode): Likewise.
16196 (move_deaths): Likewise.
16197 (distribute_notes): Likewise.
16198 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16199 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16200 (rs6000_split_multireg_move): Likewise.
16201 (rs6000_register_move_cost): Likewise.
16202 (rs6000_memory_move_cost): Likewise.
16203 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16204 (rs6000_split_multireg_move): Likewise.
16205 (rs6000_register_move_cost): Likewise.
16206 (rs6000_memory_move_cost): Likewise.
16207 * cselib.c (cselib_reset_table): Likewise.
16208 (cselib_lookup_1): Likewise.
16209 * emit-rtl.c (set_mode_and_regno): Likewise.
16210 * function.c (aggregate_value_p): Likewise.
16211 * ira-color.c (setup_profitable_hard_regs): Likewise.
16212 (check_hard_reg_p): Likewise.
16213 (calculate_saved_nregs): Likewise.
16214 (assign_hard_reg): Likewise.
16215 (improve_allocation): Likewise.
16216 (calculate_spill_cost): Likewise.
16217 * ira-emit.c (modify_move_list): Likewise.
16218 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16219 (ira_hard_reg_in_set_p): Likewise.
16220 * ira.c (setup_reg_mode_hard_regset): Likewise.
16221 (clarify_prohibited_class_mode_regs): Likewise.
16222 (check_allocation): Likewise.
16223 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16224 (lra_setup_reg_renumber): Likewise.
16225 (setup_try_hard_regno_pseudos): Likewise.
16226 (spill_for): Likewise.
16227 (assign_hard_regno): Likewise.
16228 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16229 * lra-constraints.c (in_class_p): Likewise.
16230 (lra_constraint_offset): Likewise.
16231 (simplify_operand_subreg): Likewise.
16232 (lra_constraints): Likewise.
16233 (split_reg): Likewise.
16234 (split_if_necessary): Likewise.
16235 (invariant_p): Likewise.
16236 (inherit_in_ebb): Likewise.
16237 * lra-lives.c (process_bb_lives): Likewise.
16238 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16239 (get_hard_regs): Likewise.
16240 (do_remat): Likewise.
16241 * lra-spills.c (assign_spill_hard_regs): Likewise.
16242 * mode-switching.c (create_pre_exit): Likewise.
16243 * postreload.c (reload_combine_recognize_pattern): Likewise.
16244 * recog.c (peep2_find_free_register): Likewise.
16245 * regcprop.c (kill_value_regno): Likewise.
16246 (set_value_regno): Likewise.
16247 (copy_value): Likewise.
16248 (maybe_mode_change): Likewise.
16249 (find_oldest_value_reg): Likewise.
16250 (copyprop_hardreg_forward_1): Likewise.
16251 * regrename.c (check_new_reg_p): Likewise.
16252 (regrename_do_replace): Likewise.
16253 * reload.c (push_reload): Likewise.
16254 (combine_reloads): Likewise.
16255 (find_dummy_reload): Likewise.
16256 (operands_match_p): Likewise.
16257 (find_reloads): Likewise.
16258 (find_equiv_reg): Likewise.
16259 (reload_adjust_reg_for_mode): Likewise.
16260 * reload1.c (count_pseudo): Likewise.
16261 (count_spilled_pseudo): Likewise.
16262 (find_reg): Likewise.
16263 (clear_reload_reg_in_use): Likewise.
16264 (free_for_value_p): Likewise.
16265 (allocate_reload_reg): Likewise.
16266 (choose_reload_regs): Likewise.
16267 (reload_adjust_reg_for_temp): Likewise.
16268 (emit_reload_insns): Likewise.
16269 (delete_output_reload): Likewise.
16270 * rtlanal.c (subreg_get_info): Likewise.
16271 * sched-deps.c (sched_analyze_reg): Likewise.
16272 * sel-sched.c (init_regs_for_mode): Likewise.
16273 (mark_unavailable_hard_regs): Likewise.
16274 (choose_best_reg_1): Likewise.
16275 (verify_target_availability): Likewise.
16276 * valtrack.c (dead_debug_insert_temp): Likewise.
16277 * var-tracking.c (track_loc_p): Likewise.
16278 (emit_note_insn_var_location): Likewise.
16279 * varasm.c (make_decl_rtl): Likewise.
16280 * reginfo.c (choose_hard_reg_mode): Likewise.
16281 (init_reg_modes_target): Refer directly to
16282 this_target_regs->x_hard_regno_nregs.
16284 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16286 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16287 instead of hard_regno_nregs.
16289 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16291 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16292 end_hard_regno instead of hard_regno_nregs.
16293 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16294 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16295 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16296 * ira-color.c (improve_allocation): Likewise.
16297 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16298 * lra-lives.c (mark_regno_live): Likewise.
16299 (mark_regno_dead): Likewise.
16300 * lra-remat.c (operand_to_remat): Likewise.
16301 * lra.c (collect_non_operand_hard_regs): Likewise.
16302 * postreload.c (reload_combine_note_store): Likewise.
16303 (move2add_valid_value_p): Likewise.
16304 * reload.c (regno_clobbered_p): Likewise.
16306 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16308 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16310 * config/v850/v850.c (v850_reorg): Likewise.
16311 * reload.c (refers_to_regno_for_reload_p): Likewise.
16312 (find_equiv_reg): Likewise.
16313 * reload1.c (reload_reg_reaches_end_p): Likewise.
16315 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16317 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16319 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16320 * config/arm/arm.c (output_move_neon): Likewise.
16321 (arm_attr_length_move_neon): Likewise.
16322 (neon_split_vcombine): Likewise.
16323 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16324 (c6x_mark_reg_written): Likewise.
16325 (c6x_dwarf_register_span): Likewise.
16326 * config/i386/i386.c (ix86_save_reg): Likewise.
16327 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16328 (rws_access_reg): Likewise.
16329 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16330 * mode-switching.c (create_pre_exit): Likewise.
16331 * ree.c (combine_reaching_defs): Likewise.
16332 (add_removable_extension): Likewise.
16333 * regcprop.c (find_oldest_value_reg): Likewise.
16334 (copyprop_hardreg_forward_1): Likewise.
16335 * reload.c (reload_inner_reg_of_subreg): Likewise.
16336 (push_reload): Likewise.
16337 (combine_reloads): Likewise.
16338 (find_dummy_reload): Likewise.
16339 (reload_adjust_reg_for_mode): Likewise.
16340 * reload1.c (find_reload_regs): Likewise.
16341 (forget_old_reloads_1): Likewise.
16342 (reload_reg_free_for_value_p): Likewise.
16343 (reload_adjust_reg_for_temp): Likewise.
16344 (emit_reload_insns): Likewise.
16345 (delete_output_reload): Likewise.
16346 * sel-sched.c (choose_best_reg_1): Likewise.
16347 (choose_best_pseudo_reg): Likewise.
16349 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16350 Alan Hayward <alan.hayward@arm.com>
16351 David Sherwood <david.sherwood@arm.com>
16353 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16354 * target.def (slow_unaligned_access): New hook.
16355 * targhooks.h (default_slow_unaligned_access): Declare.
16356 * targhooks.c (default_slow_unaligned_access): New function.
16357 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16358 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16359 * doc/tm.texi: Regenerate.
16360 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16361 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16362 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16364 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16365 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16367 (rs6000_slow_unaligned_access): New function.
16368 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16369 (expand_block_compare): Likewise.
16370 (expand_strn_compare): Likewise.
16371 (rs6000_rtx_costs): Likewise.
16372 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16373 (riscv_slow_unaligned_access): Likewise.
16374 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16375 (riscv_slow_unaligned_access_p): ...this and make static.
16376 (riscv_option_override): Update accordingly.
16377 (riscv_slow_unaligned_access): New function.
16378 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16379 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16380 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16381 (rs6000_slow_unaligned_access): New function.
16382 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16383 (rs6000_rtx_costs): Likewise.
16384 * config/rs6000/rs6000-string.c (expand_block_compare)
16385 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16386 of SLOW_UNALIGNED_ACCESS.
16387 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16388 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16389 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16390 of SLOW_UNALIGNED_ACCESS.
16391 * expmed.c (simple_mem_bitfield_p): Likewise.
16392 * expr.c (alignment_for_piecewise_move): Likewise.
16393 (emit_group_load_1): Likewise.
16394 (emit_group_store): Likewise.
16395 (copy_blkmode_from_reg): Likewise.
16396 (emit_push_insn): Likewise.
16397 (expand_assignment): Likewise.
16398 (store_field): Likewise.
16399 (expand_expr_real_1): Likewise.
16400 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16401 * lra-constraints.c (simplify_operand_subreg): Likewise.
16402 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16403 * gimple-ssa-store-merging.c: Likewise in block comment at start
16405 * tree-ssa-strlen.c: Include target.h.
16406 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16407 of SLOW_UNALIGNED_ACCESS.
16408 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16410 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16412 PR rtl-optimization/82185
16413 * expmed.c (emit_store_flag_int): Only test tem if it has been
16416 2017-09-12 Richard Biener <rguenther@suse.de>
16418 PR middle-end/82149
16419 * match.pd ((FTYPE) N CMP CST): Fix typo.
16421 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16423 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16425 (mips_near_type_p): Add 'short_call' attribute as a synonym
16427 * doc/extend.texi (short_call): Document new function attribute.
16429 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16432 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16433 assertion check that in the condition.
16434 (get_atomic_generic_size): Likewise. Before testing if parameter
16435 has pointer type, if it has array type, call for C++
16436 default_conversion to perform array-to-pointer conversion.
16438 2017-09-12 Richard Biener <rguenther@suse.de>
16440 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16441 for operations we cannot scalarize.
16443 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16445 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16446 vectors heap vectors. Clean up comments.
16447 Make visited_bbs a reference.
16448 (profitable_jump_thread_path): Make GC
16449 vectors heap vectors. Clean up comments.
16451 (convert_and_register_jump_thread_path): Make GC vectors heap
16453 (check_subpath_and_update_thread_path): Same. Clean up comments.
16454 Make visited_bbs a reference.
16455 (handle_phi): Abstract common code to to
16456 register_jump_thread_path_if_profitable.
16457 Rename VAR_BB to DEF_BB.
16459 Make GC vectors heap vectors.
16460 Make visited_bbs a reference.
16461 (handle_assignment): Same.
16462 (register_jump_thread_path_if_profitable): New.
16463 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16465 Make GC vectors heap vectors. Clean up comments.
16466 Make visited_bbs a reference.
16467 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16468 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16471 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16474 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16475 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16477 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16479 Revert r251800 and r251799.
16481 2017-09-11 Martin Jambor <mjambor@suse.cz>
16484 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16485 arguments in advance.
16486 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16487 use it to find predecessor edges.
16488 (naive_outof_ssa): Collect vector of predecessors.
16490 2017-09-08 Jason Merrill <jason@redhat.com>
16492 PR c++/70029 - ICE with ref-qualifier and -flto
16493 * langhooks.h (struct lang_hooks_for_types): Add
16494 copy_lang_qualifiers.
16495 * attribs.c (build_type_attribute_qual_variant): Use it.
16496 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16498 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16499 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16501 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16504 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16505 (*mulsi3_sp64): New instruction.
16506 (mulsi3): New expander.
16508 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16510 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16512 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16514 * sancov.c: Include memmodel.h.
16516 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16519 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16522 2017-09-07 Carl Love <cel@us.ibm.com>
16524 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16525 the sldi instruction.
16527 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16529 * sancov.c: Include tm_p.h.
16531 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16534 * output.h (switch_to_other_text_partition): New declaration.
16535 * varasm.c (switch_to_other_text_partition): New function.
16536 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16537 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16538 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16539 to the other text partition before emitting LCL label and switch back
16540 after emitting the word after it.
16542 2017-09-07 Richard Biener <rguenther@suse.de>
16544 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16545 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16546 critical edges here, after loop init.
16547 (pass_data_pre): Remove PROP_no_crit_edges flags.
16548 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16549 for valueization of call args to avoid leaking VN_TOP.
16550 (visit_use): Assert we do not visit default defs.
16551 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16552 Use error_mark_node to more easily detect leaking VN_TOP.
16553 All default-defs are varying, not VN_TOP. Mark them visited.
16554 (run_scc_vn): Make code match comment.
16556 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16558 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16559 OPTION_MASK_FLOAT128_KEYWORD.
16560 (POWERPC_MASKS): Likewise.
16561 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16562 support for the -mfloat128-type option, and make -mfloat128
16563 default on PowerPC Linux systems. Define or undefine
16564 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16565 Define __float128 to be __ieee128 if IEEE 128-bit support is
16566 enabled, or undefine it.
16567 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16568 Delete defining __FLOAT128_TYPE__.
16569 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16570 -mfloat128-type option and make -mfloat128 default on PowerPC
16572 (TARGET_FLOAT128_TYPE): Likewise.
16573 (-mfloat128-type): Likewise.
16574 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16575 Delete the -mfloat128-type option and make -mfloat128 default on
16576 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16577 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16578 128-bit floating point is enabled. Change tests from using
16579 -mfloat128-type to -mfloat128.
16580 (rs6000_mangle_type): Use the correct mangling for the __float128
16581 type even if normal long double is restricted to 64-bits.
16582 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16584 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16585 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16586 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16588 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16589 documentation for -mfloat128.
16591 2017-09-06 Olivier Hainque <hainque@adacore.com>
16593 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16595 2017-09-06 Wish Wu <wishwu007@gmail.com>
16596 Jakub Jelinek <jakub@redhat.com>
16598 * asan.c (initialize_sanitizer_builtins): Add
16599 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16600 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16601 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16602 BT_FN_VOID_UINT64_PTR variables.
16603 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16604 (BT_FN_VOID_UINT16_UINT16): Likewise.
16605 (BT_FN_VOID_UINT32_UINT32): Likewise.
16606 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16607 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16608 (BT_FN_VOID_UINT64_PTR): Likewise.
16609 * common.opt (flag_sanitize_coverage): New variable.
16610 (fsanitize-coverage=trace-pc): Remove.
16611 (fsanitize-coverage=): Add.
16612 * flag-types.h (enum sanitize_coverage_code): New enum.
16613 * fold-const.c (fold_range_test): Disable non-short-circuit
16614 optimization if flag_sanitize_coverage.
16615 (fold_truth_andor): Likewise.
16616 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16617 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16618 (coverage_sanitizer_opts): New array.
16619 (get_closest_sanitizer_option): Add OPTS argument, handle also
16620 OPT_fsanitize_coverage_.
16621 (parse_sanitizer_options): Adjusted to also handle
16622 OPT_fsanitize_coverage_.
16623 (common_handle_option): Add OPT_fsanitize_coverage_.
16624 * sancov.c (instrument_comparison, instrument_switch): New function.
16625 (sancov_pass): Add trace-cmp support.
16626 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16627 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16628 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16629 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16630 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16631 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16632 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16633 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16634 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16635 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16637 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16639 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16640 error. Only quit immediately if parsing is complete.
16641 (BEGIN): Initialize fatal_err and parse_done.
16642 (begin fpu, end fpu): Check number of arguments.
16643 (begin arch, end arch): Likewise.
16644 (begin cpu, end cpu): Likewise.
16645 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16646 (optalias): Likewise.
16648 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16650 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16651 * config/arm/arm-isa.h: Delete. Move definitions to ...
16652 * arm-cpus.in: ... here. Use new feature and fgroup values.
16653 * config/arm/arm.c (arm_option_override): Use lower case for feature
16655 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16656 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16657 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16659 (print_isa_bits_for): New function.
16660 (gen_isa): New function.
16661 (gen_comm_data): Use print_isa_bits_for.
16662 (define feature): New keyword.
16663 (define fgroup): New keyword.
16664 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16665 (arm-isa.h): Add rule to generate file.
16666 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16667 case for feature bit names.
16669 2017-09-06 Richard Biener <rguenther@suse.de>
16671 * tree-ssa-pre.c (NECESSARY): Remove.
16672 (create_expression_by_pieces): Do not touch pass-local flags.
16673 (insert_into_preds_of_block): Likewise.
16674 (do_pre_regular_insertion): Likewise.
16675 (eliminate_insert): Likewise.
16676 (eliminate_dom_walker::before_dom_children): Likewise.
16677 (fini_eliminate): Do not look at inserted_exprs.
16678 (mark_operand_necessary): Remove.
16679 (remove_dead_inserted_code): Replace with simple work-list
16680 algorithm based on inserted_exprs and SSA uses.
16681 (pass_pre::execute): Re-order fini_eliminate and
16682 remove_dead_inserted_code.
16684 2017-09-06 Olivier Hainque <hainque@adacore.com>
16686 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16687 for VxWorks 7. Adjust surrounding comments.
16689 2017-09-06 Richard Biener <rguenther@suse.de>
16691 * gimple-ssa-strength-reduction.c
16692 (find_candidates_dom_walker::before_dom_children): Also allow
16695 2017-09-06 Richard Biener <rguenther@suse.de>
16697 PR tree-optimization/82108
16698 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16699 for gap in the non-permutation SLP case.
16701 2017-09-06 Martin Jambor <mjambor@suse.cz>
16703 PR tree-optimization/82078
16704 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16705 add_access_to_work_queue...
16706 (build_accesses_from_assign): ...here.
16707 (propagate_all_subaccesses): Make sure racc is the group
16708 representative, if there is one.
16710 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16712 PR middle-end/82095
16713 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16716 2017-09-06 Richard Biener <rguenther@suse.de>
16718 * gimple-ssa-strength-reduction.c
16719 (find_candidates_dom_walker::before_doom_children): Use a
16720 type and not a mode check.
16722 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16725 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16727 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16729 2017-09-05 Jeff Law <law@redhat.com>
16731 PR tree-optimization/64910
16732 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16733 cases where we have 3 or more operands.
16735 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16737 PR middle-end/81768
16738 * omp-low.c (lower_omp_for): Recompute tree invariant if
16739 gimple_omp_for_initial/final is ADDR_EXPR.
16741 PR middle-end/81768
16742 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16743 into gimple val before gimplification fo the COND_EXPR.
16745 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16747 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16748 REGION_COPY argument.
16749 (thread_through_all_blocks): Remove unused argument to
16750 duplicate_thread_path.
16752 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16753 Alan Hayward <alan.hayward@arm.com>
16754 David Sherwood <david.sherwood@arm.com>
16756 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16757 Take a scalar_mode rather than a machine_mode.
16758 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16759 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16760 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16761 (aarch64_gen_adjusted_ldpstp): Likewise.
16762 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16764 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16765 Alan Hayward <alan.hayward@arm.com>
16766 David Sherwood <david.sherwood@arm.com>
16768 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16769 Take a scalar_int_mode instead of a machine_mode.
16770 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16771 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16772 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16773 (aarch64_simd_attr_length_rglist): Delete.
16774 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16775 a scalar_int_mode instead of a machine_mode.
16776 (aarch64_add_offset): Likewise.
16777 (aarch64_internal_mov_immediate): Likewise
16778 (aarch64_add_constant_internal): Likewise.
16779 (aarch64_add_constant): Likewise.
16780 (aarch64_movw_imm): Likewise.
16781 (aarch64_rtx_arith_op_extract_p): Likewise.
16782 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16783 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16784 Remove assert that the mode isn't a vector.
16785 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16786 (aarch64_expand_mov_immediate): Update calls after above changes.
16787 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16788 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16789 (aarch64_move_imm): Likewise.
16790 (aarch64_can_const_movi_rtx_p): Likewise.
16791 (aarch64_strip_extend): Likewise.
16792 (aarch64_extr_rtx_p): Likewise.
16793 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16794 a CONST_INT when the mode parameter is VOIDmode.
16795 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16797 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16799 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
16800 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16801 (bitwise_type_for_mode): Update accordingly.
16803 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16805 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
16806 * stor-layout.c (mode_for_size_tree): Likewise.
16807 (mode_for_array): Update accordingly.
16808 (layout_decl): Likewise.
16809 (compute_record_mode): Likewise. Only set the mode once.
16811 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16813 * target.def (get_mask_mode): Change return type to opt_mode.
16815 * doc/tm.texi: Regenerate.
16816 * targhooks.h (default_get_mask_mode): Return an opt_mode.
16817 * targhooks.c (default_get_mask_mode): Likewise.
16818 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16819 * optabs-query.c (can_vec_mask_load_store_p): Update use of
16820 targetm.get_mask_mode.
16821 * tree.c (build_truth_vector_type): Likewise.
16823 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16825 * machmode.h (mode_for_vector): Return an opt_mode.
16826 * stor-layout.c (mode_for_vector): Likewise.
16827 (mode_for_int_vector): Update accordingly.
16828 (layout_type): Likewise.
16829 * config/i386/i386.c (emit_memmov): Likewise.
16830 (ix86_expand_set_or_movmem): Likewise.
16831 (ix86_expand_vector_init): Likewise.
16832 (ix86_get_mask_mode): Likewise.
16833 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16835 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16836 * expmed.c (extract_bit_field_1): Likewise.
16837 * expr.c (expand_expr_real_2): Likewise.
16838 * optabs-query.c (can_vec_perm_p): Likewise.
16839 (can_vec_mask_load_store_p): Likewise.
16840 * optabs.c (expand_vec_perm): Likewise.
16841 * targhooks.c (default_get_mask_mode): Likewise.
16842 * tree-vect-stmts.c (vectorizable_store): Likewise.
16843 (vectorizable_load): Likewise.
16844 (get_vectype_for_scalar_type_and_size): Likewise.
16846 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16848 * machmode.h (mode_for_int_vector): New function.
16849 * stor-layout.c (mode_for_int_vector): Likewise.
16850 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16851 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
16852 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16853 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16854 (s390_expand_vcond): Likewise.
16856 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16858 * machmode.h (opt_machine_mode): New type.
16859 (opt_mode<T>): Allow construction from anything that can be
16861 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16862 (mode_for_size): Return an opt_machine_mode.
16863 * stor-layout.c (mode_for_size): Likewise.
16864 (mode_for_size_tree): Update call accordingly.
16865 (bitwise_mode_for_mode): Likewise.
16866 (make_fract_type): Likewise.
16867 (make_accum_type): Likewise.
16868 * caller-save.c (replace_reg_with_saved_mem): Update call
16870 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16871 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16872 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16873 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16874 * expmed.c (extract_bit_field_1): Likewise.
16875 * reload.c (get_secondary_mem): Likewise.
16876 * varasm.c (assemble_integer): Likewise.
16877 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
16880 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16882 * machmode.h (decimal_float_mode_for_size): New function.
16883 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16884 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16885 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16886 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16887 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16888 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16890 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16892 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16893 (get_builtin_sync_mode): Likewise.
16894 (expand_ifn_atomic_compare_exchange): Likewise.
16895 (expand_builtin_atomic_clear): Likewise.
16896 (expand_builtin_atomic_test_and_set): Likewise.
16897 (fold_builtin_atomic_always_lock_free): Likewise.
16898 * calls.c (compute_argument_addresses): Likewise.
16899 (emit_library_call_value_1): Likewise.
16900 (store_one_arg): Likewise.
16901 * combine.c (combine_instructions): Likewise.
16902 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16903 * config/arm/arm.c (arm_function_value): Likewise.
16904 (aapcs_allocate_return_reg): Likewise.
16905 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16906 * config/i386/i386.c (construct_container): Likewise.
16907 (ix86_gimplify_va_arg): Likewise.
16908 (ix86_expand_sse_cmp): Likewise.
16909 (emit_memmov): Likewise.
16910 (emit_memset): Likewise.
16911 (expand_small_movmem_or_setmem): Likewise.
16912 (ix86_expand_pextr): Likewise.
16913 (ix86_expand_pinsr): Likewise.
16914 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16915 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16917 * config/mips/mips.c (mips_function_value_1) Likewise.
16918 (mips_block_move_straight): Likewise.
16919 (mips_expand_ins_as_unaligned_store): Likewise.
16920 * config/powerpcspe/powerpcspe.c
16921 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16922 (rs6000_darwin64_record_arg_flush): Likewise.
16923 * config/rs6000/rs6000.c
16924 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16925 (rs6000_darwin64_record_arg_flush): Likewise.
16926 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16927 (sparc_function_value_1): Likewise.
16928 * config/spu/spu.c (adjust_operand): Likewise.
16929 (spu_emit_branch_or_set): Likewise.
16930 (arith_immediate_p): Likewise.
16931 * emit-rtl.c (gen_lowpart_common): Likewise.
16932 * expr.c (expand_expr_real_1): Likewise.
16933 * function.c (assign_parm_setup_block): Likewise.
16934 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16935 * reload1.c (alter_reg): Likewise.
16936 * stor-layout.c (mode_for_vector): Likewise.
16937 (layout_type): Likewise.
16939 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16941 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16942 (spu_convert_move): Likewise.
16943 * lower-subreg.c (resolve_simple_move): Likewise.
16945 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16948 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16949 define_insn to a define_expand.
16950 (altivec_vsum2sws_direct): New define_insn.
16951 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16953 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16955 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16958 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16963 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16964 realign stack if stack alignment needed is less than incoming
16967 2017-09-05 Marek Polacek <polacek@redhat.com>
16970 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16973 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16975 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16977 2017-09-05 Richard Biener <rguenther@suse.de>
16979 PR tree-optimization/82084
16980 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16982 2017-09-05 Richard Biener <rguenther@suse.de>
16984 PR tree-optimization/82102
16985 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16987 2017-09-05 Martin Liska <mliska@suse.cz>
16989 PR tree-optimization/82032
16990 * tree-cfg.c (generate_range_test): New function.
16991 * tree-cfg.h (generate_range_test): Declared here.
16992 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16993 (cleanup_control_expr_graph): Use it.
16994 * tree-switch-conversion.c (try_switch_expansion): Remove
16996 (emit_case_nodes): Use generate_range_test.
16998 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17001 * config/i386/i386.md (*<btsc><mode>_mask): Add
17002 TARGET_USE_BT to insn constraint.
17003 (*btr<mode>_mask): Ditto.
17005 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
17007 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
17008 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17010 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17013 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17014 TARGET_NEON and TARGET_IWMMXT.
17015 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17016 TARGET_NEON and TARGET_IWMMXT.
17017 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17019 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17021 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17022 (ix86_rewrite_tls_address): Ditto.
17023 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17024 (ix86_rewrite_tls_address_1): Ditto.
17025 (ix86_rewrite_tls_address): Ditto.
17026 * config/i386/predicates.md (tls_address_pattern): New predicate.
17027 * config/i386/i386.md (TLS address splitter): New splitter.
17029 2017-09-04 Richard Biener <rguenther@suse.de>
17031 PR tree-optimization/82084
17032 * fold-const.h (can_native_encode_string_p): Declare.
17033 * fold-const.c (can_native_encode_string_p): Factor out from ...
17034 (native_encode_string): ... here.
17035 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17036 vectorizing stores from constants we later cannot handle.
17038 2017-09-04 Marek Polacek <polacek@redhat.com>
17041 * doc/invoke.texi: Update -Wtautological-compare documentation.
17043 2017-09-04 Jeff Law <law@redhat.com>
17045 PR tree-optimization/64910
17046 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17047 swap the first and last operand if the last is a constant.
17049 2017-09-04 Marek Polacek <polacek@redhat.com>
17052 * convert.c (do_narrow): When sanitizing signed integer overflows,
17053 bail out for signed types.
17054 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17056 2017-09-04 Richard Biener <rguenther@suse.de>
17058 PR tree-optimization/82060
17059 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17060 Move devirtualization after stmt folding and before EH/AB/noreturn
17061 cleanup to get the stmt refs canonicalized. Use a bool instead
17062 of gimple_modified_p since that doesn't work for NOPs. Schedule
17063 NOPs generated by folding for removal.
17065 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17066 Alan Hayward <alan.hayward@arm.com>
17067 David Sherwood <david.sherwood@arm.com>
17069 * coretypes.h (pad_direction): New enum.
17070 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17071 (FUNCTION_ARG_PADDING): Likewise.
17072 * target.def (function_arg_padding): New hook.
17073 * targhooks.h (default_function_arg_padding): Declare.
17074 * targhooks.c (default_function_arg_padding): New function.
17075 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17076 (TARGET_FUNCTION_ARG_PADDING): ...this.
17077 * doc/tm.texi: Regenerate.
17078 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17079 instead of direction.
17080 (compute_argument_addresses): Likewise.
17081 (load_register_parameters): Likewise.
17082 (emit_library_call_value_1): Likewise.
17083 (store_one_arg): Use targetm.calls.function_arg_padding instead
17084 of FUNCTION_ARG_PADDING.
17085 (must_pass_in_stack_var_size_or_pad): Likewise.
17086 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17087 (emit_group_store): Likewise.
17088 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17089 instead of FUNCTION_ARG_PADDING.
17090 (emit_push_insn): Likewise, and propagate enum change throughout
17092 * function.h (direction): Delete.
17093 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17095 * function.c (assign_parm_find_stack_rtl): Likewise.
17096 (assign_parm_setup_block_p): Likewise.
17097 (assign_parm_setup_block): Likewise.
17098 (gimplify_parameters): Likewise.
17099 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17100 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17102 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17103 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17104 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17105 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17106 (aarch64_function_arg_padding): ...this new function.
17107 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17108 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17109 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17110 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17111 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17112 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17113 (arm_pad_arg_upward): Replace with...
17114 (arm_function_arg_padding): ...this new function.
17115 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17117 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17118 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17119 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17120 (ia64_hpux_function_arg_padding): Replace with...
17121 (ia64_function_arg_padding): ...this new function. Use pad_direction
17122 instead of direction. Check for TARGET_HPUX.
17123 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17124 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17125 (iq2000_function_arg_padding): New function.
17126 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17127 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17128 (mips_function_arg_padding): ...this new function.
17129 (mips_pad_reg_upward): Update accordingly.
17130 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17131 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17132 targetm.calls.function_arg_padding.
17133 (FUNCTION_ARG_PADDING): Delete.
17134 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17135 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17136 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17137 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17138 (nios2_block_reg_padding): Return pad_direction instead of direction.
17139 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17140 instead of direction.
17141 (nios2_function_arg_padding): Likewise. Make static.
17142 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17143 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17144 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17145 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17146 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17147 (pa_function_arg_padding): Make static. Return pad_direction instead
17149 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17150 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17151 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17152 instead of direction. Use targetm.calls.function_arg_padding.
17153 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17154 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17155 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17156 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17157 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17159 (function_arg_padding): Rename to...
17160 (rs6000_function_arg_padding): ...this. Make static. Return
17161 pad_direction instead of direction.
17162 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17163 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17164 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17165 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17166 instead of direction. Use targetm.calls.function_arg_padding.
17167 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17168 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17169 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17170 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17171 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17172 (function_arg_padding): Rename to...
17173 (rs6000_function_arg_padding): ...this. Make static. Return
17174 pad_direction instead of direction.
17175 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17176 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17177 * config/s390/s390.c (s390_function_arg_padding): New function.
17178 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17179 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17180 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17181 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17182 (function_arg_padding): Rename to...
17183 (sparc_function_arg_padding): ...this. Make static. Return
17184 pad_direction instead of direction.
17185 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17186 * config/spu/spu.c (spu_function_arg_padding): New function.
17187 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17188 * system.h (FUNCTION_ARG_PADDING): Poison.
17190 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17191 Alan Hayward <alan.hayward@arm.com>
17192 David Sherwood <david.sherwood@arm.com>
17194 * target.def (modes_tieable_p): New hook.
17195 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17196 (TARGET_MODES_TIEABLE_P): ...this.
17197 * doc/tm.texi.in: Regenerate.
17198 * hooks.h (hook_bool_mode_mode_true): Declare.
17199 * hooks.c (hook_bool_mode_mode_true): New function.
17200 * combine.c (subst): Use targetm.modes_tieable_p instead of
17202 * dse.c (find_shift_sequence): Likewise.
17203 * expmed.c (extract_low_bits): Likewise.
17204 * lower-subreg.c: Include target.h.
17205 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17207 * rtlanal.c (rtx_cost): Likewise.
17208 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17209 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17210 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17211 (TARGET_MODES_TIEABLE_P): Redefine.
17212 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17213 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17214 (TARGET_MODES_TIEABLE_P): Redefine.
17215 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17216 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17217 (arc_modes_tieable_p): New function.
17218 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17219 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17220 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17221 (arm_modes_tieable_p): Make static.
17222 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17223 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17224 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17225 (TARGET_MODES_TIEABLE_P): Redefine.
17226 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17227 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17228 (TARGET_MODES_TIEABLE_P): Redefine.
17229 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17230 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17231 (cr16_modes_tieable_p): New function.
17232 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17233 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17234 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17235 (TRULY_NOOP_TRUNCATION): Update comment.
17236 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17237 (TRULY_NOOP_TRUNCATION): Update comment.
17238 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17239 (frv_modes_tieable_p): New function.
17240 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17241 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17242 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17243 (TARGET_MODES_TIEABLE_P): Redefine.
17244 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17245 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17246 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17247 (TARGET_MODES_TIEABLE_P): Redefine.
17248 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17249 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17250 (ia64_modes_tieable_p): New function.
17251 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17252 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17253 (iq2000_modes_tieable_p): New function.
17254 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17255 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17256 (lm32_modes_tieable_p): New function.
17257 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17258 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17259 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17260 (TARGET_MODES_TIEABLE_P): Redefine.
17261 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17262 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17263 (m32r_modes_tieable_p): New function.
17264 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17265 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17266 (m68k_modes_tieable_p): New function.
17267 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17268 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17269 (mcore_modes_tieable_p): New function.
17270 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17271 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17273 (TARGET_MODES_TIEABLE_P): Redefine.
17274 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17275 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17276 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17277 (TARGET_MODES_TIEABLE_P): Redefine.
17278 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17279 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17280 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17281 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17282 (mn10300_modes_tieable_p): ...this and make static.
17283 (TARGET_MODES_TIEABLE_P): Redefine.
17284 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17285 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17286 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17287 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17288 (msp430_modes_tieable_p): Make static.
17289 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17290 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17291 (TARGET_MODES_TIEABLE_P): Redefine.
17292 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17293 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17294 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17295 (TARGET_MODES_TIEABLE_P): Redefine.
17296 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17297 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17298 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17299 (TARGET_MODES_TIEABLE_P): Redefine.
17300 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17301 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17302 (pdp11_modes_tieable_p): New function.
17303 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17304 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17305 (rs6000_modes_tieable_p): New function.
17306 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17307 * config/powerpcspe/powerpcspe.md: Update comment.
17308 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17309 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17310 (TARGET_MODES_TIEABLE_P): Redefine.
17311 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17312 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17313 (rl78_modes_tieable_p): New function.
17314 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17315 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17316 (rs6000_modes_tieable_p): New function.
17317 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17318 * config/rs6000/rs6000.md: Update comment.
17319 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17320 * config/rx/rx.c (rx_modes_tieable_p): New function.
17321 (TARGET_MODES_TIEABLE_P): Redefine.
17322 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17323 * config/s390/s390.c (s390_modes_tieable_p): New function.
17324 (TARGET_MODES_TIEABLE_P): Redefine.
17325 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17326 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17327 (sh_modes_tieable_p): New function.
17328 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17329 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17330 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17331 (sparc_modes_tieable_p): Make static.
17332 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17333 * config/spu/spu.c (spu_modes_tieable_p): New function.
17334 (TARGET_MODES_TIEABLE_P): Redefine.
17335 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17336 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17337 (TARGET_MODES_TIEABLE_P): Redefine.
17338 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17339 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17340 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17341 * config/v850/v850.c (v850_modes_tieable_p): New function.
17342 (TARGET_MODES_TIEABLE_P): Redefine.
17343 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17344 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17345 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17346 (visium_modes_tieable_p): New function.
17347 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17348 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17349 (xtensa_modes_tieable_p): New function.
17350 * system.h (MODES_TIEABLE_P): Poison.
17352 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17353 Alan Hayward <alan.hayward@arm.com>
17354 David Sherwood <david.sherwood@arm.com>
17356 * target.def (hard_regno_mode_ok): New hook.
17357 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17358 (TARGET_HARD_REGNO_MODE_OK): ...this.
17359 * doc/tm.texi.in: Regenerate.
17360 * hooks.h (hook_bool_uint_mode_true): Declare.
17361 * hooks.c (hook_bool_uint_mode_true): New function.
17362 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17363 HARD_REGNO_MODE_OK.
17364 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17365 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17366 instead of HARD_REGNO_MODE_OK.
17367 * caller-save.c: Include target.h.
17368 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17369 HARD_REGNO_MODE_OK.
17370 * combine.c (can_combine_p): Likewise.
17371 (combinable_i3pat): Likewise.
17372 (can_change_dest_mode): Likewise.
17373 * expr.c (init_expr_target): Likewise.
17374 (convert_move): Likewise.
17375 (convert_modes): Likewise.
17376 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17377 (setup_prohibited_mode_move_regs): Likewise.
17378 * ira.h (target_ira): Likewise.
17379 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17380 * lra-constraints.c (process_alt_operands): Likewise.
17381 (split_reg): Likewise.
17382 * recog.c (peep2_find_free_register): Likewise.
17383 * ree.c (combine_reaching_defs): Likewise.
17384 * regcprop.c (maybe_mode_change): Likewise.
17385 * reginfo.c (init_reg_sets_1): Likewise.
17386 (choose_hard_reg_mode): Likewise.
17387 (simplifiable_subregs): Likewise.
17388 * regrename.c (check_new_reg_p): Likewise.
17389 * reload.c (find_valid_class): Likewise.
17390 (find_valid_class_1): Likewise.
17391 (reload_inner_reg_of_subreg): Likewise.
17392 (push_reload): Likewise.
17393 (combine_reloads): Likewise.
17394 (find_dummy_reload): Likewise.
17395 (find_reloads): Likewise.
17396 * reload1.c (find_reg): Likewise.
17397 (set_reload_reg): Likewise.
17398 (allocate_reload_reg): Likewise.
17399 (choose_reload_regs): Likewise.
17400 (reload_adjust_reg_for_temp): Likewise.
17401 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17402 (simplify_subreg_regno): Likewise.
17403 * sel-sched.c (init_regs_for_mode): Likewise.
17404 * varasm.c (make_decl_rtl): Likewise.
17405 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17406 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17407 HARD_REGNO_MODE_OK.
17408 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17409 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17410 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17411 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17412 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17413 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17414 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17415 (arc_mode_class): Delete.
17416 (HARD_REGNO_MODE_OK): Delete.
17417 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17418 (arc_hard_regno_mode_ok): Rename old array to...
17419 (arc_hard_regno_mode_ok_modes): ...this.
17420 (arc_conditional_register_usage): Update accordingly.
17421 (arc_mode_class): Make static.
17422 (arc_hard_regno_mode_ok): New function.
17423 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17424 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17425 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17426 (arm_hard_regno_mode_ok): Make static.
17427 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17428 HARD_REGNO_MODE_OK.
17429 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17430 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17431 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17433 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17434 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17435 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17436 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17437 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17438 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17439 * config/bfin/predicates.md (valid_reg_operand): Use
17440 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17441 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17442 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17443 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17444 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17445 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17446 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17447 (cr16_hard_regno_mode_ok): Make static and return a bool.
17448 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17449 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17450 (cris_hard_regno_mode_ok): New function.
17451 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17452 (epiphany_mode_class): Delete.
17453 (HARD_REGNO_MODE_OK): Delete.
17454 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17455 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17456 (hard_regno_mode_ok): Rename to...
17457 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17458 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17459 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17460 HARD_REGNO_MODE_OK.
17461 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17462 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17463 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17464 (frv_hard_regno_mode_ok): Make static and return a bool.
17465 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17466 HARD_REGNO_MODE_OK.
17467 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17468 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17469 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17470 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17472 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17473 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17474 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17475 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17477 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17478 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17479 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17480 (ia64_hard_regno_mode_ok): New function.
17481 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17482 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17483 (iq2000_hard_regno_mode_ok): New function.
17484 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17485 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17486 (lm32_hard_regno_mode_ok): New function.
17487 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17488 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17489 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17490 instead of HARD_REGNO_MODE_OK.
17491 (m32c_hard_regno_ok): Rename to...
17492 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17493 (m32c_cannot_change_mode_class): Update accordingly.
17494 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17495 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17496 (m32r_mode_class): Delete.
17497 (HARD_REGNO_MODE_OK): Delete.
17498 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17499 (m32r_hard_regno_mode_ok): Rename to...
17500 (m32r_hard_regno_modes): ...this.
17501 (m32r_mode_class): Make static.
17502 (m32r_hard_regno_mode_ok): New function.
17503 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17504 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17505 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17506 (m68k_hard_regno_mode_ok): Make static.
17507 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17508 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17509 (mcore_hard_regno_mode_ok): New function.
17510 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17511 (HARD_REGNO_MODE_OK): Delete.
17512 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17514 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17515 (microblaze_hard_regno_mode_ok): New function.
17516 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17517 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17518 (mips_hard_regno_mode_ok): Delete.
17519 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17520 (mips_hard_regno_mode_ok_p): ...this and make static.
17521 (mips_hard_regno_mode_ok_p): Rename to...
17522 (mips_hard_regno_mode_ok_uncached): ...this.
17523 (mips_hard_regno_mode_ok): New function.
17524 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17525 of HARD_REGNO_MODE_OK.
17526 (mips_option_override): Update after above name changes.
17527 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17528 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17529 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17530 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17531 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17532 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17533 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17534 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17535 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17536 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17537 (msp430_hard_regno_mode_ok): Make static and return a bool.
17538 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17539 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17540 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17542 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17543 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17544 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17545 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17546 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17547 (PA_HARD_REGNO_MODE_OK): ...this
17548 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17549 (PA_HARD_REGNO_MODE_OK): ...this.
17550 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17551 (pa_hard_regno_mode_ok): New function.
17552 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17553 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17554 (pdp11_hard_regno_mode_ok): New function.
17555 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17556 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17558 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17560 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17561 (rs6000_hard_regno_mode_ok): Rename to...
17562 (rs6000_hard_regno_mode_ok_uncached): ...this.
17563 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17564 (rs6000_hard_regno_mode_ok): New function.
17565 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17566 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17567 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17568 (riscv_hard_regno_mode_ok): ...this and make static.
17569 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17570 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17571 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17572 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17573 (rl78_hard_regno_mode_ok): Make static and return bool.
17574 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17575 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17577 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17578 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17579 (rs6000_hard_regno_mode_ok): Rename to...
17580 (rs6000_hard_regno_mode_ok_uncached): ...this.
17581 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17582 (rs6000_hard_regno_mode_ok): New function.
17583 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17584 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17585 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17586 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17587 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17588 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17589 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17590 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17591 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17592 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17593 (sh_hard_regno_mode_ok): Make static.
17594 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17595 instead of HARD_REGNO_MODE_OK.
17596 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17597 (sparc_mode_class): Delete.
17598 (HARD_REGNO_MODE_OK): Delete.
17599 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17600 (hard_regno_mode_classes): Make static.
17601 (sparc_mode_class): Likewise.
17602 (sparc_hard_regno_mode_ok): New function.
17603 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17604 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17605 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17607 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17608 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17609 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17610 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17611 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17612 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17613 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17614 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17615 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17616 (visium_hard_regno_mode_ok): New function.
17617 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17618 instead of HARD_REGNO_MODE_OK.
17619 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17620 (HARD_REGNO_MODE_OK): Delete.
17621 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17622 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17623 (xtensa_option_override): Update accordingly.
17624 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17625 (xtensa_hard_regno_mode_ok): New function.
17626 * system.h (HARD_REGNO_MODE_OK): Poison.
17628 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17629 Alan Hayward <alan.hayward@arm.com>
17630 David Sherwood <david.sherwood@arm.com>
17632 * target.def (hard_regno_call_part_clobbered): New hook.
17633 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17634 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17635 * doc/tm.texi: Regenerate.
17636 * hooks.h (hook_bool_uint_mode_false): Declare.
17637 * hooks.c (hook_bool_uint_mode_false): New function.
17638 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17639 * cselib.c (cselib_process_insn): Use
17640 targetm.hard_regno_call_part_clobbered instead of
17641 HARD_REGNO_CALL_PART_CLOBBERED.
17642 * ira-conflicts.c (ira_build_conflicts): Likewise.
17643 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17644 * lra-constraints.c (need_for_call_save_p): Likewise.
17645 * lra-lives.c: Include target.h.
17646 (check_pseudos_live_through_calls): Use
17647 targetm.hard_regno_call_part_clobbered instead of
17648 HARD_REGNO_CALL_PART_CLOBBERED.
17649 * regcprop.c: Include target.h.
17650 (copyprop_hardreg_forward_1): Use
17651 targetm.hard_regno_call_part_clobbered instead of
17652 HARD_REGNO_CALL_PART_CLOBBERED.
17653 * reginfo.c (choose_hard_reg_mode): Likewise.
17654 * regrename.c (check_new_reg_p): Likewise.
17655 * reload.c (find_equiv_reg): Likewise.
17656 * reload1.c (emit_reload_insns): Likewise.
17657 * sched-deps.c (deps_analyze_insn): Likewise.
17658 * sel-sched.c (init_regs_for_mode): Likewise.
17659 (mark_unavailable_hard_regs): Likewise.
17660 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17661 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17662 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17664 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17665 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17666 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17668 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17670 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17671 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17672 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17674 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17675 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17676 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17678 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17679 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17681 * config/powerpcspe/powerpcspe.c
17682 (rs6000_hard_regno_call_part_clobbered): New function.
17683 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17684 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17685 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17687 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17688 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17689 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17691 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17692 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17693 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17695 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17696 Alan Hayward <alan.hayward@arm.com>
17697 David Sherwood <david.sherwood@arm.com>
17699 * rtl.h (subreg_memory_offset): Declare.
17700 * emit-rtl.c (subreg_memory_offset): New function.
17701 * expmed.c (store_bit_field_1): Use it.
17702 * expr.c (undefined_operand_subword_p): Likewise.
17703 * simplify-rtx.c (simplify_subreg): Likewise.
17705 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17707 PR rtl-optimization/57448
17710 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17711 using atomic_load pattern.
17712 (expand_atomic_store): Likewise.
17714 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17717 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17718 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17719 handling. Use replace_call_with_value with NULL instead of
17720 gsi_replace, unlink_stmt_vdef and release_defs.
17722 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17725 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17727 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17730 * rtl.h (emit_library_call_value_1): Declare.
17731 (emit_library_call): Replace declaration with a series of overloads.
17732 Remove the parameter count argument.
17733 (emit_library_call_value): Likewise.
17734 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17735 with an "rtx_mode_t *".
17736 (emit_library_call_value): Delete.
17737 (emit_library_call): Likewise.
17738 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17739 (asan_emit_allocas_unpoison): Likewise.
17740 * builtins.c (expand_builtin_powi): Likewise.
17741 (expand_asan_emit_allocas_unpoison): Likewise.
17742 * cfgexpand.c (expand_main_function): Likewise.
17743 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17744 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17745 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17746 * config/arm/arm.c (arm_trampoline_init): Likewise.
17747 (arm_call_tls_get_addr): Likewise.
17748 (arm_expand_divmod_libfunc): Likewise.
17749 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17750 (smulsi3_highpart): Likewise.
17751 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17752 (c6x_expand_compare): Likewise.
17753 (c6x_expand_movmem): Likewise.
17754 * config/frv/frv.c (frv_trampoline_init): Likewise.
17755 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17756 (ix86_expand_divmod_libfunc): Likewise.
17757 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17758 (ia64_expand_compare): Likewise.
17759 (ia64_profile_hook): Likewise.
17760 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17761 (nonlocal_goto): Likewise.
17762 (restore_stack_nonlocal): Likewise.
17763 * config/m32r/m32r.c (block_move_call): Likewise.
17764 (m32r_trampoline_init): Likewise.
17765 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17766 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17767 (m68k_call_m68k_read_tp): Likewise.
17768 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17769 (microblaze_expand_divide): Likewise.
17770 * config/mips/mips.h (mips_args): Likewise.
17771 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17772 (MIPS_ICACHE_SYNC): Likewise.
17773 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17774 (nios2_trampoline_init): Likewise.
17775 * config/pa/pa.c (hppa_tls_call): Likewise.
17776 (pa_trampoline_init): Likewise.
17777 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17778 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17779 (expand_strn_compare): Likewise.
17780 (rs6000_generate_compare): Likewise.
17781 (rs6000_expand_float128_convert): Likewise.
17782 (output_profile_hook): Likewise.
17783 (rs6000_trampoline_init): Likewise.
17784 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17785 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17786 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17787 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17788 (rs6000_generate_compare): Likewise.
17789 (rs6000_expand_float128_convert): Likewise.
17790 (output_profile_hook): Likewise.
17791 (rs6000_trampoline_init): Likewise.
17792 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17793 * config/sh/sh.c (sh_trampoline_init): Likewise.
17794 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17795 (sparc_emit_float_lib_cmp): Likewise.
17796 (sparc32_initialize_trampoline): Likewise.
17797 (sparc64_initialize_trampoline): Likewise.
17798 (sparc_profile_hook): Likewise.
17799 * config/spu/spu.c (ea_load_store): Likewise.
17800 * config/spu/spu.md (floatunssidf2): Likewise.
17801 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
17802 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
17803 * config/visium/visium.c (expand_block_move_4): Likewise.
17804 (expand_block_move_2): Likewise.
17805 (expand_block_move_1): Likewise.
17806 (expand_block_set_4): Likewise.
17807 (expand_block_set_2): Likewise.
17808 (expand_block_set_1): Likewise.
17809 (visium_trampoline_init): Likewise.
17810 (visium_profile_hook): Likewise.
17811 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
17812 (xtensa_setup_frame_addresses): Likewise.
17813 (xtensa_trampoline_init): Likewise.
17814 * except.c (sjlj_emit_function_enter): Likewise.
17815 (sjlj_emit_function_exit): Likewise.
17816 * explow.c (allocate_dynamic_stack_space): Likewise.
17817 (probe_stack_range): Likewise.
17818 * expr.c (convert_mode_scalar): Likewise.
17819 * optabs.c (expand_binop): Likewise.
17820 (expand_twoval_binop_libfunc): Likewise.
17821 (expand_unop): Likewise.
17822 (prepare_cmp_insn): Likewise.
17823 (prepare_float_lib_cmp): Likewise.
17824 (expand_float): Likewise.
17825 (expand_fix): Likewise.
17826 (expand_fixed_convert): Likewise.
17827 (maybe_emit_sync_lock_test_and_set): Likewise.
17828 (expand_atomic_compare_and_swap): Likewise.
17829 (expand_mem_thread_fence): Likewise.
17830 (expand_atomic_fetch_op): Likewise.
17832 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17834 * doc/generic.texi (OpenACC): Adjust URL.
17835 * doc/invoke.texi (C Dialect Options): Ditto.
17837 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17839 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17840 predicate for operand 1. Add (m,<S>) constraint.
17841 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17842 Prevent memory operand 1 with register operand 2.
17844 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17846 PR rtl-optimization/82024
17847 * combine.c (try_combine): If the combination result is a PARALLEL,
17848 and we only need to retain the SET in there that would be placed
17849 at I2, check that we can place that at I3 instead, before doing so.
17851 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17854 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
17856 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
17857 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17860 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17861 Jeff Law <law@redhat.com>
17863 * varasm.c (bss_initializer_p): Do not put constants into .bss
17864 (categorize_decl_for_section): Handle bss_initializer_p returning
17865 false when DECL_INITIAL is NULL.
17867 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17870 * config/s390/s390.c (s390_can_inline_p): New function.
17872 2017-09-01 Jeff Law <law@redhat.com>
17874 PR tree-optimization/82052
17875 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17876 Always initialize the returned slot after a hash table miss
17877 when INSERT is true.
17879 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
17881 * config/s390/s390.md (mem_signal_fence): Remove.
17882 * doc/md.texi (mem_signal_fence): Remove.
17883 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17885 * target-insns.def (mem_signal_fence): Remove.
17887 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17890 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17893 * asan.c (create_odr_indicator): Strip name encoding from assembler
17894 name before appending it after __odr_asan_.
17896 2017-09-01 Martin Liska <mliska@suse.cz>
17898 PR tree-optimization/82059
17899 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17900 frequency only when an edge is redirected.
17902 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17904 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17905 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17906 (arc_conditional_register_usage): Remove ARC600 lp_count
17908 (arc_file_start): Emit Tag_ARC_CPU_variation.
17909 (arc_can_use_doloop_p): New conditions to use ZOLs.
17910 (hwloop_fail): New function.
17911 (hwloop_optimize): Likewise.
17912 (hwloop_pattern_reg): Likewise.
17913 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17914 (arc_reorg_loops): New function, calls reorg_loops.
17915 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17916 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17918 (arc_loop_hazard): Remove function, functionality moved into
17920 (arc_hazard): Remove arc_loop_hazard call.
17921 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17922 into hwloop_optimize.
17923 (arc_label_align): Remove ZOL handling.
17924 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17925 * config/arc/arc.md (doloop_begin): Remove pattern.
17926 (doloop_begin_i): Likewise.
17927 (doloop_end_i): Likewise.
17928 (doloop_fallback): Likewise.
17929 (doloop_fallback_m): Likewise.
17930 (doloop_end): Reimplement expand.
17931 (arc_lp): New pattern for LP instruction.
17932 (loop_end): New pattern.
17933 (loop_fail): Likewise.
17934 (decrement_and_branch_until_zero): Likewise.
17935 * config/arc/arc.opt (mlpc-width): New option.
17936 * doc/invoke.texi (mlpc-width): Document option.
17938 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17940 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17941 (arc_ccfsm_advance): Fix checking for delay slots.
17942 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17944 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17946 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17948 (movhi_insn): Update store instruction constraint which are saving
17949 6-bit short immediates.
17950 (movsi_insn): Consider also short scaled load operations.
17951 (zero_extendhisi2_i): Use Usd constraint instead of T.
17952 (extendhisi2_i): Add q constraint.
17953 (arc_clzsi2): Add type and length attributes.
17954 (arc_ctzsi2): Likewise.
17955 * config/arc/constraints.md (Usc): Update constraint, the
17956 assembler can parse two relocations for a single instruction.
17958 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17960 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17961 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17963 2017-08-31 Olivier Hainque <hainque@adacore.com>
17965 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17966 match as powerpc-wrs-vxworks*.
17968 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17970 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17971 register constraint for by-element operand.
17972 (aarch64_mls_elt_merge<mode>): Likewise.
17974 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17976 * config/arc/arc.c (arc_can_follow_jump): Check for short
17979 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17981 * config.gcc: Use g.opt for arc.
17982 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17983 functionality moved to ...
17984 (legitimate_scaled_address_p): New function, ...here.
17985 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17986 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17987 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17989 (arc_override_options): Handle G option.
17990 (arc_output_pic_addr_const): Correct function definition.
17991 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17992 (arc_decl_anon_ns_mem_p): Delete.
17993 (arc_in_small_data_p): Overhaul this function to take into
17994 consideration the value given via G option.
17995 (arc_rewrite_small_data_1): Renamed and corrected old
17996 arc_rewrite_small_data function.
17997 (arc_rewrite_small_data): New function.
17998 (small_data_pattern): Don't use pic_offset_table_rtx.
17999 * config/arc/arc.h (CC1_SPEC): Recognize G option.
18000 * config/arc/simdext.md (movmisalignv2hi): Use
18001 prepare_move_operands function.
18003 (movmisalign*): Likewise.
18004 * doc/invoke.texi (ARC options): Document -G option.
18006 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18008 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18010 * config/arc/arc.c (arc_print_operand): Output scalled address for
18011 sdata whenever is possible.
18012 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18013 load/stores are available.
18014 (compact_sda_memory_operand): Check for the alignment required by
18015 code density instructions.
18016 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18018 * config/arc/constraints.md (Usd): Update constraint.
18019 (Us0): New constraint.
18020 (Usc): Update constraint.
18022 2017-08-31 Richard Biener <rguenther@suse.de>
18024 PR middle-end/82054
18025 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18026 function only once.
18028 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18030 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18031 Resize type_signature.
18033 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18034 Alan Hayward <alan.hayward@arm.com>
18035 David Sherwood <david.sherwood@arm.com>
18037 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18038 subregs whose inner modes can be stored in GPRs.
18039 (aarch64_classify_index): Likewise.
18041 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18042 Alan Hayward <alan.hayward@arm.com>
18043 David Sherwood <david.sherwood@arm.com>
18045 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18046 (V_INT_EQUIV): ...this.
18047 (v_cmp_result): Rename to...
18048 (v_int_equiv): ...this.
18049 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18050 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18051 (copysign<mode>3): Likewise.
18052 (aarch64_simd_bsl<mode>_internal): Likewise.
18053 (aarch64_simd_bsl<mode>): Likewise.
18054 (vec_cmp<mode><mode>): Likewise.
18055 (vcond<mode><mode>): Likewise.
18056 (vcond<v_cmp_mixed><mode>): Likewise.
18057 (vcondu<mode><v_cmp_mixed>): Likewise.
18058 (aarch64_cm<optab><mode>): Likewise.
18059 (aarch64_cmtst<mode>): Likewise.
18060 (aarch64_fac<optab><mode>): Likewise.
18061 (vec_perm_const<mode>): Likewise.
18062 (vcond_mask_<mode><v_cmp_result>): Rename to...
18063 (vcond_mask_<mode><v_int_equiv>): ...this.
18064 (vec_cmp<mode><v_cmp_result>): Rename to...
18065 (vec_cmp<mode><v_int_equiv>): ...this.
18067 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18068 Alan Hayward <alan.hayward@arm.com>
18069 David Sherwood <david.sherwood@arm.com>
18071 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18073 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18074 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18075 (UNSPEC_LD4_DREG): New unspecs.
18076 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18077 (aarch64_ld2<mode>_dreg_be): Replace with...
18078 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18080 (aarch64_ld3<mode>_dreg_le)
18081 (aarch64_ld3<mode>_dreg_be): Replace with...
18082 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18084 (aarch64_ld4<mode>_dreg_le)
18085 (aarch64_ld4<mode>_dreg_be): Replace with...
18086 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18089 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18091 PR tree-optimization/81987
18092 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18093 insert an initializer in a location not dominated by the stride
18096 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18098 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18099 on the entire header of the finally block in the fallthru case.
18101 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18103 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18105 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18107 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18108 rs6000_emit_move_from_cr and call renamed function.
18109 (rs6000_emit_prologue): Call renamed functions.
18110 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18111 movesi_from_cr, remove volatile CRs.
18113 2017-08-30 Jon Beniston <jon@beniston.com>
18114 Richard Biener <rguenther@suse.de>
18116 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18117 of VECTOR_MODE_P check.
18118 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18119 element vector types.
18121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18123 * df.h (df_read_modify_subreg_p): Remove in favor of...
18124 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18125 const_rtx instead of an rtx.
18126 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18127 * df-problems.c (df_word_lr_mark_ref): Likewise.
18128 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18129 (mark_pseudo_reg_dead): Likewise.
18130 (mark_ref_dead): Likewise.
18131 * reginfo.c (init_subregs_of_mode): Likewise.
18132 * sched-deps.c (sched_analyze_1): Likewise.
18133 * df-scan.c (df_def_record_1): Likewise.
18134 (df_uses_record): Likewise.
18135 (df_read_modify_subreg_p): Remove in favor of...
18136 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18137 const_rtx instead of an rtx.
18139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18140 Alan Hayward <alan.hayward@arm.com>
18141 David Sherwood <david.sherwood@arm.com>
18143 * rtl.h (partial_subreg_p): New function.
18144 * caller-save.c (save_call_clobbered_regs): Use it.
18145 * calls.c (expand_call): Likewise.
18146 * combine.c (combinable_i3pat): Likewise.
18147 (simplify_set): Likewise.
18148 (make_extraction): Likewise.
18149 (make_compound_operation_int): Likewise.
18150 (gen_lowpart_or_truncate): Likewise.
18151 (force_to_mode): Likewise.
18152 (make_field_assignment): Likewise.
18153 (reg_truncated_to_mode): Likewise.
18154 (record_truncated_value): Likewise.
18155 (move_deaths): Likewise.
18156 * cse.c (record_jump_cond): Likewise.
18157 (cse_insn): Likewise.
18158 * cselib.c (cselib_lookup_1): Likewise.
18159 * expmed.c (extract_bit_field_using_extv): Likewise.
18160 * function.c (assign_parm_setup_reg): Likewise.
18161 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18162 * ira-build.c (create_insn_allocnos): Likewise.
18163 * lra-coalesce.c (merge_pseudos): Likewise.
18164 * lra-constraints.c (match_reload): Likewise.
18165 (simplify_operand_subreg): Likewise.
18166 (curr_insn_transform): Likewise.
18167 * lra-lives.c (process_bb_lives): Likewise.
18168 * lra.c (new_insn_reg): Likewise.
18169 (lra_substitute_pseudo): Likewise.
18170 * regcprop.c (mode_change_ok): Likewise.
18171 (maybe_mode_change): Likewise.
18172 (copyprop_hardreg_forward_1): Likewise.
18173 * reload.c (push_reload): Likewise.
18174 (find_reloads): Likewise.
18175 (find_reloads_subreg_address): Likewise.
18176 * reload1.c (alter_reg): Likewise.
18177 (eliminate_regs_1): Likewise.
18178 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18180 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18182 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18183 back to if statements, including unpack.
18185 2017-08-30 Martin Liska <mliska@suse.cz>
18187 PR inline-asm/82001
18188 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18190 (func_checker::compare_asm_inputs_outputs): ... this function.
18191 (func_checker::compare_gimple_asm): Use the function to compare
18192 also ASM constrains.
18193 * ipa-icf-gimple.h: Rename the function.
18195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18196 Alan Hayward <alan.hayward@arm.com>
18197 David Sherwood <david.sherwood@arm.com>
18199 * coretypes.h (complex_mode): New type.
18200 * gdbhooks.py (build_pretty_printer): Handle it.
18201 * machmode.h (complex_mode): New class.
18202 (complex_mode::includes_p): New function.
18203 (is_complex_int_mode): Likewise.
18204 (is_complex_float_mode): Likewise.
18205 * genmodes.c (get_mode_class): Handle complex mode classes.
18206 * function.c (expand_function_end): Use is_complex_int_mode.
18208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18209 Alan Hayward <alan.hayward@arm.com>
18210 David Sherwood <david.sherwood@arm.com>
18212 * coretypes.h (scalar_mode_pod): New typedef.
18213 * gdbhooks.py (build_pretty_printer): Handle it.
18214 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18215 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18216 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18217 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18218 as_a <scalar_mode>.
18220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18221 Alan Hayward <alan.hayward@arm.com>
18222 David Sherwood <david.sherwood@arm.com>
18224 * machmode.h (mode_for_vector): Take a scalar_mode instead
18226 * stor-layout.c (mode_for_vector): Likewise.
18227 * explow.c (promote_mode): Use as_a <scalar_mode>.
18228 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18231 Alan Hayward <alan.hayward@arm.com>
18232 David Sherwood <david.sherwood@arm.com>
18234 * target.def (preferred_simd_mode): Take a scalar_mode
18235 instead of a machine_mode.
18236 * targhooks.h (default_preferred_simd_mode): Likewise.
18237 * targhooks.c (default_preferred_simd_mode): Likewise.
18238 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18239 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18240 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18241 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18242 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18243 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18244 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18245 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18247 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18248 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18249 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18250 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18251 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18252 * doc/tm.texi: Regenerate.
18253 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18257 Alan Hayward <alan.hayward@arm.com>
18258 David Sherwood <david.sherwood@arm.com>
18260 * target.def (scalar_mode_supported_p): Take a scalar_mode
18261 instead of a machine_mode.
18262 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18263 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18264 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18265 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18266 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18267 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18268 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18269 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18270 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18271 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18272 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18273 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18274 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18275 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18277 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18278 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18279 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18280 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18281 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18283 * doc/tm.texi: Regenerate.
18285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18286 Alan Hayward <alan.hayward@arm.com>
18287 David Sherwood <david.sherwood@arm.com>
18289 * coretypes.h (opt_scalar_mode): New typedef.
18290 * gdbhooks.py (build_pretty_printers): Handle it.
18291 * machmode.h (mode_iterator::get_2xwider): Add overload for
18293 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18295 * expr.c (convert_mode_scalar): Likewise.
18296 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18297 * optabs.c (expand_float): Likewise.
18298 (expand_fix): Likewise.
18299 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18302 Alan Hayward <alan.hayward@arm.com>
18303 David Sherwood <david.sherwood@arm.com>
18305 * optabs.c (expand_float): Explicitly check for scalars before
18306 using a branching expansion.
18307 (expand_fix): Likewise.
18309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18310 Alan Hayward <alan.hayward@arm.com>
18311 David Sherwood <david.sherwood@arm.com>
18313 * expr.c (convert_mode): Split scalar handling out into...
18314 (convert_mode_scalar): ...this new function. Treat the modes
18317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18318 Alan Hayward <alan.hayward@arm.com>
18319 David Sherwood <david.sherwood@arm.com>
18321 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18323 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18326 Alan Hayward <alan.hayward@arm.com>
18327 David Sherwood <david.sherwood@arm.com>
18329 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18330 rather than a machine_mode.
18331 (fixed_from_string): Likewise.
18332 (fixed_convert): Likewise.
18333 (fixed_convert_from_int): Likewise.
18334 (fixed_convert_from_real): Likewise.
18335 (real_convert_from_fixed): Likewise.
18336 * fixed-value.c (fixed_from_double_int): Likewise.
18337 (fixed_from_string): Likewise.
18338 (fixed_convert): Likewise.
18339 (fixed_convert_from_int): Likewise.
18340 (fixed_convert_from_real): Likewise.
18341 (real_convert_from_fixed): Likewise.
18342 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18345 Alan Hayward <alan.hayward@arm.com>
18346 David Sherwood <david.sherwood@arm.com>
18348 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18349 of separate mode class checks. Do not allow vector modes here.
18350 (immed_wide_int_const): Use as_a <scalar_mode>.
18351 * explow.c (trunc_int_for_mode): Likewise.
18352 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18353 (wi::shwi): Likewise.
18354 (wi::min_value): Likewise.
18355 (wi::max_value): Likewise.
18356 * dwarf2out.c (loc_descriptor): Likewise.
18357 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18358 for CONST_WIDE_INT.
18360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18361 Alan Hayward <alan.hayward@arm.com>
18362 David Sherwood <david.sherwood@arm.com>
18364 * tree.h (SCALAR_TYPE_MODE): New macro.
18365 * expr.c (expand_expr_addr_expr_1): Use it.
18366 (expand_expr_real_2): Likewise.
18367 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18368 (fold_convert_const_fixed_from_int): Likewise.
18369 (fold_convert_const_fixed_from_real): Likewise.
18370 (native_encode_fixed): Likewise
18371 (native_encode_complex): Likewise
18372 (native_encode_vector): Likewise.
18373 (native_interpret_fixed): Likewise.
18374 (native_interpret_real): Likewise.
18375 (native_interpret_complex): Likewise.
18376 (native_interpret_vector): Likewise.
18377 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18378 (simd_clone_adjust_argument_types): Likewise.
18379 (simd_clone_init_simd_arrays): Likewise.
18380 (simd_clone_adjust): Likewise.
18381 * stor-layout.c (layout_type): Likewise.
18382 * tree.c (build_minus_one_cst): Likewise.
18383 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18384 * tree-inline.c (estimate_move_cost): Likewise.
18385 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18386 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18387 (vectorizable_reduction): Likewise.
18388 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18389 (vect_recog_mixed_size_cond_pattern): Likewise.
18390 (check_bool_pattern): Likewise.
18391 (adjust_bool_pattern): Likewise.
18392 (search_type_for_mask_1): Likewise.
18393 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18394 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18395 (vectorizable_load): Likewise.
18396 (vectorizable_store): Likewise.
18397 * ubsan.c (ubsan_encode_value): Likewise.
18398 * varasm.c (output_constant): Likewise.
18400 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18401 Alan Hayward <alan.hayward@arm.com>
18402 David Sherwood <david.sherwood@arm.com>
18404 * coretypes.h (scalar_mode): New class.
18405 * machmode.h (scalar_mode): Likewise.
18406 (scalar_mode::includes_p): New function.
18407 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18408 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18409 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18410 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18411 * expmed.c (store_bit_field_1): Likewise.
18412 (extract_bit_field_1): Likewise.
18413 * expr.c (write_complex_part): Likewise.
18414 (read_complex_part): Likewise.
18415 (emit_move_complex_push): Likewise.
18416 (expand_expr_real_2): Likewise.
18417 * function.c (assign_parm_setup_reg): Likewise.
18418 (assign_parms_unsplit_complex): Likewise.
18419 * optabs.c (expand_binop): Likewise.
18420 * rtlanal.c (subreg_get_info): Likewise.
18421 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18422 * varasm.c (output_constant_pool_2): Likewise.
18424 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18425 Alan Hayward <alan.hayward@arm.com>
18426 David Sherwood <david.sherwood@arm.com>
18428 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18430 (expmed_mult_highpart_optab): Likewise.
18431 (expmed_mult_highpart): Likewise.
18433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18434 Alan Hayward <alan.hayward@arm.com>
18435 David Sherwood <david.sherwood@arm.com>
18437 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18438 instead of a machine_mode.
18439 (builtin_memset_read_str): Likewise.
18440 * builtins.c (c_readstr): Likewise.
18441 (builtin_memcpy_read_str): Likewise.
18442 (builtin_strncpy_read_str): Likewise.
18443 (builtin_memset_read_str): Likewise.
18444 (builtin_memset_gen_str): Likewise.
18445 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18446 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18447 instead of a machine_mode.
18448 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18450 (make_extraction): Likewise.
18451 (try_widen_shift_mode): Take and return scalar_int_modes instead
18453 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18454 a scalar_int_mode instead of a machine_mode.
18455 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18456 (avr_addr_space_pointer_mode): Likewise.
18457 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18458 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18459 (msp430_unwind_word_mode): Likewise.
18460 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18461 (spu_addr_space_pointer_mode): Likewise.
18462 (spu_addr_space_address_mode): Likewise.
18463 (spu_libgcc_cmp_return_mode): Likewise.
18464 (spu_libgcc_shift_count_mode): Likewise.
18465 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18466 (rl78_addr_space_pointer_mode): Likewise.
18467 (fl78_unwind_word_mode): Likewise.
18468 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18470 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18471 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18472 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18473 (mips_valid_pointer_mode): Likewise.
18474 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18475 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18476 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18478 (ft32_addr_space_address_mode): Likewise.
18479 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18480 scalar_int_mode instead of a machine_mode.
18481 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18483 (m32c_addr_space_address_mode): Likewise.
18484 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18485 (rs6000_eh_return_filter_mode): Likewise.
18486 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18487 (rs6000_eh_return_filter_mode): Likewise.
18488 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18489 (s390_libgcc_shift_count_mode): Likewise.
18490 (s390_unwind_word_mode): Likewise.
18491 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18493 * target.def (mode_rep_extended): Likewise.
18494 (valid_pointer_mode): Likewise.
18495 (addr_space.valid_pointer_mode): Likewise.
18496 (eh_return_filter_mode): Return a scalar_int_mode rather than
18498 (libgcc_cmp_return_mode): Likewise.
18499 (libgcc_shift_count_mode): Likewise.
18500 (unwind_word_mode): Likewise.
18501 (addr_space.pointer_mode): Likewise.
18502 (addr_space.address_mode): Likewise.
18503 * doc/tm.texi: Regenerate.
18504 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18506 (do_jump): Use scalar_int_mode for local variables.
18507 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18508 rather than a machine_mode.
18509 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18510 (scompare_loc_descriptor_wide): Likewise.
18511 (scompare_loc_descriptor_narrow): Likewise.
18512 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18514 * except.c (sjlj_emit_dispatch_table): Likewise.
18515 (expand_builtin_eh_copy_values): Likewise.
18516 * explow.c (convert_memory_address_addr_space_1): Likewise.
18517 Take a scalar_int_mode rather than a machine_mode.
18518 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18519 than a machine_mode.
18520 (memory_address_addr_space): Use scalar_int_mode for local variables.
18521 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18522 rather than a machine_mode.
18523 * expmed.c (mask_rtx): Likewise.
18524 (init_expmed_one_conv): Likewise.
18525 (expand_mult_highpart_adjust): Likewise.
18526 (extract_high_half): Likewise.
18527 (expmed_mult_highpart_optab): Likewise.
18528 (expmed_mult_highpart): Likewise.
18529 (expand_smod_pow2): Likewise.
18530 (expand_sdiv_pow2): Likewise.
18531 (emit_store_flag_int): Likewise.
18532 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18534 (extract_low_bits): Likewise.
18535 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18537 * expr.c (pieces_addr::adjust): Likewise.
18538 (can_store_by_pieces): Likewise.
18539 (store_by_pieces): Likewise.
18540 (clear_by_pieces_1): Likewise.
18541 (expand_expr_addr_expr_1): Likewise.
18542 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18543 (expand_expr_real_1): Likewise.
18544 (try_casesi): Likewise.
18545 * final.c (shorten_branches): Likewise.
18546 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18547 type of "mode" to machine_mode.
18548 * internal-fn.c (expand_arith_overflow_result_store): Take a
18549 scalar_int_mode rather than a machine_mode.
18550 (expand_mul_overflow): Use scalar_int_mode for local variables.
18551 * loop-doloop.c (doloop_modify): Likewise.
18552 (doloop_optimize): Likewise.
18553 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18554 than a machine_mode.
18555 (expand_doubleword_shift_condmove): Likewise.
18556 (expand_doubleword_shift): Likewise.
18557 (expand_doubleword_clz): Likewise.
18558 (expand_doubleword_popcount): Likewise.
18559 (expand_doubleword_parity): Likewise.
18560 (expand_absneg_bit): Use scalar_int_mode for local variables.
18561 (prepare_float_lib_cmp): Likewise.
18562 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18563 rather than a machine_mode.
18564 (convert_memory_address_addr_space): Likewise.
18565 (get_mode_bounds): Likewise.
18566 (get_address_mode): Return a scalar_int_mode rather than a
18568 * rtlanal.c (get_address_mode): Likewise.
18569 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18570 than a machine_mode.
18571 * targhooks.c (default_mode_rep_extended): Likewise.
18572 (default_valid_pointer_mode): Likewise.
18573 (default_addr_space_valid_pointer_mode): Likewise.
18574 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18575 than a machine_mode.
18576 (default_libgcc_cmp_return_mode): Likewise.
18577 (default_libgcc_shift_count_mode): Likewise.
18578 (default_unwind_word_mode): Likewise.
18579 (default_addr_space_pointer_mode): Likewise.
18580 (default_addr_space_address_mode): Likewise.
18581 * targhooks.h (default_eh_return_filter_mode): Likewise.
18582 (default_libgcc_cmp_return_mode): Likewise.
18583 (default_libgcc_shift_count_mode): Likewise.
18584 (default_unwind_word_mode): Likewise.
18585 (default_addr_space_pointer_mode): Likewise.
18586 (default_addr_space_address_mode): Likewise.
18587 (default_mode_rep_extended): Take a scalar_int_mode rather than
18589 (default_valid_pointer_mode): Likewise.
18590 (default_addr_space_valid_pointer_mode): Likewise.
18591 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18593 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18594 rather than a machine_mode.
18595 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18596 for local variables.
18597 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18598 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18599 than a machine_mode.
18601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18602 Alan Hayward <alan.hayward@arm.com>
18603 David Sherwood <david.sherwood@arm.com>
18605 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18606 the mode argument to scalar_int_mode.
18607 (do_jump_by_parts_zero_rtx): Likewise.
18608 (do_jump_by_parts_equality_rtx): Likewise.
18609 (do_jump_by_parts_greater): Take a mode argument.
18610 (do_jump_by_parts_equality): Likewise.
18611 (do_jump_1): Update calls accordingly.
18613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18614 Alan Hayward <alan.hayward@arm.com>
18615 David Sherwood <david.sherwood@arm.com>
18617 * is-a.h (safe_dyn_cast): New function.
18618 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18619 (jump_table_for_label): Likewise.
18620 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18621 instead of an rtx_insn *.
18622 (shorten_branches): Use dyn_cast instead of LABEL_P and
18623 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18624 rtx_jump_table_data::get_data_mode.
18625 (final_scan_insn): Likewise.
18627 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18628 Alan Hayward <alan.hayward@arm.com>
18629 David Sherwood <david.sherwood@arm.com>
18631 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18632 trying to combine a full-register integer set with a subreg
18635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18636 Alan Hayward <alan.hayward@arm.com>
18637 David Sherwood <david.sherwood@arm.com>
18639 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18640 that is always either address_mode or pointer_mode.
18642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18643 Alan Hayward <alan.hayward@arm.com>
18644 David Sherwood <david.sherwood@arm.com>
18646 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18647 when the two are known to be equal.
18649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18650 Alan Hayward <alan.hayward@arm.com>
18651 David Sherwood <david.sherwood@arm.com>
18653 * simplify-rtx.c (simplify_const_unary_operation): Use
18654 is_a <scalar_int_mode> instead of checking for a nonzero
18655 precision. Forcibly convert op_mode to a scalar_int_mode
18656 in that case. More clearly differentiate the operand and
18657 result modes and use the former when deciding what the value
18658 of a count-bits operation should be. Use is_int_mode instead
18659 of checking for a MODE_INT. Remove redundant check for whether
18660 this mode has a zero precision.
18662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18663 Alan Hayward <alan.hayward@arm.com>
18664 David Sherwood <david.sherwood@arm.com>
18666 * optabs.c (widen_leading): Change the type of the mode argument
18667 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18668 (widen_bswap): Likewise.
18669 (expand_parity): Likewise.
18670 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18671 (expand_ffs): Likewise.
18672 (epand_unop): Check for scalar integer modes before calling the
18675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18676 Alan Hayward <alan.hayward@arm.com>
18677 David Sherwood <david.sherwood@arm.com>
18679 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18681 (expand_expr_real_1): Update call accordingly.
18683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18684 Alan Hayward <alan.hayward@arm.com>
18685 David Sherwood <david.sherwood@arm.com>
18687 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18688 value_mode arguments. Use scalar_int_mode internally.
18689 (store_bit_field_1): Rename the new integer mode from imode
18690 to op0_mode and use it instead of GET_MODE (op0). Update calls
18691 to store_split_bit_field, store_bit_field_using_insv and
18692 store_fixed_bit_field.
18693 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18694 Use scalar_int_mode internally. Use a bit count rather than a mode
18695 when calculating the largest bit size for get_best_mode.
18696 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18697 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18698 Remove assertion that OP0 has a scalar integer mode.
18699 (store_split_bit_field): Add op0_mode and value_mode arguments.
18700 Update calls to extract_fixed_bit_field.
18701 (extract_bit_field_using_extv): Add an op0_mode argument.
18702 Use scalar_int_mode internally.
18703 (extract_bit_field_1): Rename the new integer mode from imode to
18704 op0_mode and use it instead of GET_MODE (op0). Update calls to
18705 extract_split_bit_field, extract_bit_field_using_extv and
18706 extract_fixed_bit_field.
18707 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18708 to extract_split_bit_field and extract_fixed_bit_field_1.
18709 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18710 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18711 on the target mode.
18712 (extract_split_bit_field): Add an op0_mode argument. Update call
18713 to extract_fixed_bit_field.
18715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18716 Alan Hayward <alan.hayward@arm.com>
18717 David Sherwood <david.sherwood@arm.com>
18719 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18720 * explow.c (hard_function_value): Likewise.
18721 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18722 convert_to_mode call outside the loop.
18723 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18724 for the mode iterator. Require the mode specified by max_pieces
18726 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18728 (copy_blkmode_to_reg): Likewise.
18729 (set_storage_via_setmem): Likewise.
18730 * optabs.c (prepare_cmp_insn): Likewise.
18731 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18732 * stor-layout.c (finish_bitfield_representative): Likewise.
18734 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18735 Alan Hayward <alan.hayward@arm.com>
18736 David Sherwood <david.sherwood@arm.com>
18738 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18739 * expr.c (convert_move): Use them.
18740 (convert_modes): Likewise.
18741 (store_expr_with_bounds): Likewise.
18743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18744 Alan Hayward <alan.hayward@arm.com>
18745 David Sherwood <david.sherwood@arm.com>
18747 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18748 parameter for the mode of "x". Remove the "known_x", "known_mode"
18749 and "known_ret" arguments. Change the type of the mode argument
18750 to scalar_int_mode.
18751 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18752 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18753 (reg_num_sign_bit_copies_for_combine): Likewise.
18754 * rtlanal.c (nonzero_bits1): Likewise.
18755 (num_sign_bit_copies1): Likewise.
18756 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18757 (reg_num_sign_bit_copies_general): Likewise.
18758 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18759 (reg_nonzero_bits_general): Likewise.
18761 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18762 Alan Hayward <alan.hayward@arm.com>
18763 David Sherwood <david.sherwood@arm.com>
18765 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18766 than in subroutines. Return 1 for non-integer modes.
18767 (cached_num_sign_bit_copies): Change the type of the mode parameter
18768 to scalar_int_mode.
18769 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18770 classes. Handle CONST_INT_P first and then check whether X also
18771 has a scalar integer mode. Check the same thing for inner registers
18772 of a SUBREG and for values that are being extended or truncated.
18774 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18775 Alan Hayward <alan.hayward@arm.com>
18776 David Sherwood <david.sherwood@arm.com>
18778 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18779 in subroutines. Return the mode mask for non-integer modes.
18780 (cached_nonzero_bits): Change the type of the mode parameter
18781 to scalar_int_mode.
18782 (nonzero_bits1): Likewise. Remove early exit for other mode
18783 classes. Handle CONST_INT_P first and then check whether X
18784 also has a scalar integer mode.
18786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18787 Alan Hayward <alan.hayward@arm.com>
18788 David Sherwood <david.sherwood@arm.com>
18790 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18791 Return a scalar_int_mode and assert that the size is greater than
18793 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18794 (op_by_pieces_d::op_by_pieces_d): Likewise.
18795 (op_by_pieces_d::run): Likewise.
18796 (can_store_by_pieces): Likewise.
18798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18799 Alan Hayward <alan.hayward@arm.com>
18800 David Sherwood <david.sherwood@arm.com>
18802 * combine.c (extract_left_shift): Add a mode argument and update
18804 (make_compound_operation_int): Change the type of the mode parameter
18805 to scalar_int_mode and update the call to extract_left_shift.
18807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18808 Alan Hayward <alan.hayward@arm.com>
18809 David Sherwood <david.sherwood@arm.com>
18811 * combine.c (simplify_and_const_int): Change the type of the mode
18812 parameter to scalar_int_mode.
18813 (simplify_and_const_int_1): Likewise. Update recursive call.
18815 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18816 Alan Hayward <alan.hayward@arm.com>
18817 David Sherwood <david.sherwood@arm.com>
18819 * combine.c (simplify_compare_const): Check that the mode is a
18820 scalar_int_mode (rather than VOIDmode) before testing its
18822 (simplify_comparison): Move COMPARISON_P handling out of the
18823 loop and restrict the latter part of the loop to scalar_int_modes.
18824 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
18825 and when considering SUBREG_REGs. Use is_int_mode instead of
18826 checking GET_MODE_CLASS against MODE_INT.
18828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18829 Alan Hayward <alan.hayward@arm.com>
18830 David Sherwood <david.sherwood@arm.com>
18832 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18833 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18834 shift_unit_mode and for modes involved in scalar shifts.
18836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18837 Alan Hayward <alan.hayward@arm.com>
18838 David Sherwood <david.sherwood@arm.com>
18840 * combine.c (force_int_to_mode): New function, split out from...
18841 (force_to_mode): ...here. Keep xmode up-to-date and use it
18842 instead of GET_MODE (x).
18844 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18845 Alan Hayward <alan.hayward@arm.com>
18846 David Sherwood <david.sherwood@arm.com>
18848 * optabs-query.h (extraction_insn::struct_mode): Change type to
18849 opt_scalar_int_mode and update comment.
18850 (extraction_insn::field_mode): Change type to scalar_int_mode.
18851 (extraction_insn::pos_mode): Likewise.
18852 * combine.c (make_extraction): Update accordingly.
18853 * optabs-query.c (get_traditional_extraction_insn): Likewise.
18854 (get_optab_extraction_insn): Likewise.
18855 * recog.c (simplify_while_replacing): Likewise.
18856 * expmed.c (narrow_bit_field_mem): Change the type of the mode
18857 parameter to opt_scalar_int_mode.
18859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18860 Alan Hayward <alan.hayward@arm.com>
18861 David Sherwood <david.sherwood@arm.com>
18863 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18864 to a scalar_int_mode instead of a machine_mode.
18865 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
18866 (get_best_mode): Return a boolean and use a pointer argument to store
18867 the selected mode. Replace the limit mode parameter with a bit limit.
18868 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
18869 for the values returned by bit_field_mode_iterator::next_mode.
18870 (store_bit_field): Update call to get_best_mode.
18871 (store_fixed_bit_field): Likewise.
18872 (extract_fixed_bit_field): Likewise.
18873 * expr.c (optimize_bitfield_assignment_op): Likewise.
18874 * fold-const.c (optimize_bit_field_compare): Likewise.
18875 (fold_truth_andor_1): Likewise.
18876 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
18877 Update for new type of m_mode.
18878 (get_best_mode): As above.
18880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18881 Alan Hayward <alan.hayward@arm.com>
18882 David Sherwood <david.sherwood@arm.com>
18884 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18885 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18886 (store_bit_field): Check is_a <scalar_int_mode> before calling
18887 strict_volatile_bitfield_p.
18888 (extract_bit_field): Likewise.
18890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18891 Alan Hayward <alan.hayward@arm.com>
18892 David Sherwood <david.sherwood@arm.com>
18894 * target.def (cstore_mode): Return a scalar_int_mode.
18895 * doc/tm.texi: Regenerate.
18896 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18897 * targhooks.h (default_cstore_mode): Likewise.
18898 * targhooks.c (default_cstore_mode): Likewise, using a forced
18900 * expmed.c (emit_cstore): Expect the target of the cstore to be
18903 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18904 Alan Hayward <alan.hayward@arm.com>
18905 David Sherwood <david.sherwood@arm.com>
18907 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18909 (niter_desc): Likewise mode.
18910 (iv_analyze): Add a mode parameter.
18912 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18913 and change its type to scalar_int_mode.
18914 * loop-iv.c: Update commentary at head of file.
18915 (iv_constant): Pass the mode paraeter before the rtx it describes
18916 and change its type to scalar_int_mode. Remove VOIDmode handling.
18917 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18918 (iv_extend): Likewise.
18919 (shorten_into_mode): Likewise.
18920 (iv_add): Use scalar_int_mode.
18921 (iv_mult): Likewise.
18922 (iv_shift): Likewise.
18923 (canonicalize_iv_subregs): Likewise.
18924 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18925 it describes and change its mode to scalar_int_mode. Also change
18926 the type of the returned inner_mode to scalar_int_mode.
18927 (get_biv_step): Likewise, turning outer_mode from a pointer
18928 into a direct parameter. Update call to get_biv_step_1.
18929 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18930 iv_constant and get_biv_step.
18931 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18932 and change its type to scalar_int_mode. Don't initialise iv->mode
18933 to VOIDmode and remove later checks for its still being VOIDmode.
18934 Update calls to iv_analyze_op and iv_analyze_expr. Check
18935 is_a <scalar_int_mode> when changing the mode under consideration.
18936 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18937 Update call to iv_analyze_expr.
18938 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18939 inner register is not also a scalar_int_mode. Update call to
18941 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18942 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18943 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18944 separate mode class checks. Update calls to iv_analyze. Remove
18945 fix-up of VOIDmodes after iv_analyze_biv.
18946 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18947 don't have a scalar_int_mode. Update call to biv_p.
18949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18950 Alan Hayward <alan.hayward@arm.com>
18951 David Sherwood <david.sherwood@arm.com>
18953 * cfgexpand.c (convert_debug_memory_address): Use
18954 as_a <scalar_int_mode>.
18955 * combine.c (expand_compound_operation): Likewise.
18956 (make_extraction): Likewise.
18957 (change_zero_ext): Likewise.
18958 (simplify_comparison): Likewise.
18959 * cse.c (cse_insn): Likewise.
18960 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18961 (mem_loc_descriptor): Likewise.
18962 (loc_descriptor): Likewise.
18963 * expmed.c (init_expmed_one_mode): Likewise.
18964 (synth_mult): Likewise.
18965 (emit_store_flag_1): Likewise.
18966 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18967 of a comparison with size.
18968 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18969 (reduce_to_bit_field_precision): Likewise.
18970 * function.c (expand_function_end): Likewise.
18971 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18972 * loop-doloop.c (doloop_modify): Likewise.
18973 * optabs.c (expand_binop): Likewise.
18974 (expand_unop): Likewise.
18975 (expand_copysign_absneg): Likewise.
18976 (prepare_cmp_insn): Likewise.
18977 (maybe_legitimize_operand): Likewise.
18978 * recog.c (const_scalar_int_operand): Likewise.
18979 * rtlanal.c (get_address_mode): Likewise.
18980 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18981 (simplify_cond_clz_ctz): Likewise.
18982 * tree-nested.c (get_nl_goto_field): Likewise.
18983 * tree.c (build_vector_type_for_mode): Likewise.
18984 * var-tracking.c (use_narrower_mode): Likewise.
18986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18987 Alan Hayward <alan.hayward@arm.com>
18988 David Sherwood <david.sherwood@arm.com>
18990 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18991 * builtins.c (expand_builtin_signbit): Use it.
18992 * cfgexpand.c (expand_debug_expr): Likewise.
18993 * dojump.c (do_jump): Likewise.
18994 (do_compare_and_jump): Likewise.
18995 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18996 * expmed.c (make_tree): Likewise.
18997 * expr.c (expand_expr_real_2): Likewise.
18998 (expand_expr_real_1): Likewise.
18999 (try_casesi): Likewise.
19000 * fold-const-call.c (fold_const_call_ss): Likewise.
19001 * fold-const.c (unextend): Likewise.
19002 (extract_muldiv_1): Likewise.
19003 (fold_single_bit_test): Likewise.
19004 (native_encode_int): Likewise.
19005 (native_encode_string): Likewise.
19006 (native_interpret_int): Likewise.
19007 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
19008 * internal-fn.c (expand_addsub_overflow): Likewise.
19009 (expand_neg_overflow): Likewise.
19010 (expand_mul_overflow): Likewise.
19011 (expand_arith_overflow): Likewise.
19012 * match.pd: Likewise.
19013 * stor-layout.c (layout_type): Likewise.
19014 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19015 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19016 * tree-ssanames.c (get_range_info): Likewise.
19017 * tree-switch-conversion.c (array_value_type) Likewise.
19018 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19019 (vect_recog_divmod_pattern): Likewise.
19020 (vect_recog_mixed_size_cond_pattern): Likewise.
19021 * tree-vrp.c (extract_range_basic): Likewise.
19022 (simplify_float_conversion_using_ranges): Likewise.
19023 * tree.c (int_fits_type_p): Likewise.
19024 * ubsan.c (instrument_bool_enum_load): Likewise.
19025 * varasm.c (mergeable_string_section): Likewise.
19026 (narrowing_initializer_constant_valid_p): Likewise.
19027 (output_constant): Likewise.
19029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19030 Alan Hayward <alan.hayward@arm.com>
19031 David Sherwood <david.sherwood@arm.com>
19033 * machmode.h (NARROWEST_INT_MODE): New macro.
19034 * expr.c (alignment_for_piecewise_move): Use it instead of
19035 GET_CLASS_NARROWEST_MODE (MODE_INT).
19036 (push_block): Likewise.
19037 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19039 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19042 Alan Hayward <alan.hayward@arm.com>
19043 David Sherwood <david.sherwood@arm.com>
19045 * postreload.c (move2add_valid_value_p): Change the type of the
19046 mode parameter to scalar_int_mode.
19047 (move2add_use_add2_insn): Add a mode parameter and use it instead
19049 (move2add_use_add3_insn): Likewise.
19050 (reload_cse_move2add): Update accordingly.
19052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19053 Alan Hayward <alan.hayward@arm.com>
19054 David Sherwood <david.sherwood@arm.com>
19056 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19058 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19059 * optabs.c (expand_unop): Likewise.
19061 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19062 Alan Hayward <alan.hayward@arm.com>
19063 David Sherwood <david.sherwood@arm.com>
19065 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19066 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19067 (popcount_loc_descriptor): Likewise.
19068 (bswap_loc_descriptor): Likewise.
19069 (rotate_loc_descriptor): Likewise.
19070 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19071 calling the functions above.
19073 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19074 Alan Hayward <alan.hayward@arm.com>
19075 David Sherwood <david.sherwood@arm.com>
19077 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19079 (try_combine): Likewise.
19080 (simplify_if_then_else): Likewise.
19081 * cse.c (cse_insn): Likewise.
19082 * dwarf2out.c (mem_loc_descriptor): Likewise.
19083 * emit-rtl.c (gen_lowpart_common): Likewise.
19084 * simplify-rtx.c (simplify_truncation): Likewise.
19085 (simplify_binary_operation_1): Likewise.
19086 (simplify_const_relational_operation): Likewise.
19087 (simplify_ternary_operation): Likewise.
19088 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19091 Alan Hayward <alan.hayward@arm.com>
19092 David Sherwood <david.sherwood@arm.com>
19094 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19095 * reload.c (push_reload): Likewise.
19096 (find_reloads): Likewise.
19098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19099 Alan Hayward <alan.hayward@arm.com>
19100 David Sherwood <david.sherwood@arm.com>
19102 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19103 (make_compound_operation_int): Likewise.
19104 (change_zero_ext): Likewise.
19105 * expr.c (convert_move): Likewise.
19106 (convert_modes): Likewise.
19107 * fwprop.c (forward_propagate_subreg): Likewise.
19108 * loop-iv.c (get_biv_step_1): Likewise.
19109 * optabs.c (widen_operand): Likewise.
19110 * postreload.c (move2add_valid_value_p): Likewise.
19111 * recog.c (simplify_while_replacing): Likewise.
19112 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19113 (simplify_binary_operation_1): Likewise. Remove redundant
19114 mode equality check.
19116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19117 Alan Hayward <alan.hayward@arm.com>
19118 David Sherwood <david.sherwood@arm.com>
19120 * combine.c (combine_simplify_rtx): Add checks for
19121 is_a <scalar_int_mode>.
19122 (simplify_if_then_else): Likewise.
19123 (make_field_assignment): Likewise.
19124 (simplify_comparison): Likewise.
19125 * ifcvt.c (noce_try_bitop): Likewise.
19126 * loop-invariant.c (canonicalize_address_mult): Likewise.
19127 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19130 Alan Hayward <alan.hayward@arm.com>
19131 David Sherwood <david.sherwood@arm.com>
19133 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19134 is_a <scalar_int_mode> instead of != BLKmode.
19136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19137 Alan Hayward <alan.hayward@arm.com>
19138 David Sherwood <david.sherwood@arm.com>
19140 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19141 instead of != VOIDmode.
19142 * combine.c (if_then_else_cond): Likewise.
19143 (change_zero_ext): Likewise.
19144 * dwarf2out.c (mem_loc_descriptor): Likewise.
19145 (loc_descriptor): Likewise.
19146 * rtlanal.c (canonicalize_condition): Likewise.
19147 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19150 Alan Hayward <alan.hayward@arm.com>
19151 David Sherwood <david.sherwood@arm.com>
19153 * simplify-rtx.c (simplify_binary_operation_1): Use
19154 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19157 Alan Hayward <alan.hayward@arm.com>
19158 David Sherwood <david.sherwood@arm.com>
19160 * wide-int.h (int_traits<unsigned char>) New class.
19161 (int_traits<unsigned short>) Likewise.
19162 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19163 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19165 * combine.c (set_nonzero_bits_and_sign_copies): Use
19166 is_a <scalar_int_mode>.
19167 (find_split_point): Likewise.
19168 (combine_simplify_rtx): Likewise.
19169 (simplify_logical): Likewise.
19170 (expand_compound_operation): Likewise.
19171 (expand_field_assignment): Likewise.
19172 (make_compound_operation): Likewise.
19173 (extended_count): Likewise.
19174 (change_zero_ext): Likewise.
19175 (simplify_comparison): Likewise.
19176 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19177 (ucompare_loc_descriptor): Likewise.
19178 (minmax_loc_descriptor): Likewise.
19179 (mem_loc_descriptor): Likewise.
19180 (loc_descriptor): Likewise.
19181 * expmed.c (init_expmed_one_mode): Likewise.
19182 * lra-constraints.c (lra_constraint_offset): Likewise.
19183 * optabs.c (prepare_libcall_arg): Likewise.
19184 * postreload.c (move2add_note_store): Likewise.
19185 * reload.c (operands_match_p): Likewise.
19186 * rtl.h (load_extend_op): Likewise.
19187 * rtlhooks.c (gen_lowpart_general): Likewise.
19188 * simplify-rtx.c (simplify_truncation): Likewise.
19189 (simplify_unary_operation_1): Likewise.
19190 (simplify_binary_operation_1): Likewise.
19191 (simplify_const_binary_operation): Likewise.
19192 (simplify_const_relational_operation): Likewise.
19193 (simplify_subreg): Likewise.
19194 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19195 * var-tracking.c (adjust_mems): Likewise.
19196 (prepare_call_arguments): Likewise.
19198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19199 Alan Hayward <alan.hayward@arm.com>
19200 David Sherwood <david.sherwood@arm.com>
19202 * machmode.h (is_int_mode): New fuction.
19203 * combine.c (find_split_point): Use it.
19204 (combine_simplify_rtx): Likewise.
19205 (simplify_if_then_else): Likewise.
19206 (simplify_set): Likewise.
19207 (simplify_shift_const_1): Likewise.
19208 (simplify_comparison): Likewise.
19209 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19210 * cse.c (notreg_cost): Likewise.
19211 (cse_insn): Likewise.
19212 * cselib.c (cselib_lookup_1): Likewise.
19213 * dojump.c (do_jump_1): Likewise.
19214 (do_compare_rtx_and_jump): Likewise.
19215 * dse.c (get_call_args): Likewise.
19216 * dwarf2out.c (rtl_for_decl_init): Likewise.
19217 (native_encode_initializer): Likewise.
19218 * expmed.c (emit_store_flag_1): Likewise.
19219 (emit_store_flag): Likewise.
19220 * expr.c (convert_modes): Likewise.
19221 (store_field): Likewise.
19222 (expand_expr_real_1): Likewise.
19223 * fold-const.c (fold_read_from_constant_string): Likewise.
19224 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19225 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19226 * optabs.c (expand_binop): Likewise.
19227 (expand_unop): Likewise.
19228 (expand_abs_nojump): Likewise.
19229 (expand_one_cmpl_abs_nojump): Likewise.
19230 * simplify-rtx.c (mode_signbit_p): Likewise.
19231 (val_signbit_p): Likewise.
19232 (val_signbit_known_set_p): Likewise.
19233 (val_signbit_known_clear_p): Likewise.
19234 (simplify_relational_operation_1): Likewise.
19235 * tree.c (vector_type_mode): Likewise.
19237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19238 Alan Hayward <alan.hayward@arm.com>
19239 David Sherwood <david.sherwood@arm.com>
19241 * machmode.h (smallest_mode_for_size): Fix formatting.
19242 (smallest_int_mode_for_size): New function.
19243 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19244 instead of smallest_mode_for_size.
19245 * combine.c (make_extraction): Likewise.
19246 * config/arc/arc.c (arc_expand_movmem): Likewise.
19247 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19248 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19249 * config/s390/s390.c (s390_expand_insv): Likewise.
19250 * config/sparc/sparc.c (assign_int_registers): Likewise.
19251 * config/spu/spu.c (spu_function_value): Likewise.
19252 (spu_function_arg): Likewise.
19253 * coverage.c (get_gcov_type): Likewise.
19254 (get_gcov_unsigned_t): Likewise.
19255 * dse.c (find_shift_sequence): Likewise.
19256 * expmed.c (store_bit_field_1): Likewise.
19257 * expr.c (convert_move): Likewise.
19258 (store_field): Likewise.
19259 * internal-fn.c (expand_arith_overflow): Likewise.
19260 * optabs-query.c (get_best_extraction_insn): Likewise.
19261 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19262 * stor-layout.c (layout_type): Likewise.
19263 (initialize_sizetypes): Likewise.
19264 * targhooks.c (default_get_mask_mode): Likewise.
19265 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19268 Alan Hayward <alan.hayward@arm.com>
19269 David Sherwood <david.sherwood@arm.com>
19271 * machmode.h (opt_mode::else_blk): New function.
19272 (int_mode_for_mode): Declare.
19273 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19274 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19276 * cfgexpand.c (expand_debug_expr): Likewise.
19277 * combine.c (gen_lowpart_or_truncate): Likewise.
19278 (gen_lowpart_for_combine): Likewise.
19279 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19280 * config/avr/avr.c (avr_to_int_mode): Likewise.
19281 (avr_out_plus_1): Likewise.
19282 (avr_out_plus): Likewise.
19283 (avr_out_round): Likewise.
19284 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19285 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19286 (s390_expand_vcond): Likewise.
19287 * config/spu/spu.c (spu_split_immediate): Likewise.
19288 (spu_expand_mov): Likewise.
19289 * dse.c (get_stored_val): Likewise.
19290 * expmed.c (store_bit_field_1): Likewise.
19291 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19293 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19294 (extract_low_bits): Likewise.
19295 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19296 handling rather than repeating the check.
19297 (emit_group_store): Likewise.
19298 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19299 * optabs.c (expand_absneg_bit): Likewise.
19300 (expand_copysign_absneg): Likewise.
19301 (expand_copysign_bit): Likewise.
19302 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19303 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19304 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19305 * var-tracking.c (prepare_call_arguments): Likewise.
19306 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19307 int_mode_for_mode instead of mode_for_size.
19308 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19311 Alan Hayward <alan.hayward@arm.com>
19312 David Sherwood <david.sherwood@arm.com>
19314 * machmode.h (int_mode_for_size): New function.
19315 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19316 instead of mode_for_size.
19317 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19319 * combine.c (expand_field_assignment): Use int_mode_for_size
19320 instead of mode_for_size.
19321 (make_extraction): Likewise.
19322 (simplify_shift_const_1): Likewise.
19323 (simplify_comparison): Likewise.
19324 * dojump.c (do_jump): Likewise.
19325 * dwarf2out.c (mem_loc_descriptor): Likewise.
19326 * emit-rtl.c (init_derived_machine_modes): Likewise.
19327 * expmed.c (flip_storage_order): Likewise.
19328 (convert_extracted_bit_field): Likewise.
19329 * expr.c (copy_blkmode_from_reg): Likewise.
19330 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19331 * internal-fn.c (expand_mul_overflow): Likewise.
19332 * lower-subreg.c (simple_move): Likewise.
19333 * optabs-libfuncs.c (init_optabs): Likewise.
19334 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19335 * tree.c (vector_type_mode): Likewise.
19336 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19337 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19338 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19339 * tree-vect-stmts.c (vectorizable_load): Likewise.
19340 (vectorizable_store): Likewise.
19342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19343 Alan Hayward <alan.hayward@arm.com>
19344 David Sherwood <david.sherwood@arm.com>
19346 * coretypes.h (pod_mode): New type.
19347 (scalar_int_mode_pod): New typedef.
19348 * machmode.h (pod_mode): New class.
19349 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19350 * genmodes.c (emit_mode_int_n): Update accordingly.
19351 * lower-subreg.h (target_lower_subreg): Change type to
19352 scalar_int_mode_pod.
19353 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19354 scalar_int_mode_pod.
19356 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19357 Alan Hayward <alan.hayward@arm.com>
19358 David Sherwood <david.sherwood@arm.com>
19360 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19361 machine_mode to scalar_int_mode.
19362 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19363 (rs6000_option_override_internal): Remove cast to int.
19364 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19365 machine_mode to scalar_int_mode.
19366 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19367 (rs6000_option_override_internal): Remove cast to int.
19368 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19369 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19371 * config/s390/s390.c (s390_expand_builtin): Likewise.
19372 * coretypes.h (scalar_int_mode): New type.
19373 (opt_scalar_int_mode): New typedef.
19374 * machmode.h (scalar_int_mode): New class.
19375 (scalar_int_mode::includes_p): New function.
19376 (byte_mode): Change type to scalar_int_mode.
19377 (word_mode): Likewise.
19378 (ptr_mode): Likewise.
19379 * emit-rtl.c (byte_mode): Likewise.
19380 (word_mode): Likewise.
19381 (ptr_mode): Likewise.
19382 (init_derived_machine_modes): Update accordingly.
19383 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19384 and MODE_PARTIAL_INT.
19385 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19386 opt_scalar_int_mode.
19388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19389 Alan Hayward <alan.hayward@arm.com>
19390 David Sherwood <david.sherwood@arm.com>
19392 * target.def (libgcc_floating_mode_supported_p): Take a
19394 * doc/tm.texi: Regenerate.
19395 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19397 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19398 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19402 Alan Hayward <alan.hayward@arm.com>
19403 David Sherwood <david.sherwood@arm.com>
19405 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19406 * doc/tm.texi: Regenerate.
19407 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19408 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19409 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19410 * targhooks.h (default_floatn_mode): Likewise.
19411 * targhooks.c (default_floatn_mode): Likewise.
19412 * tree.c (build_common_tree_nodes): Update accordingly.
19414 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19415 Alan Hayward <alan.hayward@arm.com>
19416 David Sherwood <david.sherwood@arm.com>
19418 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19419 (mode_iterator::iterate_p): Likewise.
19420 (mode_iterator::get_wider): Likewise.
19421 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19424 Alan Hayward <alan.hayward@arm.com>
19425 David Sherwood <david.sherwood@arm.com>
19427 * coretypes.h (opt_scalar_float_mode): New typedef.
19428 * machmode.h (float_mode_for_size): New function.
19429 * emit-rtl.c (double_mode): Delete.
19430 (init_emit_once): Use float_mode_for_size.
19431 * stor-layout.c (layout_type): Likewise.
19432 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19434 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19435 Alan Hayward <alan.hayward@arm.com>
19436 David Sherwood <david.sherwood@arm.com>
19438 * output.h (assemble_real): Take a scalar_float_mode.
19439 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19440 * config/arm/arm.md (consttable_4): Likewise.
19441 (consttable_8): Likewise.
19442 (consttable_16): Likewise.
19443 * config/mips/mips.md (consttable_float): Likewise.
19444 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19445 * varasm.c (assemble_real): Take a scalar_float_mode.
19446 (output_constant_pool_2): Update accordingly.
19447 (output_constant): Likewise.
19449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19450 Alan Hayward <alan.hayward@arm.com>
19451 David Sherwood <david.sherwood@arm.com>
19453 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19454 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19455 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19456 (native_encode_real): Likewise.
19457 (native_interpret_real): Likewise.
19458 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19459 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19461 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19462 Alan Hayward <alan.hayward@arm.com>
19463 David Sherwood <david.sherwood@arm.com>
19465 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19466 <scalar_float_mode>. Simplify.
19467 (gen_extend_conv_libfunc): Likewise.
19469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19470 Alan Hayward <alan.hayward@arm.com>
19471 David Sherwood <david.sherwood@arm.com>
19473 * coretypes.h (scalar_float_mode): New type.
19474 * machmode.h (mode_traits::from_int): Use machine_mode if
19475 USE_ENUM_MODES is defined.
19476 (is_a): New function.
19478 (dyn_cast): Likewise.
19479 (scalar_float_mode): New class.
19480 (scalar_float_mode::includes_p): New function.
19481 (is_float_mode): Likewise.
19482 * gdbhooks.py (MachineModePrinter): New class.
19483 (build_pretty_printer): Use it for scalar_float_mode.
19484 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19485 (format_helper::format_helper): Turn into a template.
19486 * genmodes.c (get_mode_class): New function.
19487 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19488 or machine_mode if none.
19489 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19490 as_a <scalar_float_mode>.
19491 * dwarf2out.c (mem_loc_descriptor): Likewise.
19492 (insert_float): Likewise.
19493 (add_const_value_attribute): Likewise.
19494 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19495 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19496 (expand_unop): Update accordingly.
19497 (expand_abs_nojump): Likewise.
19498 (expand_copysign_absneg): Take a scalar_float_mode.
19499 (expand_copysign_bit): Likewise.
19500 (expand_copysign): Update accordingly.
19502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19503 Alan Hayward <alan.hayward@arm.com>
19504 David Sherwood <david.sherwood@arm.com>
19506 * coretypes.h (opt_mode): New class.
19507 * machmode.h (opt_mode): Likewise.
19508 (opt_mode::else_void): New function.
19509 (opt_mode::require): Likewise.
19510 (opt_mode::exists): Likewise.
19511 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19512 (GET_MODE_2XWIDER_MODE): Likewise.
19513 (mode_iterator::get_wider): Update accordingly.
19514 (mode_iterator::get_2xwider): Likewise.
19515 (mode_iterator::get_known_wider): Likewise, turning into a template.
19516 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19517 forcing a wider mode to exist.
19518 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19519 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19520 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19521 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19522 * lower-subreg.c (init_lower_subreg): Likewise.
19523 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19524 on the final iteration.
19525 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19526 a wider mode exists before asking for a move pattern.
19527 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19528 forcing a wider mode to exist.
19529 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19530 returning false if no such mode exists.
19531 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19532 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19533 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19534 Avoid checking for a MODE_INT if we already know the mode is not a
19536 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19537 forcing a wider mode to exist.
19538 (expmed_mult_highpart_optab): Likewise.
19539 (expmed_mult_highpart): Likewise.
19540 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19542 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19543 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19544 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19545 * internal-fn.c (expand_mul_overflow): Update use of
19546 GET_MODE_2XWIDER_MODE.
19547 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19548 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19549 GET_MODE_WIDER_MODE.
19550 (convert_plusminus_to_widen): Likewise.
19551 * tree-switch-conversion.c (array_value_type): Likewise.
19552 * var-tracking.c (emit_note_insn_var_location): Likewise.
19553 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19554 Return false inside rather than outside the loop if no wider mode
19556 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19557 and GET_MODE_2XWIDER_MODE
19558 (can_compare_p): Use else_void.
19559 * gdbhooks.py (OptMachineModePrinter): New class.
19560 (build_pretty_printer): Use it for opt_mode.
19562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19563 Alan Hayward <alan.hayward@arm.com>
19564 David Sherwood <david.sherwood@arm.com>
19566 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19567 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19570 Alan Hayward <alan.hayward@arm.com>
19571 David Sherwood <david.sherwood@arm.com>
19573 * machmode.h (mode_traits): New structure.
19574 (get_narrowest_mode): New function.
19575 (mode_iterator::start): Likewise.
19576 (mode_iterator::iterate_p): Likewise.
19577 (mode_iterator::get_wider): Likewise.
19578 (mode_iterator::get_known_wider): Likewise.
19579 (mode_iterator::get_2xwider): Likewise.
19580 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19581 (FOR_EACH_MODE): Likewise.
19582 (FOR_EACH_MODE_FROM): Likewise.
19583 (FOR_EACH_MODE_UNTIL): Likewise.
19584 (FOR_EACH_WIDER_MODE): Likewise.
19585 (FOR_EACH_2XWIDER_MODE): Likewise.
19586 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19587 * combine.c (simplify_comparison): Likewise
19588 * config/i386/i386.c (type_natural_mode): Likewise.
19589 * cse.c (cse_insn): Likewise.
19590 * dse.c (find_shift_sequence): Likewise.
19591 * emit-rtl.c (init_derived_machine_modes): Likewise.
19592 (init_emit_once): Likewise.
19593 * explow.c (hard_function_value): Likewise.
19594 * expmed.c (extract_fixed_bit_field_1): Likewise.
19595 (extract_bit_field_1): Likewise.
19596 (expand_divmod): Likewise.
19597 (emit_store_flag_1): Likewise.
19598 * expr.c (init_expr_target): Likewise.
19599 (convert_move): Likewise.
19600 (alignment_for_piecewise_move): Likewise.
19601 (widest_int_mode_for_size): Likewise.
19602 (emit_block_move_via_movmem): Likewise.
19603 (copy_blkmode_to_reg): Likewise.
19604 (set_storage_via_setmem): Likewise.
19605 (compress_float_constant): Likewise.
19606 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19607 * optabs-query.c (get_best_extraction_insn): Likewise.
19608 * optabs.c (expand_binop): Likewise.
19609 (expand_twoval_unop): Likewise.
19610 (expand_twoval_binop): Likewise.
19611 (widen_leading): Likewise.
19612 (widen_bswap): Likewise.
19613 (expand_parity): Likewise.
19614 (expand_unop): Likewise.
19615 (prepare_cmp_insn): Likewise.
19616 (prepare_float_lib_cmp): Likewise.
19617 (expand_float): Likewise.
19618 (expand_fix): Likewise.
19619 (expand_sfix_optab): Likewise.
19620 * postreload.c (move2add_use_add2_insn): Likewise.
19621 * reg-stack.c (reg_to_stack): Likewise.
19622 * reginfo.c (choose_hard_reg_mode): Likewise.
19623 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19624 * stor-layout.c (mode_for_size): Likewise.
19625 (smallest_mode_for_size): Likewise.
19626 (mode_for_vector): Likewise.
19627 (finish_bitfield_representative): Likewise.
19628 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19629 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19630 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19631 * var-tracking.c (prepare_call_arguments): Likewise.
19633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19634 Alan Hayward <alan.hayward@arm.com>
19635 David Sherwood <david.sherwood@arm.com>
19637 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19638 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19639 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19640 * machmode.h (mode_size): Move earlier in file.
19641 (mode_precision): Likewise.
19642 (mode_inner): Likewise.
19643 (mode_nunits): Likewise.
19644 (mode_unit_size): Likewise.
19645 (unit_unit_precision): Likewise.
19646 (mode_wider): Likewise.
19647 (mode_2xwider): Likewise.
19648 (machine_mode): New class.
19649 (mode_to_bytes): New function.
19650 (mode_to_bits): Likewise.
19651 (mode_to_precision): Likewise.
19652 (mode_to_inner): Likewise.
19653 (mode_to_unit_size): Likewise.
19654 (mode_to_unit_precision): Likewise.
19655 (mode_to_nunits): Likewise.
19656 (GET_MODE_SIZE): Use mode_to_bytes.
19657 (GET_MODE_BITSIZE): Use mode_to_bits.
19658 (GET_MODE_PRECISION): Use mode_to_precision.
19659 (GET_MODE_INNER): Use mode_to_inner.
19660 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19661 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19662 (GET_MODE_NUNITS): Use mode_to_nunits.
19663 * system.h (ALWAYS_INLINE): New macro.
19664 * config/powerpcspe/powerpcspe-c.c
19665 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19666 int for arg1_mode and arg2_mode.
19668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19669 Alan Hayward <alan.hayward@arm.com>
19670 David Sherwood <david.sherwood@arm.com>
19672 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19673 Prefix mode names with E_ in case statements.
19674 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19675 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19676 (aarch64_split_simd_move): Likewise.
19677 (aarch64_gen_storewb_pair): Likewise.
19678 (aarch64_gen_loadwb_pair): Likewise.
19679 (aarch64_gen_store_pair): Likewise.
19680 (aarch64_gen_load_pair): Likewise.
19681 (aarch64_get_condition_code_1): Likewise.
19682 (aarch64_constant_pool_reload_icode): Likewise.
19683 (get_rsqrte_type): Likewise.
19684 (get_rsqrts_type): Likewise.
19685 (get_recpe_type): Likewise.
19686 (get_recps_type): Likewise.
19687 (aarch64_gimplify_va_arg_expr): Likewise.
19688 (aarch64_simd_container_mode): Likewise.
19689 (aarch64_emit_load_exclusive): Likewise.
19690 (aarch64_emit_store_exclusive): Likewise.
19691 (aarch64_expand_compare_and_swap): Likewise.
19692 (aarch64_gen_atomic_cas): Likewise.
19693 (aarch64_emit_bic): Likewise.
19694 (aarch64_emit_atomic_swap): Likewise.
19695 (aarch64_emit_atomic_load_op): Likewise.
19696 (aarch64_evpc_trn): Likewise.
19697 (aarch64_evpc_uzp): Likewise.
19698 (aarch64_evpc_zip): Likewise.
19699 (aarch64_evpc_ext): Likewise.
19700 (aarch64_evpc_rev): Likewise.
19701 (aarch64_evpc_dup): Likewise.
19702 (aarch64_gen_ccmp_first): Likewise.
19703 (aarch64_gen_ccmp_next): Likewise.
19704 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19705 (alpha_emit_xfloating_libcall): Likewise.
19706 (emit_insxl): Likewise.
19707 (alpha_arg_type): Likewise.
19708 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19709 (arc_preferred_simd_mode): Likewise.
19710 (arc_secondary_reload): Likewise.
19711 (get_arc_condition_code): Likewise.
19712 (arc_print_operand): Likewise.
19713 (arc_legitimate_constant_p): Likewise.
19714 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19715 * config/arc/arc.md (casesi_load): Likewise.
19716 (casesi_compact_jump): Likewise.
19717 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19718 (cc_use_register): Likewise.
19719 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19720 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19721 (arm_init_iwmmxt_builtins): Likewise.
19722 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19723 (neon_expand_vector_init): Likewise.
19724 (arm_attr_length_move_neon): Likewise.
19725 (maybe_get_arm_condition_code): Likewise.
19726 (arm_emit_vector_const): Likewise.
19727 (arm_preferred_simd_mode): Likewise.
19728 (arm_output_iwmmxt_tinsr): Likewise.
19729 (thumb1_output_casesi): Likewise.
19730 (thumb2_output_casesi): Likewise.
19731 (arm_emit_load_exclusive): Likewise.
19732 (arm_emit_store_exclusive): Likewise.
19733 (arm_expand_compare_and_swap): Likewise.
19734 (arm_evpc_neon_vuzp): Likewise.
19735 (arm_evpc_neon_vzip): Likewise.
19736 (arm_evpc_neon_vrev): Likewise.
19737 (arm_evpc_neon_vtrn): Likewise.
19738 (arm_evpc_neon_vext): Likewise.
19739 (arm_validize_comparison): Likewise.
19740 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19741 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19742 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19743 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19744 (c6x_preferred_simd_mode): Likewise.
19745 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19746 (epiphany_rtx_costs): Likewise.
19747 * config/epiphany/predicates.md (proper_comparison_operator):
19749 * config/frv/frv.c (condexec_memory_operand): Likewise.
19750 (frv_emit_move): Likewise.
19751 (output_move_single): Likewise.
19752 (output_condmove_single): Likewise.
19753 (frv_hard_regno_mode_ok): Likewise.
19754 (frv_matching_accg_mode): Likewise.
19755 * config/h8300/h8300.c (split_adds_subs): Likewise.
19756 (h8300_rtx_costs): Likewise.
19757 (h8300_print_operand): Likewise.
19758 (compute_mov_length): Likewise.
19759 (output_logical_op): Likewise.
19760 (compute_logical_op_length): Likewise.
19761 (compute_logical_op_cc): Likewise.
19762 (h8300_shift_needs_scratch_p): Likewise.
19763 (output_a_shift): Likewise.
19764 (compute_a_shift_length): Likewise.
19765 (compute_a_shift_cc): Likewise.
19766 (expand_a_rotate): Likewise.
19767 (output_a_rotate): Likewise.
19768 * config/i386/i386.c (classify_argument): Likewise.
19769 (function_arg_advance_32): Likewise.
19770 (function_arg_32): Likewise.
19771 (function_arg_64): Likewise.
19772 (function_value_64): Likewise.
19773 (ix86_gimplify_va_arg): Likewise.
19774 (ix86_legitimate_constant_p): Likewise.
19775 (put_condition_code): Likewise.
19776 (split_double_mode): Likewise.
19777 (ix86_avx256_split_vector_move_misalign): Likewise.
19778 (ix86_expand_vector_logical_operator): Likewise.
19779 (ix86_split_idivmod): Likewise.
19780 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19781 (ix86_build_const_vector): Likewise.
19782 (ix86_build_signbit_mask): Likewise.
19783 (ix86_match_ccmode): Likewise.
19784 (ix86_cc_modes_compatible): Likewise.
19785 (ix86_expand_branch): Likewise.
19786 (ix86_expand_sse_cmp): Likewise.
19787 (ix86_expand_sse_movcc): Likewise.
19788 (ix86_expand_int_sse_cmp): Likewise.
19789 (ix86_expand_vec_perm_vpermi2): Likewise.
19790 (ix86_expand_vec_perm): Likewise.
19791 (ix86_expand_sse_unpack): Likewise.
19792 (ix86_expand_int_addcc): Likewise.
19793 (ix86_split_to_parts): Likewise.
19794 (ix86_vectorize_builtin_gather): Likewise.
19795 (ix86_vectorize_builtin_scatter): Likewise.
19796 (avx_vpermilp_parallel): Likewise.
19797 (inline_memory_move_cost): Likewise.
19798 (ix86_tieable_integer_mode_p): Likewise.
19799 (x86_maybe_negate_const_int): Likewise.
19800 (ix86_expand_vector_init_duplicate): Likewise.
19801 (ix86_expand_vector_init_one_nonzero): Likewise.
19802 (ix86_expand_vector_init_one_var): Likewise.
19803 (ix86_expand_vector_init_concat): Likewise.
19804 (ix86_expand_vector_init_interleave): Likewise.
19805 (ix86_expand_vector_init_general): Likewise.
19806 (ix86_expand_vector_set): Likewise.
19807 (ix86_expand_vector_extract): Likewise.
19808 (emit_reduc_half): Likewise.
19809 (ix86_emit_i387_round): Likewise.
19810 (ix86_mangle_type): Likewise.
19811 (ix86_expand_round_sse4): Likewise.
19812 (expand_vec_perm_blend): Likewise.
19813 (canonicalize_vector_int_perm): Likewise.
19814 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
19815 (expand_vec_perm_1): Likewise.
19816 (expand_vec_perm_interleave3): Likewise.
19817 (expand_vec_perm_even_odd_pack): Likewise.
19818 (expand_vec_perm_even_odd_1): Likewise.
19819 (expand_vec_perm_broadcast_1): Likewise.
19820 (ix86_vectorize_vec_perm_const_ok): Likewise.
19821 (ix86_expand_vecop_qihi): Likewise.
19822 (ix86_expand_mul_widen_hilo): Likewise.
19823 (ix86_expand_sse2_abs): Likewise.
19824 (ix86_expand_pextr): Likewise.
19825 (ix86_expand_pinsr): Likewise.
19826 (ix86_preferred_simd_mode): Likewise.
19827 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19828 * config/i386/sse.md (*andnot<mode>3): Likewise.
19829 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19830 (*<code><mode>3): Likewise.
19831 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19832 (ia64_expand_atomic_op): Likewise.
19833 (ia64_arg_type): Likewise.
19834 (ia64_mode_to_int): Likewise.
19835 (ia64_scalar_mode_supported_p): Likewise.
19836 (ia64_vector_mode_supported_p): Likewise.
19837 (expand_vec_perm_broadcast): Likewise.
19838 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19839 (iq2000_function_arg_advance): Likewise.
19840 (iq2000_function_arg): Likewise.
19841 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19842 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19843 (m68k_libcall_value): Likewise.
19844 (m68k_function_value): Likewise.
19845 (sched_attr_op_type): Likewise.
19846 * config/mcore/mcore.c (mcore_output_move): Likewise.
19847 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19849 (microblaze_function_arg): Likewise.
19850 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19851 (mips_print_operand): Likewise.
19852 (mips_mode_ok_for_mov_fmt_p): Likewise.
19853 (mips_vector_mode_supported_p): Likewise.
19854 (mips_preferred_simd_mode): Likewise.
19855 (mips_expand_vpc_loongson_even_odd): Likewise.
19856 (mips_expand_vec_unpack): Likewise.
19857 (mips_expand_vi_broadcast): Likewise.
19858 (mips_expand_vector_init): Likewise.
19859 (mips_expand_vec_reduc): Likewise.
19860 (mips_expand_msa_cmp): Likewise.
19861 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19862 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19863 (cc_flags_for_mode): Likewise.
19864 * config/msp430/msp430.c (msp430_print_operand): Likewise.
19865 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
19866 (nds32_output_casesi_pc_relative): Likewise.
19867 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19868 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
19869 (nvptx_gen_unpack): Likewise.
19870 (nvptx_gen_pack): Likewise.
19871 (nvptx_gen_shuffle): Likewise.
19872 (nvptx_gen_wcast): Likewise.
19873 (nvptx_preferred_simd_mode): Likewise.
19874 * config/pa/pa.c (pa_secondary_reload): Likewise.
19875 * config/pa/predicates.md (base14_operand): Likewise.
19876 * config/powerpcspe/powerpcspe-c.c
19877 (altivec_resolve_overloaded_builtin): Likewise.
19878 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
19880 (rs6000_preferred_simd_mode): Likewise.
19881 (output_vec_const_move): Likewise.
19882 (rs6000_expand_vector_extract): Likewise.
19883 (rs6000_split_vec_extract_var): Likewise.
19884 (reg_offset_addressing_ok_p): Likewise.
19885 (rs6000_legitimate_offset_address_p): Likewise.
19886 (rs6000_legitimize_address): Likewise.
19887 (rs6000_emit_set_const): Likewise.
19888 (rs6000_const_vec): Likewise.
19889 (rs6000_emit_move): Likewise.
19890 (spe_build_register_parallel): Likewise.
19891 (rs6000_darwin64_record_arg_recurse): Likewise.
19892 (swap_selector_for_mode): Likewise.
19893 (spe_init_builtins): Likewise.
19894 (paired_init_builtins): Likewise.
19895 (altivec_init_builtins): Likewise.
19896 (do_load_for_compare): Likewise.
19897 (rs6000_generate_compare): Likewise.
19898 (rs6000_expand_float128_convert): Likewise.
19899 (emit_load_locked): Likewise.
19900 (emit_store_conditional): Likewise.
19901 (rs6000_output_function_epilogue): Likewise.
19902 (rs6000_handle_altivec_attribute): Likewise.
19903 (rs6000_function_value): Likewise.
19904 (emit_fusion_gpr_load): Likewise.
19905 (emit_fusion_p9_load): Likewise.
19906 (emit_fusion_p9_store): Likewise.
19907 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19908 (fusion_gpr_mem_load): Likewise.
19909 (fusion_addis_mem_combo_load): Likewise.
19910 (fusion_addis_mem_combo_store): Likewise.
19911 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19912 (fusion_gpr_mem_load): Likewise.
19913 (fusion_addis_mem_combo_load): Likewise.
19914 (fusion_addis_mem_combo_store): Likewise.
19915 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19917 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19918 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19919 (rs6000_preferred_simd_mode): Likewise.
19920 (output_vec_const_move): Likewise.
19921 (rs6000_expand_vector_extract): Likewise.
19922 (rs6000_split_vec_extract_var): Likewise.
19923 (reg_offset_addressing_ok_p): Likewise.
19924 (rs6000_legitimate_offset_address_p): Likewise.
19925 (rs6000_legitimize_address): Likewise.
19926 (rs6000_emit_set_const): Likewise.
19927 (rs6000_const_vec): Likewise.
19928 (rs6000_emit_move): Likewise.
19929 (rs6000_darwin64_record_arg_recurse): Likewise.
19930 (swap_selector_for_mode): Likewise.
19931 (paired_init_builtins): Likewise.
19932 (altivec_init_builtins): Likewise.
19933 (rs6000_expand_float128_convert): Likewise.
19934 (emit_load_locked): Likewise.
19935 (emit_store_conditional): Likewise.
19936 (rs6000_output_function_epilogue): Likewise.
19937 (rs6000_handle_altivec_attribute): Likewise.
19938 (rs6000_function_value): Likewise.
19939 (emit_fusion_gpr_load): Likewise.
19940 (emit_fusion_p9_load): Likewise.
19941 (emit_fusion_p9_store): Likewise.
19942 * config/rx/rx.c (rx_gen_move_template): Likewise.
19943 (flags_from_mode): Likewise.
19944 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19945 (s390_slb_comparison): Likewise.
19946 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19947 (s390_vector_mode_supported_p): Likewise.
19948 (s390_cc_modes_compatible): Likewise.
19949 (s390_match_ccmode_set): Likewise.
19950 (s390_canonicalize_comparison): Likewise.
19951 (s390_emit_compare_and_swap): Likewise.
19952 (s390_branch_condition_mask): Likewise.
19953 (s390_rtx_costs): Likewise.
19954 (s390_secondary_reload): Likewise.
19955 (__SECONDARY_RELOAD_CASE): Likewise.
19956 (s390_expand_cs): Likewise.
19957 (s390_preferred_simd_mode): Likewise.
19958 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19959 * config/sh/sh.c (sh_print_operand): Likewise.
19960 (dump_table): Likewise.
19961 (sh_secondary_reload): Likewise.
19962 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19963 * config/sh/sh.md (casesi_worker_1): Likewise.
19964 (casesi_worker_2): Likewise.
19965 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19966 (fcc_comparison_operator): Likewise.
19967 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19968 (emit_soft_tfmode_cvt): Likewise.
19969 (sparc_preferred_simd_mode): Likewise.
19970 (output_cbranch): Likewise.
19971 (sparc_print_operand): Likewise.
19972 (sparc_expand_vec_perm_bmask): Likewise.
19973 (vector_init_bshuffle): Likewise.
19974 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19975 (spu_vector_mode_supported_p): Likewise.
19976 (spu_expand_insv): Likewise.
19977 (spu_emit_branch_or_set): Likewise.
19978 (spu_handle_vector_attribute): Likewise.
19979 (spu_builtin_splats): Likewise.
19980 (spu_builtin_extract): Likewise.
19981 (spu_builtin_promote): Likewise.
19982 (spu_expand_sign_extend): Likewise.
19983 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19984 (tilegx_simd_int): Likewise.
19985 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19986 (tilepro_simd_int): Likewise.
19987 * config/v850/v850.c (const_double_split): Likewise.
19988 (v850_print_operand): Likewise.
19989 (ep_memory_offset): Likewise.
19990 * config/vax/vax.c (vax_rtx_costs): Likewise.
19991 (vax_output_int_move): Likewise.
19992 (vax_output_int_add): Likewise.
19993 (vax_output_int_subtract): Likewise.
19994 * config/visium/predicates.md (visium_branch_operator): Likewise.
19995 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19996 (visium_print_operand_address): Likewise.
19997 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19998 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19999 (xtensa_expand_conditional_branch): Likewise.
20000 (xtensa_copy_incoming_a7): Likewise.
20001 (xtensa_output_literal): Likewise.
20002 * dfp.c (decimal_real_maxval): Likewise.
20003 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
20005 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20006 Alan Hayward <alan.hayward@arm.com>
20007 David Sherwood <david.sherwood@arm.com>
20009 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20010 (mode_nunits_inline): Likewise.
20011 (mode_inner_inline): Likewise.
20012 (mode_unit_size_inline): Likewise.
20013 (mode_unit_precision_inline): Likewise.
20014 (emit_insn_modes_h): Likewise. Also emit a #define of the
20016 (emit_mode_wider): Add an E_ prefix to mode names.
20017 (emit_mode_complex): Likewise.
20018 (emit_mode_inner): Likewise.
20019 (emit_mode_adjustments): Likewise.
20020 (emit_mode_int_n): Likewise.
20021 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20022 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20023 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20024 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20025 (CRC32_BUILTIN, ENTRY): Likewise.
20026 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20027 (aarch64_pop_regs): Likewise.
20028 (aarch64_process_components): Likewise.
20029 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20030 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20031 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20032 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20033 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20034 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20036 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20037 (pcmpistr): Likewise.
20038 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20039 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20040 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20042 * config/rl78/rl78.c (mduc_regs): Likewise.
20043 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20044 (htm_expand_builtin): Likewise.
20045 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20046 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20047 * config/xtensa/xtensa.c (print_operand): Likewise.
20048 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20049 (NUM_MODE_VECTOR_INT): Likewise.
20050 * genoutput.c (null_operand): Likewise.
20051 (output_operand_data): Likewise.
20052 * genrecog.c (print_parameter_value): Likewise.
20053 * lra.c (debug_operand_data): Likewise.
20055 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20057 * dwarf2out.c (scompare_loc_descriptor_wide)
20058 (scompare_loc_descriptor_narrow): New functions, split out from...
20059 (scompare_loc_descriptor): ...here.
20060 * expmed.c (emit_store_flag_int): New function, split out from...
20061 (emit_store_flag): ...here.
20063 2017-08-30 Richard Biener <rguenther@suse.de>
20065 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20066 (dwarf2out_early_finish): Move setting of AT_pubnames from
20067 early debug output to early finish.
20069 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20071 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20072 and -mdata-region to the assembler.
20074 2017-08-30 Richard Biener <rguenther@suse.de>
20076 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20078 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20079 (dwarf2out_early_global_decl): Only generate a DIE for the
20080 abstract origin if it doesn't already exist or is a declaration DIE.
20081 (resolve_addr): Do not add the linkage name twice when
20082 generating a stub DIE for the DW_TAG_GNU_call_site target.
20084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20086 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20087 Use machine_mode rather than int for arg1_mode.
20089 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20092 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20093 that the second argument of the built-in functions to unpack
20094 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20095 switch statement instead a lot of if statements.
20096 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20097 Allow 64-bit values to be in Altivec registers as well as
20098 traditional floating point registers.
20099 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20101 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20103 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20104 MAX_REGS_PER_ADDRESS == 1.
20106 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20108 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20109 * config/i386/i386.c (ix86_option_override_internal): Simplify
20110 setting of opts->x_flag_entry.
20112 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20113 Jakub Jelinek <jakub@redhat.com>
20114 Richard Biener <rguenther@suse.de>
20116 PR tree-optimization/81503
20117 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20118 folded constant fits in the target type; reorder tests for clarity.
20120 2017-08-29 Martin Liska <mliska@suse.cz>
20122 * passes.def: Include pass_lower_switch.
20123 * stmt.c (dump_case_nodes): Remove and move to
20124 tree-switch-conversion.
20125 (case_values_threshold): Likewise.
20126 (expand_switch_as_decision_tree_p): Likewise.
20127 (emit_case_decision_tree): Likewise.
20128 (expand_case): Likewise.
20129 (balance_case_nodes): Likewise.
20130 (node_has_low_bound): Likewise.
20131 (node_has_high_bound): Likewise.
20132 (node_is_bounded): Likewise.
20133 (emit_case_nodes): Likewise.
20134 (struct simple_case_node): New struct.
20135 (add_case_node): Remove.
20136 (emit_case_dispatch_table): Use vector instead of case_list.
20137 (reset_out_edges_aux): Remove.
20138 (compute_cases_per_edge): Likewise.
20139 (expand_case): Build list of simple_case_node.
20140 (expand_sjlj_dispatch_table): Use it.
20141 * tree-switch-conversion.c (struct case_node): Moved from
20142 stmt.c and adjusted.
20143 (emit_case_nodes): Likewise.
20144 (node_has_low_bound): Likewise.
20145 (node_has_high_bound): Likewise.
20146 (node_is_bounded): Likewise.
20147 (case_values_threshold): Likewise.
20148 (reset_out_edges_aux): Likewise.
20149 (compute_cases_per_edge): Likewise.
20150 (add_case_node): Likewise.
20151 (dump_case_nodes): Likewise.
20152 (balance_case_nodes): Likewise.
20153 (expand_switch_as_decision_tree_p): Likewise.
20154 (emit_jump): Likewise.
20155 (emit_case_decision_tree): Likewise.
20156 (try_switch_expansion): Likewise.
20157 (do_jump_if_equal): Likewise.
20158 (emit_cmp_and_jump_insns): Likewise.
20159 (fix_phi_operands_for_edge): New function.
20160 (record_phi_operand_mapping): Likewise.
20161 (class pass_lower_switch): New pass.
20162 (pass_lower_switch::execute): New function.
20163 (make_pass_lower_switch): Likewise.
20164 (conditional_probability):
20165 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20166 * tree-pass.h: Add make_pass_lower_switch.
20168 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20171 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20174 2017-08-29 Richard Biener <rguenther@suse.de>
20176 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20177 we do not add a DW_AT_inline attribute twice.
20178 (gen_subprogram_die): Remove code setting DW_AT_inline on
20179 DECL_ABSTRACT_P nodes.
20181 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20183 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20184 calls to internal functions.
20185 (gimplify_modify_expr): Likewise.
20186 * tree-call-cdce.c (use_internal_fn): Likewise.
20187 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20188 (convert_to_divmod): Set the nothrow flag.
20189 * tree-if-conv.c (predicate_mem_writes): Likewise.
20190 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20191 (vectorizable_call): Likewise.
20192 (vectorizable_store): Likewise.
20193 (vectorizable_load): Likewise.
20194 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20195 (vect_recog_mask_conversion_pattern): Likewise.
20197 2017-08-29 Martin Liska <mliska@suse.cz>
20200 * gcc.c (driver_handle_option): Add new argument.
20201 * opts-common.c (handle_option): Pass
20202 target_option_override_hook.
20203 * opts-global.c (lang_handle_option): Add new option.
20204 (set_default_handlers): Add new argument.
20205 (decode_options): Likewise.
20206 * opts.c (target_handle_option): Likewise.
20207 (common_handle_option): Call target_option_override_hook.
20208 * opts.h (struct cl_option_handler_func): Add hook for
20209 target option override.
20210 (struct cl_option_handlers): Likewise.
20211 (set_default_handlers): Add new argument.
20212 (decode_options): Likewise.
20213 (common_handle_option): Likewise.
20214 (target_handle_option): Likewise.
20215 * toplev.c (toplev::main): Pass targetm.target_option.override
20218 2017-08-29 Richard Biener <rguenther@suse.de>
20219 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20221 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20222 life to the active subtree.
20224 2017-08-28 Jeff Law <law@redhat.com>
20226 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20227 derive_equivalences.
20228 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20229 Code moved into....
20230 (edge_info::derive_equivalences): New private member function
20232 * tree-ssa-dom.c (class edge_info): Changed from a struct
20233 to a class. Add ctor/dtor, methods and data members.
20234 (edge_info::edge_info): Renamed from allocate_edge_info.
20235 Initialize additional members.
20236 (edge_info::~edge_info): New.
20237 (free_dom_edge_info): Delete the edge info.
20238 (record_edge_info): Use new class & associated member functions.
20239 Tighten forms for testing for edge equivalences.
20240 (record_temporary_equivalences): Iterate over the simple
20241 equivalences rather than assuming there's only one per edge.
20242 (cprop_into_successor_phis): Iterate over the simple
20243 equivalences rather than assuming there's only one per edge.
20244 (optimize_stmt): Use operand_equal_p rather than pointer
20245 equality for mini-DSE code.
20247 2017-08-28 Nathan Sidwell <nathan@acm.org>
20249 * gcc.c (execute): Fold SIGPIPE handling into switch
20250 statement. Adjust internal error message.
20252 2017-08-28 Richard Biener <rguenther@suse.de>
20255 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20256 Do nothing for removed DIEs.
20258 2017-08-28 Richard Biener <rguenther@suse.de>
20260 PR tree-optimization/81977
20261 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20264 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20267 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20268 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20269 using targetm.gen_mem_thread_fence.
20271 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20274 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20275 predicate to register_operand. Reorder operands.
20276 (*btr<mode>): Ditto.
20277 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20278 (*btr<mode>_mask): Ditto.
20280 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20282 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20283 * config/rs6000/xmmintrin.h: New file.
20284 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20286 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20289 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20290 parameter and_insn and return it.
20291 (recombine_lvx_pattern): Insert a copy to ensure availability of
20292 the base register of the copied masking operation at the point of
20293 the instruction replacement.
20294 (recombine_stvx_pattern): Likewise.
20296 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20298 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20299 undocumented switches.
20300 (-mpower9-dform-vector): Likewise.
20301 (-mpower9-dform): Likewise.
20302 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20303 comments to delete references to -mpower9-dform* switches.
20304 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20305 Delete reference to -mpower9-dform* switches, test for
20306 -mpower9-vector instead.
20307 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20308 (OTHER_P9_VECTOR_MASKS): Likewise.
20309 (POWERPC_MASKS): Likewise.
20310 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20311 tests against -mpower9-dform* to -mpower9-vector. Delete code
20312 that checked for -mpower9-dform* consistancy with other options.
20313 Add test for -mpower9-misc to enable other power9 switches.
20314 (rs6000_init_hard_regno_mode_ok): Likewise.
20315 (rs6000_option_override_internal): Likewise.
20316 (rs6000_emit_prologue): Likewise.
20317 (rs6000_emit_epilogue): Likewise.
20318 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20319 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20320 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20322 (emit_fusion_p9_store): Likewise.
20323 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20324 resetting these macros if the assembler does not support ISA 3.0
20326 (TARGET_P9_DFORM_VECTOR): Likewise.
20327 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20328 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20330 2017-08-25 Alan Modra <amodra@gmail.com>
20332 PR rtl-optimization/81747
20333 * cse.c (cse_extended_basic_block): Don't attempt to record
20334 equivalences for degenerate conditional branches that branch
20335 to their fall-through.
20337 2017-08-24 Martin Sebor <msebor@redhat.com>
20339 PR middle-end/81908
20340 * gimple-fold.c (size_must_be_zero_p): New function.
20341 (gimple_fold_builtin_memory_op): Call it.
20343 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20345 * config/rs6000/mm_malloc.h: New file.
20347 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20349 PR tree-optimization/81913
20350 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20351 analysis when either IVs in condition can wrap.
20353 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20355 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20356 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20358 2017-08-24 Richard Biener <rguenther@suse.de>
20361 * targhooks.c (default_target_can_inline_p): Properly
20362 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20363 is present and always compare.
20364 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20365 infer -mfpmath=sse from TARGET_SSE_P.
20366 (ix86_can_inline_p): Properly use target_option_default_node when
20367 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20369 2017-08-24 Richard Biener <rguenther@suse.de>
20372 * dwarf2out.c (output_die): Handle flag_generate_offload like
20374 (output_comp_unit): Likewise.
20375 (gen_array_type_die): Likewise.
20376 (dwarf2out_early_finish): Likewise.
20377 (note_variable_value_in_expr): Likewise.
20378 (dwarf2out_finish): Likewise. Adjust assert.
20379 * cgraphunit.c (symbol_table::compile): Move setting of
20380 flag_generate_offload earlier ...
20381 (symbol_table::finalize_compilation_unit): ... here, before
20382 early debug finalization.
20384 2017-08-24 Richard Biener <rguenther@suse.de>
20386 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20387 and ipa-fnsummary.h.
20388 (ix86_can_inline_p): When ix86_fpmath flags do not match
20389 check whether the callee uses FP math at all.
20391 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20393 PR middle-end/81931
20394 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20395 instead of TYPE_PRECISION.
20397 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20398 Alan Hayward <alan.hayward@arm.com>
20399 David Sherwood <david.sherwood@arm.com>
20401 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20403 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20405 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20407 (ix86_frame::stack_realign_allocate): New field.
20408 (struct machine_frame_state): Modify comments.
20409 (machine_frame_state::sp_realigned_fp_end): New field.
20410 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20411 layout calculation.
20412 (sp_valid_at): Add assertion to assure no attempt to access invalid
20413 offset of a realigned stack.
20414 (fp_valid_at): Likewise.
20415 (choose_baseaddr): Modify comments.
20416 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20417 ix86_expand_prologue.
20418 (ix86_expand_prologue): Modify stack realignment and allocation.
20419 (ix86_expand_epilogue): Modify comments.
20420 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20421 avx2_runtime, avx512f, and avx512f_runtime.
20423 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20425 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20426 (mstackrealign): Do not init to -1.
20427 * config/i386/i386.c (ix86_option_override_internal):
20428 Check opts_set, not opts when setting default value of
20429 opts->x_ix86_force_align_arg_pointer.
20431 2017-08-23 Richard Biener <rguenther@suse.de>
20433 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20434 lang_hooks.decl_printable_name.
20435 * print-rtl-function.c (print_rtx_function): Likewise.
20436 * tree-pretty-print.c (dump_function_header): Likewise.
20438 2017-08-23 Richard Biener <rguenther@suse.de>
20441 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20442 -g0 at compile-time.
20444 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20446 PR middle-end/19706
20447 * doc/sourcebuild.texi (Other hardware attributes):
20450 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20452 PR middle-end/19706
20453 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20454 Fix single-use check.
20456 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20458 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20460 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20462 * doc/install.texi: Modify to add more details on running selected
20465 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20467 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20468 is combined with -mabi=ms.
20469 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20472 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20474 PR tree-optimization/81488
20475 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20476 and cached_basis fields.
20477 (MAX_SPREAD): New constant.
20478 (alloc_cand_and_find_basis): Initialize new fields.
20479 (clear_visited): New function.
20480 (create_phi_basis_1): Rename from create_phi_basis, set visited
20481 and cached_basis fields.
20482 (create_phi_basis): New wrapper function.
20483 (phi_add_costs_1): Rename from phi_add_costs, add spread
20484 parameter, set visited field, short-circuit when limits reached.
20485 (phi_add_costs): New wrapper function.
20486 (record_phi_increments_1): Rename from record_phi_increments, set
20488 (record_phi_increments): New wrapper function.
20489 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20490 (phi_incr_cost): New wrapper function.
20491 (all_phi_incrs_profitable_1): Rename from
20492 all_phi_incrs_profitable, set visited field.
20493 (all_phi_incrs_profitable): New wrapper function.
20495 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20496 Alan Hayward <alan.hayward@arm.com>
20497 David Sherwood <david.sherwood@arm.com>
20499 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20500 that takes the outer and inner modes.
20501 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20502 comparison as the canonical test for a paradoxical subreg.
20503 * combine.c (simplify_set): Use paradoxical_subreg_p.
20504 (make_extraction): Likewise.
20505 (force_to_mode): Likewise.
20506 (rtx_equal_for_field_assignment_p): Likewise.
20507 (gen_lowpart_for_combine): Likewise.
20508 (simplify_comparison): Likewise.
20509 * cse.c (equiv_constant): Likewise.
20510 * expmed.c (store_bit_field_1): Likewise.
20511 * final.c (alter_subreg): Likewise.
20512 * fwprop.c (propagate_rtx): Likewise.
20513 (forward_propagate_subreg): Likewise.
20514 * ira-conflicts.c (ira_build_conflicts): Likewise.
20515 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20516 * lra-constraints.c (curr_insn_transform): Likewise.
20517 (split_reg): Likewise.
20518 * lra-eliminations.c (move_plus_up): Likewise.
20519 (lra_eliminate_regs_1): Likewise.
20520 * recog.c (general_operand): Likewise.
20521 * ree.c (combine_reaching_defs): Likewise.
20522 * reload.c (push_reload): Likewise.
20523 (find_reloads): Likewise.
20524 * reload1.c (elimination_effects): Likewise.
20525 (compute_reload_subreg_offset): Likewise.
20526 (choose_reload_regs): Likewise.
20527 * rtlanal.c (subreg_lsb_1): Likewise.
20528 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20529 (simplify_subreg): Likewise.
20530 * var-tracking.c (track_loc_p): Likewise.
20531 * emit-rtl.c (byte_lowpart_offset): Likewise.
20532 (paradoxical_subreg_p): Delete out-of-line definition.
20534 2017-08-22 Jeff Law <law@redhat.com>
20536 PR tree-optimization/81741
20537 PR tree-optimization/71947
20538 * tree-ssa-dom.c: Include tree-inline.h.
20539 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20540 equivalences if one is more expensive to compute than the other.
20541 * tree-ssa-scopedtables.h (class const_or_copies): Make
20542 record_const_or_copy_raw method private.
20543 (class avail_exprs_stack): New method simplify_binary_operation.
20544 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20545 avail_exprs_stack::simplify_binary_operation as needed.
20546 (avail_exprs_stack::simplify_binary_operation): New function.
20548 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20550 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20551 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20552 (DOT_SYMBOLS): Likewise.
20553 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20554 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20555 (RS6000_ABI_NAME): Likewise.
20556 (TARGET_CMODEL): Likewise.
20557 (TOC_SECTION_ASM_OP): Likewise.
20558 (SET_CMODEL): New macro.
20559 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20561 2017-08-22 Richard Biener <rguenther@suse.de>
20563 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20564 to allow for free-lang-data replacements similar to verify_type_variant.
20566 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20567 Alan Hayward <alan.hayward@arm.com>
20568 David Sherwood <david.sherwood@arm.com>
20570 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20571 VOIDmode for the LABEL_REF.
20573 2017-08-22 Richard Biener <rguenther@suse.de>
20575 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20577 2017-08-22 Richard Biener <rguenther@suse.de>
20579 * common.opt (feliminate-dwarf2-dups): Ignore.
20580 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20581 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20582 same_die_p_wrap, compute_section_prefix,
20583 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20584 (comdat_symbol_id, comdat_symbol_number): Likewise.
20585 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20587 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20588 (output_die): Mark unreachable path unreachable.
20589 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20590 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20591 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20592 (dwarf2out_early_finish): Likewise.
20594 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20596 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20598 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20601 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20602 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20603 (avr_attribute_table) <io, io_low, address>: Initialize
20604 .decl_required with true.
20606 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20608 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20609 undocumented debugging options.
20610 (-mvsx-scalar-double): Likewise.
20611 (-mallow-df-permute): Likewise.
20612 (-mvectorize-builtins): Likewise.
20613 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20614 (rs6000_builtin_vectorized_function): Likewise.
20615 (rs6000_builtin_md_vectorized_function): Likewise.
20616 (rs6000_opt_vars): Likewise.
20618 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20621 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20622 (*btrq_imm): Rename from *btrq.
20623 (*btcq_imm): Rename from *btcq.
20624 (btsc): New code attribute.
20625 (*<btsc><mode>): New insn pattern.
20626 (*btr<mode>): Ditto.
20627 (*<btsc><mode>_mask): New insn_and_split pattern.
20628 (*btr<mode>_mask): Ditto.
20630 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20631 Alan Hayward <alan.hayward@arm.com>
20632 David Sherwood <david.sherwood@arm.com>
20634 * function.c (pad_below): Simplify padding calculation.
20636 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20637 Alan Hayward <alan.hayward@arm.com>
20638 David Sherwood <david.sherwood@arm.com>
20640 * target.def (function_prologue): Remove frame size argument.
20641 (function_epilogue): Likewise.
20642 * doc/tm.texi: Regenerate.
20643 * final.c (final_start_function): Update call to function_prologue.
20644 (final_end_function): Update call to function_epilogue.
20645 (default_function_pro_epilogue): Remove frame size argument.
20646 * output.h (default_function_pro_epilogue): Likewise.
20647 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20648 (arm_output_function_prologue): Likewise.
20649 * config/frv/frv.c (frv_function_prologue): Likewise.
20650 (frv_function_epilogue): Likewise.
20651 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20652 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20653 (ia64_output_function_epilogue): Likewise.
20654 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20655 (m32r_output_function_epilogue): Likewise.
20656 * config/microblaze/microblaze.c (microblaze_function_prologue)
20657 (microblaze_function_epilogue): Likewise.
20658 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20659 (mips_output_function_epilogue): Likewise.
20660 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20661 (mmix_target_asm_function_epilogue): Likewise.
20662 * config/msp430/msp430.c (msp430_start_function): Likewise.
20663 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20664 (nds32_asm_function_epilogue): Likewise.
20665 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20666 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20667 (pa_output_function_epilogue): Likewise.
20668 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20669 (rs6000_output_function_epilogue): Likewise.
20670 * config/rl78/rl78.c (rl78_start_function): Likewise.
20671 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20672 (rs6000_output_function_epilogue): Likewise.
20673 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20674 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20675 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20676 (sparc_asm_function_epilogue): Likewise.
20678 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20680 * tree.h (type_has_mode_precision_p): New function.
20681 * convert.c (convert_to_integer_1): Use it.
20682 * expr.c (expand_expr_real_2): Likewise.
20683 (expand_expr_real_1): Likewise.
20684 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20685 * match.pd: Likewise.
20686 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20687 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20688 * tree-tailcall.c (process_assignment): Likewise.
20689 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20690 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20691 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20692 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20693 (vectorizable_assignment): Likewise.
20694 (vectorizable_shift): Likewise.
20695 (vectorizable_operation): Likewise.
20696 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20698 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20700 * match.pd: Add pow (C, x) simplification.
20702 2017-08-21 Richard Biener <rguenther@suse.de>
20704 PR tree-optimization/81900
20705 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20706 for blocks with abnormal predecessors.
20707 (compute_antic): Do not set visited flag prematurely.
20709 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20712 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20714 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20716 * stor-layout.h (vector_type_mode): Move to...
20717 * tree.h (vector_type_mode): ...here.
20718 * stor-layout.c (vector_type_mode): Move to...
20719 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20721 2017-08-21 Richard Biener <rguenther@suse.de>
20723 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20724 register_external_die hooks.
20725 (debug_false_tree_charstarstar_uhwistar): Declare.
20726 (debug_nothing_tree_charstar_uhwi): Likewise.
20727 * debug.c (do_nothing_debug_hooks): Adjust.
20728 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20729 (debug_nothing_tree_charstar_uhwi): Likewise.
20730 * dbxout.c (dbx_debug_hooks): Adjust.
20731 (xcoff_debug_hooks): Likewise.
20732 * sdbout.c (sdb_debug_hooks): Likewise.
20733 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20734 * dwarf2out.c (macinfo_label_base): New global.
20735 (dwarf2out_register_external_die): New function for the
20736 register_external_die hook.
20737 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20738 (dwarf2_debug_hooks): Use them.
20739 (dwarf2_lineno_debug_hooks): Adjust.
20740 (struct die_struct): Add with_offset flag.
20741 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20742 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20743 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20744 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20745 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20746 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20747 defining section names for the early LTO debug variants.
20748 (reset_indirect_string): New helper.
20749 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20750 (print_dw_val): Add support for offsetted symbol references.
20751 (get_ultimate_context): Split out from is_cxx.
20752 (is_cxx): Use get_ultimate_context.
20753 (is_fortran): Add decl overload.
20754 (compute_comp_unit_symbol): Split out worker from
20755 compute_section_prefix.
20756 (compute_section_prefix): Call compute_comp_unit_symbol and
20757 set comdat_type_p here.
20758 (output_die): Skip DIE symbol output for the LTO added one.
20759 Handle DIE symbol references with offset.
20760 (output_comp_unit): Guard section name mangling properly.
20761 For LTO debug sections emit a symbol at the section beginning
20762 which we use to refer to its DIEs.
20763 (add_abstract_origin_attribute): For DIEs registered via
20764 dwarf2out_register_external_die directly refer to the early
20765 DIE rather than indirectly through the shadow one we created.
20766 Remove obsolete call to dwarf2out_abstract_function for
20767 non-function/block origins.
20768 (gen_array_type_die): When generating early LTO debug do
20769 not emit DW_AT_string_length.
20770 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20771 late when in LTO. As suggested place a gcc_unreachable for
20772 the DECL_ABSTRACT_P case.
20773 (gen_subprogram_die): Avoid another specification DIE
20774 for early built declarations/definitions for the late LTO case.
20775 (gen_variable_die): Add type references for late duplicated VLA dies
20777 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20778 we have the abstract instance already.
20779 (process_scope_var): Adjust decl DIE contexts in LTO which
20780 first puts them in limbo.
20781 (gen_decl_die): Do not generate type DIEs late apart from
20782 types for VLAs or for decls we do not yet have a DIE. Do not
20783 call dwarf2out_abstract_function late.
20784 (dwarf2out_early_global_decl): Make sure to create DIEs
20785 for abstract instances of a decl first.
20786 (dwarf2out_late_global_decl): Adjust comment.
20787 (output_macinfo_op): With multiple macro sections use
20788 macinfo_label_base to distinguish labels.
20789 (output_macinfo): Likewise. Update macinfo_label_base.
20790 Pass in the line info label.
20791 (note_variable_value_in_expr): When generating LTO resolve
20792 all variable values here by generating DIEs as needed.
20793 (init_sections_and_labels): Add early LTO debug flag parameter
20794 and generate different sections and names if set. Add generation
20795 counter for the labels so we can have multiple of them.
20796 (reset_dies): Helper to allow DIEs to be output multiple times.
20797 (dwarf2out_finish): When outputting DIEs to the fat part of an
20798 LTO object first reset DIEs.
20799 (dwarf2out_early_finish): Output early DIEs when generating LTO.
20800 (modified_type_die): Check for decl_ultimate_origin being self
20802 (gen_type_die_with_usage): Likewise.
20803 (gen_typedef_die): Allow decl_ultimate_origin being self.
20804 (set_decl_abstract_flags): Remove.
20805 (set_block_abstract_flags): Likewise.
20806 (dwarf2out_abstract_function): Treat the early generated DIEs
20807 as the abstract copy and only add DW_AT_inline and
20808 DW_AT_artificial here and call set_decl_origin_self.
20809 If the DIE has an abstract origin don't do anything.
20810 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
20811 if we have none yet (Go fails to build one, PR78628).
20812 (variably_modified_type_p): Prevent endless recursion for Ada
20813 cyclic pointer types.
20814 * lto-streamer-in.c: Include debug.h.
20815 (dref_queue): New global.
20816 (lto_read_tree_1): Stream in DIE references.
20817 (lto_input_tree): Register DIE references.
20818 (input_function): Stream DECL_DEBUG_ARGS.
20819 * lto-streamer-out.c: Include debug.h.
20820 (lto_write_tree_1): Output DIE references.
20821 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
20822 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
20823 (output_function): Stream DECL_DEBUG_ARGS.
20824 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20825 Stream DECL_ABSTRACT_ORIGIN.
20826 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20827 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20828 DECL_CONTEXT for file-scope decls.
20829 * lto-streamer.h (struct dref_entry): Declare.
20830 (dref_queue): Likewise.
20831 * cfgexpand.c (pass_expand::execute): Do not call the
20832 outlining_inline_function hook here.
20833 * lto-wrapper.c (debug_obj): New global.
20834 (tool_cleanup): Unlink it if required.
20835 (debug_objcopy): New function.
20836 (run_gcc): Handle early debug sections in the IL files by
20837 extracting them to separate files, partially linkin them and
20838 feeding the result back as result to the linker.
20839 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20840 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20841 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20842 sections into a separate segment.
20843 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20845 (darwin_asm_dwarf_section): Likewise.
20846 (darwin_asm_output_dwarf_offset): Likewise.
20847 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20849 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20850 Alan Hayward <alan.hayward@arm.com>
20851 David Sherwood <david.sherwood@arm.com>
20853 * read-md.h (md_reader::record_potential_iterator_use): Replace
20854 pointer argument with an rtx and an index.
20855 * read-rtl.c (iterator_group::apply_iterator): Likewise.
20856 (apply_mode_iterator): Likewise.
20857 (apply_code_iterator): Likewise.
20858 (apply_int_iterator): Likewise.
20859 (apply_subst_iterator): Likewise.
20860 (record_iterator_use): Likewise.
20861 (record_attribute_use): Likewise.
20862 (md_reader::record_potential_iterator_use): Likewise. Update calls
20863 to record_iterator_use and apply_iterator.
20864 (iterator_use): Replace ptr with x and index.
20865 (attribute_use): Likewise.
20866 (apply_attribute_uses): Update calls to apply_iterator.
20867 (apply_iterators): Likewise. Update initialization of iterator_use.
20868 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
20869 and record_potential_iterator_use.
20870 (rtx_reader::read_rtx_operand): Likewise.
20872 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20873 Alan Hayward <alan.hayward@arm.com>
20874 David Sherwood <david.sherwood@arm.com>
20876 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
20879 2017-08-21 Richard Biener <rguenther@suse.de>
20881 PR middle-end/81884
20882 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
20883 at struct end conservatively when comparing common bases.
20885 2017-08-21 Richard Biener <rguenther@suse.de>
20887 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20888 (mem_ref_in_stmt): Remove.
20889 (determine_max_movement): Use ref index to get at the reference.
20890 (invariantness_dom_walker::before_dom_children): Deal with
20891 lim data already initialized.
20892 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20894 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20896 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20897 (TARGET_ISA_ROUND): Ditto.
20898 (TARGET_ROUND): Ditto.
20899 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20900 * config/i386/i386.md: Ditto.
20901 * config/i386/sse.md: Ditto.
20902 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20903 with OPTION_MASK_ISA_SSE4_1.
20905 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20908 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20909 __builtin_ia32_lzcnt_u16.
20911 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20914 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20915 (rs6000_set_current_function): Rewrite function to use it.
20917 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20920 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20921 and TYPE_WARN_IF_NOT_ALIGN.
20922 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20923 (handle_warn_if_not_align): New.
20924 (place_union_field): Call handle_warn_if_not_align.
20925 (place_field): Call handle_warn_if_not_align.
20926 Copy TYPE_WARN_IF_NOT_ALIGN.
20927 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20928 (layout_type): Likewise.
20929 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20931 (tree_decl_common): Add warn_if_not_align.
20932 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20933 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20934 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20935 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20936 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20937 * doc/extend.texi: Document warn_if_not_aligned attribute.
20938 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20940 2017-08-17 Martin Liska <mliska@suse.cz>
20943 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20944 (get_data_dependence): Use it as pointer type.
20945 (distribute_loop): Likewise.
20947 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20949 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20950 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20951 (p8_vmrgow_<mode>_direct): New define_insn.
20952 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20953 handle endianness for vmrgew and vmrgow permute patterns.
20955 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20957 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20958 * config/rs6000/rs6000-cpus.def: Remove comment.
20959 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20960 (POWERPC_MASKS): Likewise.
20961 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20962 use of TARGET_VSX_TIMODE.
20963 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20964 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20965 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20966 (rs6000_option_override_internal): Remove dead code.
20967 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20968 (rs6000_legitimize_reload_address): Likewise.
20969 (rs6000_legitimate_address_p): Likewise.
20970 (rs6000_opt_masks): Delete "vsx-timode".
20971 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20972 from function comment.
20973 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20974 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20975 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20977 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20978 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20979 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20980 (VSX_TI): Likewise.
20982 (define_peephole2): Likewise.
20984 2017-08-17 Martin Sebor <msebor@redhat.com>
20987 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20988 past the end of an array.
20989 (test_pp_format): Add test cases.
20991 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20993 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20994 missing ECF_NOTHROW flags.
20996 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20999 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
21000 operands residing in integer registers.
21001 (*vsx_le_perm_load_<mode>): Likewise.
21002 (*vsx_le_perm_store_<mode>): Likewise.
21003 (define_peephole2): Add peepholes to optimize the above.
21005 2017-08-17 Marek Polacek <polacek@redhat.com>
21007 PR middle-end/81814
21008 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21009 to mimic what shorten_compare did. Change the return type to bool.
21010 (fold_cond_expr_with_comparison): Update call to
21011 operand_equal_for_comparison_p.
21012 (fold_ternary_loc): Likewise.
21014 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21016 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21018 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21020 2017-08-17 Richard Biener <rguenther@suse.de>
21022 * tree-ssa-structalias.c (solve_graph): When propagating
21023 to successors update the graphs succ edges and avoid duplicate work.
21025 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21028 * config/i386/i386.c (ix86_option_override_internal): Save target
21029 specific options after ix86_stack_protector_guard_reg was changed.
21031 2017-08-17 Richard Biener <rguenther@suse.de>
21033 PR tree-optimization/81827
21034 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21035 (new_var_info): Initialize it conservatively.
21036 (get_call_vi): Mark register vars.
21037 (new_scalar_tmp_constraint_exp): Likewise.
21038 (handle_rhs_call): Likewise.
21039 (handle_const_call): Likewise.
21040 (create_function_info_for): Likewise.
21041 (solve_constraints): Sort varinfos to separate register from
21042 non-register vars to pack points-to solution bitmaps during
21045 2017-08-17 Marek Polacek <polacek@redhat.com>
21047 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21049 2017-08-17 Richard Biener <rguenther@suse.de>
21051 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21052 to true when overflow is undefined and we saturated the result.
21054 2017-08-17 Alan Modra <amodra@gmail.com>
21057 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21058 Don't use store multiple if only one reg needs saving.
21059 (interesting_frame_related_regno): New function.
21060 (rs6000_frame_related): Don't emit frame info for regs that
21062 (rs6000_emit_epilogue): Likewise.
21064 2017-08-16 Nathan Sidwell <nathan@acm.org>
21066 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21067 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21068 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21069 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21070 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21071 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21072 (verify_type): Adjust for TYPE_BINFO move.
21073 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21074 process TYPE_BINFO directly.
21075 (hash_tree): Likewise.
21076 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21078 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21081 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21083 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21085 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21088 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21089 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21090 Add ix86_binary_operator_ok to insn constraint.
21091 (*iordi_1_bts): Ditto.
21092 (*xordi_1_btc): Ditto.
21093 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21094 Update corresponding peephole2 pattern.
21098 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21100 PR tree-optimization/81832
21101 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21102 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21104 2017-08-16 Marek Polacek <polacek@redhat.com>
21107 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21108 regarding size_zero_node.
21110 2017-08-16 Martin Liska <mliska@suse.cz>
21113 * config.gcc: Respect previously set extra_objs in case
21116 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21118 PR tree-optimization/81835
21119 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21120 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21121 not depend on the phi.
21123 2017-08-16 Alan Modra <amodra@gmail.com>
21125 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21128 2017-08-16 Alan Modra <amodra@gmail.com>
21130 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21131 (save_reg_p): ..into this. Update all callers.
21132 (first_reg_to_save): Simplify.
21134 2017-08-16 Alan Modra <amodra@gmail.com>
21136 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21139 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21143 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21144 constant count if that count is less than 32.
21146 2017-08-15 Nathan Sidwell <nathan@acm.org>
21148 * gcc.c (execute): Emit friendlier message if inferior is killed
21149 by an external cause.
21151 2017-08-15 Richard Biener <rguenther@suse.de>
21153 PR tree-optimization/81790
21154 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21155 CONSTRUCTORs from simplifying and VN.
21157 2017-08-14 Martin Sebor <msebor@redhat.com>
21159 * builtin-attrs.def: Add comments.
21161 2017-08-14 Martin Sebor <msebor@redhat.com>
21164 * doc/extend.texi (attribute nonstring): Document new attribute.
21166 2017-08-14 Martin Sebor <msebor@redhat.com>
21169 * tree-diagnostic.c (default_tree_printer): Handle %G.
21170 * gimple-pretty-print.h (percent_G_format): Declare new function.
21171 * gimple-pretty-print.c (percent_G_format): Define.
21172 * tree-pretty-print.c (percent_K_format): Add argument.
21174 2017-08-14 Martin Sebor <msebor@redhat.com>
21176 PR translation/79998
21177 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21178 Remove a stray space.
21180 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21183 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21184 (*iordi_1_bts): Ditto.
21185 (*xordi_1_btc): Ditto.
21187 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21190 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21191 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21193 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21194 quoted strings, and make more translator-friendly.
21195 (darwin_rs6000_override_options): Likewise.
21196 (rs6000_option_override_internal): Likewise.
21197 (rs6000_return_in_memory): Fix overlong line.
21198 (init_cmulative_args): Use quoted strings, and make more
21199 translator-friendly.
21200 (rs6000_pass_by_reference): Fix overlong line.
21201 (def_builtin): Use quoted strings.
21202 (altivec_expand_predicate_builtin): Use quoted strings, and make
21203 more translator-friendly.
21204 (htm_expand_builtin): Use quoted strings.
21205 (cpu_expand_builtin): Use quoted strings, and make more
21206 translator-friendly.
21207 (altivec_expand_builtin): Likewise.
21208 (paired_expand_predicate_builtin): Likewise.
21209 (rs6000_invalid_builtin): Likewise.
21210 (builtin_function_type): Use quoted strings.
21211 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21212 more translator-friendly.
21213 (rs6000_trampoline_init): Likewise.
21214 (rs6000_handle_altivec_attribute): Likewise.
21215 (rs6000_inner_target_options): Use quoted strings.
21216 (rs6000_disable_incompatible_switches): Likewise.
21217 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21218 strings, and make more translator-friendly.
21219 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21221 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21223 PR tree-optimization/81799
21224 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21225 cond_expr to simple gimple operand.
21227 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21229 PR middle-end/46932
21230 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21232 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21236 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21237 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21238 TARGET_GASISR_PROLOGUES.
21239 * config/avr/avr.c (avr_option_override): Same.
21240 (avr_pass_pre_proep::execute): Same.
21242 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21245 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21246 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21248 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21250 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21251 define_insn_and_split. Split to a memory load from 0 in
21252 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21253 using PTR mode iterator.
21254 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21255 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21256 (*add_tp_<mode>): Redefine as define_insn_and_split.
21257 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21258 address space. Merge with *add_tp_x32 using PTR mode iterator.
21259 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21260 Split to an add with a memory load from 0 in
21261 DEFAULT_TLS_SEG_REG address space.
21263 2017-08-12 Andrew Pinski <apinski@cavium.com>
21265 * config/aarch64/aarch64-option-extensions.def (rdma):
21266 Fix feature string to what Linux prints out in /proc/cpuinfo.
21268 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21271 * dwarf2out.c (modified_type_die): For C typedef types that have
21272 an ultimate origin, process the ultimate origin instead of the
21274 (gen_typedef_die): Assert that input DECLs have no ultimate
21276 (gen_type_die_with_usage): For typedef variants that have an
21277 ultimate origin, just call gen_decl_die on the original DECL.
21278 (process_scope_var): Avoid creating DIEs for local typedefs and
21279 concrete static variables.
21281 2017-08-12 Alan Modra <amodra@gmail.com>
21285 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21286 match gnu-user.h startfile.
21287 (ENDFILE_LINUX_SPEC): Similarly.
21289 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21292 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21294 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21296 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21297 * config/aarch64/aarch64.md (mov<mode>): Change.
21298 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21299 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21300 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21302 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21304 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21305 for storage order barriers.
21307 2017-08-11 Martin Liska <mliska@suse.cz>
21310 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21311 variables of void type.
21313 2017-08-11 Martin Liska <mliska@suse.cz>
21315 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21316 TARGET_SUPPORTS_ALIASES.
21317 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21318 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21319 (optimize_weakref): Likewise.
21320 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21321 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21322 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21324 2017-08-11 Martin Liska <mliska@suse.cz>
21327 * config/i386/i386.c (make_resolver_func): Do complete
21328 refactoring of the function.
21330 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21333 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21334 * config/i386/i386.c (ix86_stack_protect_guard): Use
21335 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21336 * doc/invoke.texi (x86 Options): Document
21337 -mstack-protector-guard-symbol= option.
21339 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21341 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21342 * config/i386/i386.c (ix86_split_stack_guard): New function.
21343 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21344 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21345 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21346 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21347 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21348 (split_stack_space_check): Call ix86_split_stack_guard.
21350 2017-08-10 Martin Sebor <msebor@redhat.com>
21352 * print-tree.c (print_node): Print location using the established
21354 Replace spaces with colons.
21355 (debug_raw, debug): Ditto.
21357 2017-08-10 Martin Sebor <msebor@redhat.com>
21360 * pretty-print.c (pp_format): Correct the handling of %s precision.
21362 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21365 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21367 (ix86_finalize_stack_frame_flags): This. Also clear
21368 frame_pointer_needed if -fno-omit-frame-pointer is used without
21370 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21371 with ix86_finalize_stack_frame_flags.
21372 (ix86_expand_epilogue): Likewise.
21373 (ix86_expand_split_stack_prologue): Likewise.
21374 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21376 2017-08-10 Martin Liska <mliska@suse.cz>
21379 * c-attribs.c (handle_target_attribute):
21380 Report warning for an empty string argument of target attribute.
21382 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21385 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21387 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21388 or DECL_NONLOCAL labels.
21389 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21390 or DECL_NONLOCAL labels here.
21392 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21394 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21395 to indicate when early gimple folding has been disabled.
21396 (rs6000_gimple_fold_builtin): Add debug content.
21397 (rs6000_invalid_builtin): Fix whitespace.
21398 (rs6000_expand_builtin): Fix whitespace.
21399 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21401 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21404 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21405 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21407 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21409 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21410 (qdf24xx): Likewise.
21411 * config/aarch64/aarch64.md: Include falkor.md.
21412 * config/aarch64/falkor.md: New.
21414 2017-08-09 Marek Polacek <polacek@redhat.com>
21417 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21418 * diagnostic.c (emit_diagnostic): Add a comment.
21419 (emit_diagnostic_valist): New function.
21421 2017-08-09 Marek Polacek <polacek@redhat.com>
21424 * input.c (make_location): New overload.
21425 * input.h (make_location): Declare.
21427 2017-08-08 Alan Modra <amodra@gmail.com>
21428 H.J. Lu <hongjiu.lu@intel.com>
21431 * gcc.c (NO_PIE_SPEC): Delete.
21432 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21434 (LINK_PIE_SPEC): ..to here.
21435 (LINK_COMMAND_SPEC): Support -no-pie.
21436 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21437 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21439 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21440 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21441 (ENDFILE_CRTEND_SPEC): Similarly.
21443 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21446 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21447 (mstack-protector-guard-offset=): Ditto.
21448 * config/i386/i386.c (ix86_option_override): Handle
21449 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21451 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21452 ix86_stack_protect_guard_offset variables.
21453 (TARGET_STACK_PROTECT_GUARD): Always define.
21454 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21455 and -mstack-protector-guard-offset= options.
21457 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21459 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21460 boundary case for the last candidate.
21462 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21464 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21465 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21467 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21469 PR middle-end/19706
21470 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21471 * config/aarch64/aarch64-builtins.c
21472 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21473 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21474 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21476 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21477 Andrew Pinski <pinskia@gmail.com>
21479 PR middle-end/19706
21480 * internal-fn.def (XORSIGN): New.
21481 * optabs.def (xorsign_optab): New.
21482 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21483 (convert_expand_mult_copysign): New.
21484 (pass_optimize_widening_mul::execute): Call
21485 convert_expand_mult_copysign.
21487 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21489 PR tree-optimization/81354
21490 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21491 Insert on edges rather than explicitly creating landing pads.
21492 (analyze_candidates_and_replace): Commit edge inserts.
21494 2017-08-08 Richard Biener <rguenther@suse.de>
21496 PR middle-end/81719
21497 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21498 (expand_simple_operations): Also look through ADDR_EXPRs with
21499 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21501 2017-08-08 Richard Biener <rguenther@suse.de>
21503 PR tree-optimization/81723
21504 * tree-vect-slp.c (struct bst_traits): New hash traits.
21505 (bst_fail): New global.
21506 (vect_build_slp_tree_2): New worker, split out from ...
21507 (vect_build_slp_tree): ... this now wrapping it with using
21508 bst_fail set to cache SLP tree build fails. Properly handle
21510 (vect_analyze_slp_instance): Allocate and free bst_fail.
21512 2017-08-08 Martin Liska <mliska@suse.cz>
21515 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21516 LABEL_DECLs that can be from a different function.
21518 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21520 PR tree-optimization/81744
21521 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21522 loop's number of iterations.
21524 2017-08-08 Martin Liska <mliska@suse.cz>
21526 * asan.c: Include header files.
21527 * attribs.c (build_decl_attribute_variant): New function moved
21529 (build_type_attribute_qual_variant): Likewise.
21530 (cmp_attrib_identifiers): Likewise.
21531 (simple_cst_list_equal): Likewise.
21532 (omp_declare_simd_clauses_equal): Likewise.
21533 (attribute_value_equal): Likewise.
21534 (comp_type_attributes): Likewise.
21535 (build_type_attribute_variant): Likewise.
21536 (lookup_ident_attribute): Likewise.
21537 (remove_attribute): Likewise.
21538 (merge_attributes): Likewise.
21539 (merge_type_attributes): Likewise.
21540 (merge_decl_attributes): Likewise.
21541 (merge_dllimport_decl_attributes): Likewise.
21542 (handle_dll_attribute): Likewise.
21543 (attribute_list_equal): Likewise.
21544 (attribute_list_contained): Likewise.
21545 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21546 (lookup_attribute_by_prefix): Likewise.
21547 * bb-reorder.c: Include header files.
21548 * builtins.c: Likewise.
21549 * calls.c: Likewise.
21550 * cfgexpand.c: Likewise.
21551 * cgraph.c: Likewise.
21552 * cgraphunit.c: Likewise.
21553 * convert.c: Likewise.
21554 * dwarf2out.c: Likewise.
21555 * final.c: Likewise.
21556 * fold-const.c: Likewise.
21557 * function.c: Likewise.
21558 * gimple-expr.c: Likewise.
21559 * gimple-fold.c: Likewise.
21560 * gimple-pretty-print.c: Likewise.
21561 * gimple.c: Likewise.
21562 * gimplify.c: Likewise.
21563 * hsa-common.c: Likewise.
21564 * hsa-gen.c: Likewise.
21565 * internal-fn.c: Likewise.
21566 * ipa-chkp.c: Likewise.
21567 * ipa-cp.c: Likewise.
21568 * ipa-devirt.c: Likewise.
21569 * ipa-fnsummary.c: Likewise.
21570 * ipa-inline.c: Likewise.
21571 * ipa-visibility.c: Likewise.
21573 * lto-cgraph.c: Likewise.
21574 * omp-expand.c: Likewise.
21575 * omp-general.c: Likewise.
21576 * omp-low.c: Likewise.
21577 * omp-offload.c: Likewise.
21578 * omp-simd-clone.c: Likewise.
21579 * opts-global.c: Likewise.
21580 * passes.c: Likewise.
21581 * predict.c: Likewise.
21582 * sancov.c: Likewise.
21583 * sanopt.c: Likewise.
21584 * symtab.c: Likewise.
21585 * toplev.c: Likewise.
21586 * trans-mem.c: Likewise.
21587 * tree-chkp.c: Likewise.
21588 * tree-eh.c: Likewise.
21589 * tree-into-ssa.c: Likewise.
21590 * tree-object-size.c: Likewise.
21591 * tree-parloops.c: Likewise.
21592 * tree-profile.c: Likewise.
21593 * tree-ssa-ccp.c: Likewise.
21594 * tree-ssa-live.c: Likewise.
21595 * tree-ssa-loop.c: Likewise.
21596 * tree-ssa-sccvn.c: Likewise.
21597 * tree-ssa-structalias.c: Likewise.
21598 * tree-ssa.c: Likewise.
21599 * tree-streamer-in.c: Likewise.
21600 * tree-vectorizer.c: Likewise.
21601 * tree-vrp.c: Likewise.
21602 * tsan.c: Likewise.
21603 * ubsan.c: Likewise.
21604 * varasm.c: Likewise.
21605 * varpool.c: Likewise.
21606 * tree.c: Remove functions moved to attribs.[ch].
21607 * tree.h: Likewise.
21608 * config/aarch64/aarch64.c: Add attrs.h header file.
21609 * config/alpha/alpha.c: Likewise.
21610 * config/arc/arc.c: Likewise.
21611 * config/arm/arm.c: Likewise.
21612 * config/avr/avr.c: Likewise.
21613 * config/bfin/bfin.c: Likewise.
21614 * config/c6x/c6x.c: Likewise.
21615 * config/cr16/cr16.c: Likewise.
21616 * config/cris/cris.c: Likewise.
21617 * config/darwin.c: Likewise.
21618 * config/epiphany/epiphany.c: Likewise.
21619 * config/fr30/fr30.c: Likewise.
21620 * config/frv/frv.c: Likewise.
21621 * config/ft32/ft32.c: Likewise.
21622 * config/h8300/h8300.c: Likewise.
21623 * config/i386/winnt.c: Likewise.
21624 * config/ia64/ia64.c: Likewise.
21625 * config/iq2000/iq2000.c: Likewise.
21626 * config/lm32/lm32.c: Likewise.
21627 * config/m32c/m32c.c: Likewise.
21628 * config/m32r/m32r.c: Likewise.
21629 * config/m68k/m68k.c: Likewise.
21630 * config/mcore/mcore.c: Likewise.
21631 * config/microblaze/microblaze.c: Likewise.
21632 * config/mips/mips.c: Likewise.
21633 * config/mmix/mmix.c: Likewise.
21634 * config/mn10300/mn10300.c: Likewise.
21635 * config/moxie/moxie.c: Likewise.
21636 * config/msp430/msp430.c: Likewise.
21637 * config/nds32/nds32-isr.c: Likewise.
21638 * config/nds32/nds32.c: Likewise.
21639 * config/nios2/nios2.c: Likewise.
21640 * config/nvptx/nvptx.c: Likewise.
21641 * config/pa/pa.c: Likewise.
21642 * config/pdp11/pdp11.c: Likewise.
21643 * config/powerpcspe/powerpcspe.c: Likewise.
21644 * config/riscv/riscv.c: Likewise.
21645 * config/rl78/rl78.c: Likewise.
21646 * config/rx/rx.c: Likewise.
21647 * config/s390/s390.c: Likewise.
21648 * config/sh/sh.c: Likewise.
21649 * config/sol2.c: Likewise.
21650 * config/sparc/sparc.c: Likewise.
21651 * config/spu/spu.c: Likewise.
21652 * config/stormy16/stormy16.c: Likewise.
21653 * config/tilegx/tilegx.c: Likewise.
21654 * config/tilepro/tilepro.c: Likewise.
21655 * config/v850/v850.c: Likewise.
21656 * config/vax/vax.c: Likewise.
21657 * config/visium/visium.c: Likewise.
21658 * config/xtensa/xtensa.c: Likewise.
21660 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21663 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21664 constraints since the -mupper-regs-* switches have been
21666 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21667 into a vector from a double word element that was extracted from
21668 another vector, and eliminate extra XXPERMDI instructions.
21669 (vsx_concat_<mode>_2): Likewise.
21670 (vsx_concat_<mode>_3): Likewise.
21671 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21672 concat to allow optimizing inserts from previous extracts.
21674 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21676 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21677 memory reference to a SSP offset in TLS address space.
21678 (ix86_print_operand) <case '@'>: Remove.
21679 (ix86_print_operand_punct_valid_p): Remove '@' code.
21680 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21681 UNSPEC_SP_TLS_TEST.
21682 (stack_tls_protect_set_<mode>): Remove.
21683 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21684 (stack_tls_protect_test_<mode>): Remove.
21685 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21687 2017-08-07 Olivier Hainque <hainque@adacore.com>
21690 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21692 2017-08-07 Douglas Rupp <rupp@adacore.com>
21694 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21695 variable was referenced as multidir in command.
21697 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21700 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21703 2017-08-07 Martin Liska <mliska@suse.cz>
21705 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21706 * config/rl78/rl78.c: Add include of attribs.h.
21707 * config/sh/sh.c: Likewise.
21708 * config/v850/v850.c: Likewise.
21710 2017-08-07 Tom de Vries <tom@codesourcery.com>
21712 PR middle-end/78266
21713 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21715 2017-08-07 Martin Liska <mliska@suse.cz>
21717 * config/mips/mips.c: Include attribs.h.
21719 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21722 * doc/invoke.texi: Document change in behvaior for -Ofast for
21725 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21727 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21729 (aarch64_pop_regs): Likewise.
21730 (aarch64_gen_load_pair): Likewise.
21731 (aarch64_save_callee_saves): Likewise.
21732 (aarch64_restore_callee_saves): Likewise.
21734 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21736 * config/i386/i386.c: Revert the last change.
21738 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21741 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21743 (ix86_finalize_stack_frame_flags): This. Also clear
21744 frame_pointer_needed if -fno-omit-frame-pointer is used without
21746 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21747 with ix86_finalize_stack_frame_flags.
21748 (ix86_expand_epilogue): Likewise.
21749 (ix86_expand_split_stack_prologue): Likewise.
21751 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21754 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21755 to P_AES for Westmere.
21757 2017-08-07 Jonathan Yong <10walls@gmail.com>
21759 * config/i386/mingw.opt (fset-stack-executable): Removed.
21760 * config/i386/cygming.opt (fset-stack-executable): Moved
21762 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21764 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21766 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21768 2017-08-07 Marek Polacek <polacek@redhat.com>
21770 PR middle-end/81737
21771 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21773 2017-08-07 Martin Liska <mliska@suse.cz>
21775 * attribs.h (canonicalize_attr_name): New function.
21776 (cmp_attribs): Move from c-format.c and adjusted.
21777 (is_attribute_p): Moved from tree.h.
21778 * tree-inline.c: Add new includes.
21779 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21780 (private_is_attribute_p): Remove.
21781 (private_lookup_attribute): Likewise.
21782 (private_lookup_attribute_by_prefix): Simplify.
21783 (remove_attribute): Use is_attribute_p.
21784 * tree.h: Remove removed declarations.
21786 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21788 PR middle-end/81698
21789 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21790 instead of computing it in the function. Formatting fix.
21791 (expand_case): Don't rely on default_edge being the first edge,
21792 clear it if removing it, pass default_edge to
21793 emit_case_dispatch_table.
21794 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21797 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
21799 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
21800 insn in the function, emit NOP after the insn.
21802 2017-08-06 Tom de Vries <tom@codesourcery.com>
21804 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
21807 2017-08-06 Tom de Vries <tom@codesourcery.com>
21809 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
21812 2017-08-04 Yury Gribov <tetra2005@gmail.com>
21814 PR tree-optimization/57371
21815 * match.pd: New pattern.
21817 2017-08-04 Marek Polacek <polacek@redhat.com>
21819 PR middle-end/81695
21820 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
21821 perform the computation in offset_int.
21823 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21825 PR tree-optimization/81136
21826 * tree-vectorizer.h: Include tree-hash-traits.h.
21827 (vec_base_alignments): New typedef.
21828 (vec_info): Add a base_alignments field.
21829 (vect_record_base_alignments): Declare.
21830 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21832 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21833 (create_data_ref): Add an is_conditional_in_stmt argument.
21834 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21835 the is_conditional_in_stmt field.
21836 (data_ref_loc): Add an is_conditional_in_stmt field.
21837 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21838 (find_data_references_in_stmt): Update call to create_data_ref.
21839 (graphite_find_data_references_in_stmt): Likewise.
21840 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21841 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21842 (vect_record_base_alignment): New function.
21843 (vect_record_base_alignments): Likewise.
21844 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21845 for nested statements even if we fail to compute a misalignment.
21846 Use pooled base alignments for unconditional references.
21847 (vect_find_same_alignment_drs): Compare base addresses instead
21849 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21850 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
21852 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21854 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
21855 Add an explicit name for the enum. Use auto_vec for slp_instances
21856 and grouped_stores.
21857 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
21859 (_bb_vec_info): Add a constructor and destructor.
21860 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21861 (destroy_loop_vec_info): Delete.
21862 (vect_destroy_datarefs): Likewise.
21863 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21864 (vec_info::vec_info): New function.
21865 (vec_info::~vec_info): Likewise.
21866 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
21867 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
21868 destroy_loop_vec_info.
21869 * tree-vect-loop.c (new_loop_vec_info): Replace with...
21870 (_loop_vec_info::_loop_vec_info): ...this.
21871 (destroy_loop_vec_info): Replace with...
21872 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
21873 the stmt_vec_infos. Leave handling of vec_info information to its
21874 destructor. Remove explicit vector releases.
21875 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
21876 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
21877 * tree-vect-slp.c (new_bb_vec_info): Replace with...
21878 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
21879 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
21880 (destroy_bb_vec_info): Replace with...
21881 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
21882 information to its destructor.
21883 (vect_slp_analyze_bb_1): Use new and delete instead of
21884 new_bb_vec_info and destroy_bb_vec_info.
21885 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21888 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21890 * tree-data-ref.h (subscript): Add access_fn field.
21891 (data_dependence_relation): Add could_be_independent_p.
21892 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21893 (same_access_functions): Move to tree-data-ref.c.
21894 * tree-data-ref.c (ref_contains_union_access_p): New function.
21895 (access_fn_component_p): Likewise.
21896 (access_fn_components_comparable_p): Likewise.
21897 (dr_analyze_indices): Add a reference to access_fn_component_p.
21898 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21900 (constant_access_functions): Likewise.
21901 (add_other_self_distances): Likewise.
21902 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21903 (initialize_data_dependence_relation): Use XCNEW and remove
21904 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21905 of access functions that have the same type. Allow the
21906 subsequence to end with different bases in some circumstances.
21907 Record the chosen access functions in SUB_ACCESS_FN.
21908 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21909 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21910 (subscript_dependence_tester_1): Likewise dra and drb.
21911 (build_classic_dist_vector): Update calls accordingly.
21912 (subscript_dependence_tester): Likewise.
21913 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21914 DDR_COULD_BE_INDEPENDENT_P.
21915 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21916 comp_alias_ddrs instead of may_alias_ddrs.
21917 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21919 (vect_analyze_data_ref_dependence): Use it if
21920 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21921 distance vectors if that fails.
21922 (dependence_distance_ge_vf): New function.
21923 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21924 LOOP_VINFO_MAY_ALIAS_DDRS.
21926 2017-08-04 Richard Biener <rguenther@suse.de>
21928 PR middle-end/81705
21929 * fold-const.c (fold_binary_loc): Properly restrict
21930 minus_var0 && minus_var1 case when associating undefined overflow
21933 2017-08-04 Tom de Vries <tom@codesourcery.com>
21935 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21937 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21939 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21940 Don't start diagnostic messages with a capital letter.
21941 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21943 (rs6000_invalid_builtin): Likewise.
21944 (rs6000_trampoline_init): Likewise.
21946 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21949 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21950 after setting changeable df flags.
21952 2017-08-03 Richard Biener <rguenther@suse.de>
21954 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21955 up if the use is in USE - X.
21957 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21959 * toplev.c (dumpfile.h): New include.
21960 (internal_error_reentered): New static function. Use it...
21961 (internal_error_function): ...here to handle reentered internal_error.
21963 2017-08-03 Richard Biener <rguenther@suse.de>
21965 PR middle-end/81148
21966 * fold-const.c (split_tree): Add minus_var and minus_con
21967 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21968 here but always use minus_*.
21969 (associate_trees): Assert we never associate with MINUS_EXPR
21970 and NULL first operand. Do not recurse for PLUS_EXPR operands
21971 when associating as MINUS_EXPR either.
21972 (fold_binary_loc): Track minus_var and minus_con.
21974 2017-08-03 Tom de Vries <tom@codesourcery.com>
21977 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21978 ACCEL_COMPILER, apply finish_options on
21979 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21981 2017-08-03 Tom de Vries <tom@codesourcery.com>
21984 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21985 function_entry_patch_area_size > 0.
21987 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21990 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21991 instead of 10??LU, perform unit multiplication in wide_int,
21992 don't change alloc_object_size_limit if the limit is larger
21995 PR tree-optimization/81655
21996 PR tree-optimization/81588
21997 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21998 the case when ranges[i].low and high are 1 for unsigned type with
22001 PR middle-end/81052
22002 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
22003 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
22005 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22007 * tree-vrp.h: Add include guard.
22009 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22012 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22013 (ud2): New insn pattern.
22014 * config/i386/i386.c (ix86_expand_epilogue):
22015 For naked functions, generate ud2 instead of trap insn.
22017 2017-08-02 Marek Polacek <polacek@redhat.com>
22020 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22022 2017-08-02 Tom de Vries <tom@codesourcery.com>
22023 Cesar Philippidis <cesar@codesourcery.com>
22025 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22026 Add missing edge probabilities.
22028 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22030 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22031 Correct endianness.
22033 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22035 PR middle-end/79499
22036 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22037 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22038 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22039 prologue_seq sequences - if any.
22041 2017-08-02 Richard Biener <rguenther@suse.de>
22043 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22044 via vectors if supported, integer extracts via punning if supported
22045 or otherwise vector extracts.
22047 2017-08-02 Richard Biener <rguenther@suse.de>
22049 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22051 (bitmap_insert_into_set): ... this.
22053 2017-08-02 Richard Biener <rguenther@suse.de>
22055 PR tree-optimization/81633
22057 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22059 PR tree-optimization/71752
22060 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22062 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22064 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22065 (machine_function::call_ms2sysv_pad_out): Remove field.
22066 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22067 (ix86_compute_frame_layout): Likewise.
22069 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22072 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22073 attribute with interrupt attribute.
22075 2017-08-01 Andrew Pinski <apinski@cavium.com>
22077 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22078 BIT_INSERT_EXPR's operand 1
22079 to see if the types precision matches.
22081 2017-08-01 Martin Liska <mliska@suse.cz>
22083 PR middle-end/70140
22084 * builtins.c (expand_builtin_memcpy_args): Remove.
22085 (expand_builtin_memcpy): Call newly added function
22086 expand_builtin_memory_copy_args.
22087 (expand_builtin_memcpy_with_bounds): Likewise.
22088 (expand_builtin_mempcpy): Remove last argument.
22089 (expand_builtin_mempcpy_with_bounds): Likewise.
22090 (expand_builtin_memory_copy_args): New function created from
22091 expand_builtin_mempcpy_args with small modifications.
22092 (expand_builtin_mempcpy_args): Remove.
22093 (expand_builtin_stpcpy): Remove unused argument.
22094 (expand_builtin): Likewise.
22095 (expand_builtin_with_bounds): Likewise.
22097 2017-08-01 Martin Liska <mliska@suse.cz>
22100 Make mempcpy more optimal (PR middle-end/70140).
22102 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22105 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22106 __builtin_vec_cmpne verify both arguments are compatible vectors
22107 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22108 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22109 move computation of aligned to after checking the argument types.
22113 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22114 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22117 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22118 Doug Rupp <rupp@adacore.com>
22119 Olivier Hainque <hainque@adacore.com>
22121 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22122 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22124 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22125 for TARGET_OS_CPP_BUILTIN.
22126 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22127 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22129 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22130 passing required abi options to the assembler for EABI configurations.
22131 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22132 of .text.hot and .text.unlikely sections for kernel modules when
22133 using ARM style exceptions.
22134 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22135 options. Add EXTRA_CC1_SPEC.
22136 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22138 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22140 (ARM_TARGET2_DWARF_FORMAT): Define.
22141 * config/arm/t-vxworks: Adjust multilib control to removal of the
22142 Diab command line options.
22144 2017-08-01 Martin Liska <mliska@suse.cz>
22146 PR gcov-profile/81561
22147 * gcov.c (unblock): Make unblocking safe as we need to preserve
22148 index correspondence of blocks and block_lists.
22150 2017-08-01 Richard Biener <rguenther@suse.de>
22152 PR tree-optimization/81181
22153 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22154 (compute_antic): ... end of iteration here.
22156 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22158 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22159 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22160 (ftree-slp-vectorize): Likewise.
22161 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22162 can no longer be set independent of flag_tree_loop_vectorize.
22163 * omp-general.c (emp_max_vf): Likewise.
22164 * opts.c (enable_fdo_optimizations): Remove references to
22165 flag_tree_vectorize, these are now implicit.
22166 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22167 and leave it for the options machinery.
22169 2017-08-01 Martin Liska <mliska@suse.cz>
22171 PR middle-end/70140
22172 * builtins.c (expand_builtin_memcpy_args): Remove.
22173 (expand_builtin_memcpy): Call newly added function
22174 expand_builtin_memory_copy_args.
22175 (expand_builtin_memcpy_with_bounds): Likewise.
22176 (expand_builtin_mempcpy): Remove last argument.
22177 (expand_builtin_mempcpy_with_bounds): Likewise.
22178 (expand_builtin_memory_copy_args): New function created from
22179 expand_builtin_mempcpy_args with small modifications.
22180 (expand_builtin_mempcpy_args): Remove.
22181 (expand_builtin_stpcpy): Remove unused argument.
22182 (expand_builtin): Likewise.
22183 (expand_builtin_with_bounds): Likewise.
22185 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22188 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22189 print "ds:" only for immediates in generic address space.
22191 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22194 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22195 (ix86_function_ok_for_sibcall): Return false for naked functions.
22197 2017-08-01 Richard Biener <rguenther@suse.de>
22199 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22200 (compute_antic): Seed worklist with exit block predecessors.
22201 * cfganal.c (dfs_find_deadend): For a cycle return the source
22202 of the edge closing it.
22204 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22206 * config/aarch64/aarch64.c
22207 (aarch64_can_const_movi_rtx_p): Move 0 check.
22209 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22211 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22212 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22214 * match.pd: Ditto in address comparison pattern.
22216 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22218 PR tree-optimization/81627
22219 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22220 closed ssa form for store-store chain.
22222 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22224 PR tree-optimization/81620
22225 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22226 for store-store chain.
22228 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22230 PR tree-optimization/81588
22231 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22232 ranges[i].in_p, invert comparison code ccode. For >/>=,
22233 swap rhs1 and rhs2 and comparison code unconditionally,
22234 for </<= don't do that. Don't swap rhs1/rhs2 again if
22235 ranges[i].in_p, instead invert comparison code ccode if
22236 opcode or oe->rank is BIT_IOR_EXPR.
22239 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22240 a direct optab to conversion optab.
22241 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22242 with GET_MODE_INNER as last argument instead of optab_handler.
22243 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22244 vector extraction if possible and optab is available.
22245 * expr.c (store_constructor): Use convert_optab_handler instead
22246 of optab_handler. Use vector initialization from smaller
22247 vectors if possible and optab is available.
22248 * tree-vect-stmts.c (vectorizable_load): Likewise.
22249 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22250 now have two modes.
22251 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22252 of vec_init from half-sized vectors with the same element mode.
22253 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22254 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22255 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22256 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22257 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22258 after mode in gen_vec_extract* calls.
22259 (vec_extract<mode>): Renamed to ...
22260 (vec_extract<mode><ssescalarmodelower>): ... this.
22261 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22262 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22263 element mode after mode in gen_vec_init* calls.
22264 (VEC_INIT_HALF_MODE): New mode iterator.
22265 (vec_init<mode>): Renamed to ...
22266 (vec_init<mode><ssescalarmodelower>): ... this.
22267 (vec_init<mode><ssehalfvecmodelower>): New expander.
22268 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22269 (vec_extractv2sfsf): ... this.
22270 (vec_initv2sf): Renamed to ...
22271 (vec_initv2sfsf): ... this.
22272 (vec_extractv2si): Renamed to ...
22273 (vec_extractv2sisi): ... this.
22274 (vec_initv2si): Renamed to ...
22275 (vec_initv2sisi): ... this.
22276 (vec_extractv4hi): Renamed to ...
22277 (vec_extractv4hihi): ... this.
22278 (vec_initv4hi): Renamed to ...
22279 (vec_initv4hihi): ... this.
22280 (vec_extractv8qi): Renamed to ...
22281 (vec_extractv8qiqi): ... this.
22282 (vec_initv8qi): Renamed to ...
22283 (vec_initv8qiqi): ... this.
22284 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22285 (vec_init<mode>): Renamed to ...
22286 (vec_init<mode><VEC_base_l>): ... this.
22287 (vec_extract<mode>): Renamed to ...
22288 (vec_extract<mode><VEC_base_l>): ... this.
22289 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22290 (vec_initv2sfsf): ... this.
22291 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22292 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22293 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22294 element mode after mode in gen_vec_init* calls.
22295 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22296 (vec_init<mode><Vel>): ... this.
22297 (vec_extract<mode>): Renamed to ...
22298 (vec_extract<mode><Vel>): ... this.
22299 * config/aarch64/iterators.md (Vel): New mode attribute.
22300 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22301 Add element mode after mode in gen_vec_extract* calls.
22302 * config/s390/vector.md (non_vec_l): New mode attribute.
22303 (vec_extract<mode>): Renamed to ...
22304 (vec_extract<mode><non_vec_l>): ... this.
22305 (vec_init<mode>): Renamed to ...
22306 (vec_init<mode><non_vec_l>): ... this.
22307 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22308 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22310 * config/arm/iterators.md (V_elem_l): New mode attribute.
22311 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22312 (vec_extract<mode><V_elem_l>): ... this.
22313 (vec_extractv2di): Renamed to ...
22314 (vec_extractv2didi): ... this.
22315 (vec_init<mode>): Renamed to ...
22316 (vec_init<mode><V_elem_l>): ... this.
22317 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22318 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22319 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22320 Add element mode after gen_vec_extract* calls.
22321 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22322 (vec_init<mode><unitmode>): ... this.
22323 (vec_extract<mode>): Renamed to ...
22324 (vec_extract<mode><unitmode>): ... this.
22325 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22326 (vec_init<mode><unitmode>): ... this.
22327 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22328 (vec_initv2sfsf): ... this.
22329 (vec_extractv2sf): Renamed to ...
22330 (vec_extractv2sfsf): ... this.
22331 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22332 Add element mode after gen_vec_extract* calls.
22333 * config/mips/mips.md (unitmode): New mode iterator.
22334 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22335 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22336 * config/spu/spu.md (inner_l): New mode attribute.
22337 (vec_init<mode>): Renamed to ...
22338 (vec_init<mode><inner_l>): ... this.
22339 (vec_extract<mode>): Renamed to ...
22340 (vec_extract<mode><inner_l>): ... this.
22341 * config/sparc/sparc.md (veltmode): New mode iterator.
22342 (vec_init<VMALL:mode>): Renamed to ...
22343 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22344 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22345 (vec_initv2sisi): ... this.
22346 (vec_initv2sf): Renamed to ...
22347 (vec_initv2sfsf): ... this.
22348 (vec_extractv2sf): Renamed to ...
22349 (vec_extractv2sfsf): ... this.
22350 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22351 (vec_init<mode>): Renamed to ...
22352 (vec_init<mode><VEC_base_l>): ... this.
22353 (vec_extract<mode>): Renamed to ...
22354 (vec_extract<mode><VEC_base_l>): ... this.
22355 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22356 (vec_initv2sfsf): ... this.
22357 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22358 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22359 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22360 gen_vec_init* calls.
22362 2017-08-01 Richard Biener <rguenther@suse.de>
22364 PR tree-optimization/81297
22365 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22366 TREE_OVERFLOW from INTEGER_CSTs.
22368 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22370 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22372 2017-07-31 Carl Love <cel@us.ibm.com>
22374 * config/rs6000/rs6000-c: Add support for built-in functions
22375 vector signed char vec_xl_be (signed long long, signed char *);
22376 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22377 vector signed int vec_xl_be (signed long long, signed int *);
22378 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22379 vector signed long long vec_xl_be (signed long long, signed long long *);
22380 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22381 vector signed short vec_xl_be (signed long long, signed short *);
22382 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22383 vector double vec_xl_be (signed long long, double *);
22384 vector float vec_xl_be (signed long long, float *);
22385 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22386 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22387 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22389 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22390 (altivec_expand_builtin): Add switch statement to call
22391 altivec_expand_xl_be for each builtin.
22392 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22393 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22394 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22395 __builtin_vsx_le_be_v16qi.
22396 * doc/extend.texi: Update the built-in documentation file for the
22397 new built-in functions.
22399 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22402 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22404 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22406 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22408 * config.gcc: Add z14.
22409 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22410 CPU model numbers for z13s and z14.
22411 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22413 * config/s390/s390-opts.h (enum processor_type): Rename
22414 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22415 * config/s390/s390.c (processor_table): Add field for CPU name to
22416 be passed to Binutils.
22417 (s390_asm_output_machine_for_arch): Use the new field in
22418 processor_table for Binutils.
22419 (s390_expand_builtin): Replace arch12 with z14.
22420 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22421 (s390_get_sched_attrmask): Likewise.
22422 (s390_get_unit_mask): Likewise.
22423 * config/s390/s390.opt: Add z14 to processor_type enum.
22425 2017-07-31 Martin Jambor <mjambor@suse.cz>
22428 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22429 regardless of optimization level.
22431 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22432 Martin Liska <mliska@suse.cz>
22434 * predict.def: Remove old comment and adjust probability.
22435 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22436 PREDICT statements.
22438 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22441 * config/i386/i386.c (ix86_function_naked): New function.
22442 (ix86_can_use_return_insn_p): Return false for naked functions.
22443 (ix86_expand_prologue): Skip prologue for naked functions.
22444 (ix86_expand_epilogue): Skip epilogue for naked functions
22445 and emit trap instruction.
22446 (ix86_warn_func_return): New function.
22447 (ix86_attribute_table): Add "naked" attribute specification.
22448 (TARGET_WARN_FUNC_RETURN): Define.
22449 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22451 2017-07-31 Martin Liska <mliska@suse.cz>
22453 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22454 (dump_gimple_bb_header): Always dump BB info.
22455 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22457 2017-07-31 Martin Liska <mliska@suse.cz>
22460 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22461 also with current_function_decl non-null equality.
22463 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22466 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22467 change type to the element type, instead add eltype variable and
22468 use it where we are interested in the element type.
22470 PR tree-optimization/81603
22471 * ipa-polymorphic-call.c
22472 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22473 offset arithmetic in offset_int, bail out if the resulting bit offset
22474 doesn't fit into shwi.
22476 2017-07-31 Martin Liska <mliska@suse.cz>
22478 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22479 (gimplify_save_expr): Fix comment.
22481 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22484 * config/i386/i386.c (ix86_function_arg): Update arguments for
22486 (ix86_compute_frame_layout): Set the initial stack offset to
22487 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22488 INCOMING_FRAME_SP_OFFSET.
22489 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22490 stack before exception handler returns.
22491 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22492 the 'ERROR_CODE' for exception handler.
22494 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22496 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22497 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22498 (ASM_OUTPUT_REG_POP): Ditto.
22499 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22500 instead of asm_fprintf to output pure string.
22502 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22504 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22505 to imported_module_or_decl hook.
22506 (debug_nothing_tree_tree_tree_bool): Remove.
22507 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22508 * debug.c (do_nothing_debug_hooks): Use
22509 debug_nothing_tree_tree_tree_bool_bool instead of
22510 debug_nothing_tree_tree_tree_bool.
22511 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22512 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22513 * sdbout.c (sdb_debug_hooks): Likewise.
22514 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22515 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22517 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22518 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22519 attribute, don't add anything.
22521 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22523 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22524 (fold_build2_stat_loc): Likewise.
22525 (fold_build3_stat_loc): Likewise.
22526 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22527 (fold_build1_loc): Remove macro.
22528 (fold_build2_loc): Likewise.
22529 (fold_build3_loc): Likewise.
22531 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22533 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22534 (gimple_build_debug_bind_source_stat): Likewise.
22535 * gimple.h (gimple_build_debug_bind): Remove macro.
22536 (gimple_build_debug_bind_source): Likewise.
22538 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22540 * bitmap.c (bitmap_alloc): Adjust.
22541 (bitmap_gc_alloc): Likewise.
22542 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22544 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22546 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22547 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22548 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22549 (bitmap_gc_alloc_stat): Likewise.
22550 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22552 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22554 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22555 * rtl.h (shallow_copy_rtx): Remove macro.
22557 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22559 * emit-rtl.c (gen_raw_REG): Adjust.
22560 * gengenrtl.c (gendef): Likewise.
22561 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22562 * rtl.h (rtx_alloc): Remove macro.
22564 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22566 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22567 (build_tree_list_stat): Likewise.
22568 * tree.h (build_tree_list): Remove macro.
22569 (build_tree_list_vec): Likewise.
22571 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22573 * tree.c (make_vector_stat): Remove _stat from name.
22574 (build_vector_stat): Likewise.
22575 * tree.h (make_vector_stat): Remove macro.
22576 (build_vector_stat): Likewise.
22578 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22580 * tree.h (build_var_debug_value): Remove prototype.
22582 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22584 * tree.c (tree_cons_stat): Remove _stat from name.
22585 * tree.h (tree_cons): Remove macro.
22587 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22589 * tree.c (build_vl_exp_stat): Remove _stat from name.
22590 * tree.h (build_vl_exp): Remove macro.
22592 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22594 * tree.c (build_decl_stat): Remove _stat from name.
22595 * tree.h (build_decl): Remove macro.
22597 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22599 * gimple.c (gimple_build_with_ops_stat): Adjust.
22600 (gimple_alloc_stat): Remove _stat from name.
22601 * gimple.h (gimple_alloc): Remove macro.
22603 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22605 * tree.c (make_tree_vec_stat): Remove _stat from name.
22606 (grow_tree_vec_stat): Likewise.
22607 * tree.h (make_tree_vec_stat): Adjust prototype.
22608 (grow_tree_vec_stat): Likewise.
22609 (make_tree_vec): Remove macro.
22610 (grow_tree_vec): Likewise.
22612 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22614 * fold-const.c (fold_build1_stat_loc): Adjust.
22615 (fold_build2_stat_loc): Likewise.
22616 (fold_build3_stat_loc): Likewise.
22617 * tree.c (build0_stat): Remove _stat from name.
22618 (build1_stat): Likewise.
22619 (build2_stat): Likewise.
22620 (build3_stat): Likewise.
22621 (build4_stat): Likewise.
22622 (build5_stat): Likewise.
22623 * tree.h (build1_loc): Remove macro, and rename _stat function
22625 (build2_loc): Likewise.
22626 (build3_loc): Likewise.
22627 (build4_loc): Likewise.
22628 (build5_loc): Likewise.
22630 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22632 * tree.c (make_int_cst_stat): Remove _stat from name.
22633 * tree.h (make_int_cst_stat): Adjust prototype.
22634 (make_int_cst): Remove macro.
22636 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22638 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22639 * tree.h (make_tree_binfo_stat): Adjust prototype.
22640 (make_tree_binfo): Remove.
22642 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22644 * tree.c (copy_node_stat): Rename to copy_node.
22645 (build_distinct_type_copy): Adjust.
22646 * tree.h (copy_node_stat): Adjust prototype.
22647 (copy_node): Remove macro.
22649 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22651 * tree.c (make_node_stat): rename to make_node.
22652 (build_tree_list_stat): Adjust.
22653 (build0_stat): Likewise.
22654 (build2_stat): Likewise.
22655 (build3_stat): Likewise.
22656 (build4_stat): Likewise.
22657 (build5_stat): Likewise.
22658 (build_decl_stat): Likewise.
22659 * tree.h (make_node_stat): Adjust prototype.
22660 (make_node): remove macro.
22662 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22664 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22665 (PPC_FEATURE2_SCV): Likewise.
22666 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22668 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22670 * config/aarch64/aarch64.c
22671 (aarch64_internal_mov_immediate): Add new special pattern.
22672 * config/aarch64/aarch64.md (*movdi_aarch64):
22673 Add reg/32bit const mov case.
22675 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22676 Richard Sandiford <richard.sandiford@linaro.org>
22678 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22679 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22680 Add integer and movi cases.
22681 (movi-split-hf-df-sf split, fp16): New.
22682 (enabled): Added TARGET_FP_F16INST.
22683 * config/aarch64/iterators.md (GPF_HF): New.
22684 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22686 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22688 * config/aarch64/aarch64.c
22689 (aarch64_simd_container_mode): Add prototype.
22690 (aarch64_expand_mov_immediate): Add HI support.
22691 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22692 (aarch64_can_const_movi_rtx_p): New.
22693 (aarch64_preferred_reload_class):
22694 Remove restrictions of using FP registers for certain SIMD operations.
22695 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22696 (aarch64_valid_floating_const): Add integer move validation.
22697 (aarch64_simd_imm_scalar_p): Remove.
22698 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22699 (aarch64_legitimate_constant_p): Expand list of supported cases.
22700 * config/aarch64/aarch64-protos.h
22701 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22702 (aarch64_reinterpret_float_as_int): New.
22703 (aarch64_simd_imm_scalar_p): Remove.
22704 * config/aarch64/constraints.md (Uvi): New.
22705 (Dd): Split into Ds and new Dd.
22706 * config/aarch64/aarch64.md (*movsi_aarch64):
22708 (*movdi_aarch64): Add SIMD mov case.
22710 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22712 * tree-predcom.c: (struct chain): Handle store-store chain in which
22713 stores for elimination only store loop invariant values.
22714 (execute_pred_commoning_chain): Ditto.
22715 (prepare_initializers_chain_store_elim): Ditto.
22716 (prepare_finalizers): Ditto.
22717 (is_inv_store_elimination_chain): New function.
22718 (initialize_root_vars_store_elim_1): New function.
22720 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22722 * tree-predcom.c: Revise general description of the pass.
22723 (enum chain_type): New enum type for store elimination.
22724 (struct chain): New field supporting store elimination.
22725 (struct component): Ditto.
22726 (dump_chain): Dump store-stores chain.
22727 (release_chain): Release resources.
22728 (split_data_refs_to_components): Compute and create component
22729 contains only stores for elimination.
22730 (get_chain_last_ref_at): New function.
22731 (make_invariant_chain): Initialization.
22732 (make_rooted_chain): Specify chain type in parameter and record it.
22733 (add_looparound_copies): Skip for store-stores chain.
22734 (determine_roots_comp): Compute type of chain and pass it to
22736 (initialize_root_vars_store_elim_2): New function.
22737 (finalize_eliminated_stores): New function.
22738 (remove_stmt): Handle store for elimination.
22739 (execute_pred_commoning_chain): Execute predictive commoning on
22740 store-store chains.
22741 (determine_unroll_factor): Skip unroll for store-stores chain.
22742 (prepare_initializers_chain_store_elim): New function.
22743 (prepare_initializers_chain): Hanlde store-store chain.
22744 (prepare_finalizers_chain, prepare_finalizers): New function.
22745 (tree_predictive_commoning_loop): Return integer value indicating
22746 if loop is unrolled or lcssa form is corrupted.
22747 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22749 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22751 * tree-predcom.c (initialize_root): Delete.
22752 (execute_pred_commoning_chain): Initialize root vars and replace
22753 reference of non-combined chain directly, rather than call above
22756 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22758 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22759 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22761 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22763 * tree-predcom.c (struct chain): New field init_seq.
22764 (release_chain): Release init_seq.
22765 (prepare_initializers_chain): Record intialization stmts in above
22767 (insert_init_seqs): New function.
22768 (tree_predictive_commoning_loop): Call insert_init_seqs.
22770 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22772 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22774 2017-07-28 Richard Biener <rguenther@suse.de>
22776 * match.pd: Remove superfluous :c.
22777 * genmatch.c (simplify::id): Add member.
22778 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22780 (current_id): New global.
22781 (dt_node::parent): Move from ...
22782 (dt_operand::parent): ... here. Add for_id member.
22783 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22784 (decision_tree::find_node): Relax order requirement when
22785 merging DT_TRUE nodes to ones inbetween the current simplify
22786 and the one we try to merge with. Add diagnostic whenever
22787 we need to enforce pattern order by not merging.
22788 (decision_tree::insert): Set current_id.
22789 (decision_tree::print_node): Dump parent node and for_id.
22790 (parser::last_id): Add member.
22791 (parser::push_simplify): Assign unique id.
22792 (parser::parser): Initialize last_id.
22794 2017-07-28 Martin Liska <mliska@suse.cz>
22797 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22798 gimple_build_debug_bind.
22800 2017-07-28 Richard Biener <rguenther@suse.de>
22802 PR tree-optimization/81502
22803 * match.pd: Add pattern combining BIT_INSERT_EXPR with
22805 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
22807 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
22808 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
22809 for BIT_FIELD_REF args.
22810 * fold-const.c (make_bit_field_ref): Likewise.
22811 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
22813 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22816 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
22817 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
22818 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
22819 Or it into SANITIZER_UNDEFINED.
22820 * ubsan.c: Include gimple-fold.h and varasm.h.
22821 (ubsan_expand_ptr_ifn): New function.
22822 (instrument_pointer_overflow): New function.
22823 (maybe_instrument_pointer_overflow): New function.
22824 (instrument_object_size): Formatting fix.
22825 (pass_ubsan::execute): Call instrument_pointer_overflow
22826 and maybe_instrument_pointer_overflow.
22827 * internal-fn.c (expand_UBSAN_PTR): New function.
22828 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22829 * sanitizer.def (__ubsan_handle_pointer_overflow,
22830 __ubsan_handle_pointer_overflow_abort): New builtins.
22831 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22832 * internal-fn.def (UBSAN_PTR): New internal function.
22833 * opts.c (sanitizer_opts): Add pointer-overflow.
22834 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22835 * fold-const.c (build_range_check): Compute pointer range check in
22836 integral type if pointer arithmetics would be needed. Formatting
22839 2017-07-28 Martin Liska <mliska@suse.cz>
22842 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22843 parameters that are of a variable-length.
22845 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22847 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22849 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22850 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22851 (CRT_CALL_STATIC_FUNCTION): Likewise.
22852 (ASM_DEFAULT_SPEC): New define.
22853 (ASM_SPEC32): Likewise.
22854 (ASM_SPEC64): Likewise.
22855 (ASM_SPEC_COMMON): Likewise.
22856 (ASM_SPEC): Likewise.
22857 (INVALID_64BIT): Likewise.
22858 (LINK_OS_DEFAULT_SPEC): Likewise.
22859 (LINK_OS_SPEC32): Likewise.
22860 (LINK_OS_SPEC64): Likewise.
22861 (POWERPC_LINUX): Likewise.
22862 (PTRDIFF_TYPE): Likewise.
22863 (RESTORE_FP_PREFIX): Likewise.
22864 (RESTORE_FP_SUFFIX): Likewise.
22865 (SAVE_FP_PREFIX): Likewise.
22866 (SAVE_FP_SUFFIX): Likewise.
22867 (SIZE_TYPE): Likewise.
22868 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
22869 (TARGET_64BIT): Likewise.
22870 (TARGET_64BIT): Likewise.
22871 (TARGET_AIX): Likewise.
22872 (WCHAR_TYPE_SIZE): Likewise.
22873 (WCHAR_TYPE): Undefine.
22874 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
22875 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
22876 (CPP_OS_RTEMS_SPEC): Delete.
22877 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
22878 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
22880 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
22882 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22884 PR tree-optimization/81578
22885 * tree-parloops.c (build_new_reduction): Bail out if
22886 reduction_code isn't one of the standard OpenMP reductions.
22887 Move the details printing after that decision.
22889 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22891 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22892 related to reload_in_progress.
22893 (splat_input_operand): Likewise.
22894 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22896 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22898 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22899 (TARGET_INSTANTIATE_DECLS): Likewise.
22900 (legitimate_indexed_address_p): Delete reload_in_progress code.
22901 (rs6000_debug_legitimate_address_p): Likewise.
22902 (rs6000_eliminate_indexed_memrefs): Likewise.
22903 (rs6000_emit_le_vsx_store): Likewise.
22904 (rs6000_emit_move_si_sf_subreg): Likewise.
22905 (rs6000_emit_move): Likewise.
22906 (register_to_reg_type): Likewise.
22907 (rs6000_pre_atomic_barrier): Likewise.
22908 (rs6000_machopic_legitimize_pic_address): Likewise.
22909 (rs6000_allocate_stack_temp): Likewise.
22910 (rs6000_address_for_fpconvert): Likewise.
22911 (rs6000_address_for_altivec): Likewise.
22912 (rs6000_secondary_memory_needed_rtx): Delete function.
22913 (rs6000_check_sdmode): Likewise.
22914 (rs6000_alloc_sdmode_stack_slot): Likewise.
22915 (rs6000_instantiate_decls): Likewise.
22916 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22917 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22918 Delete reload_in_progress.
22919 (*vec_reload_and_plus_<mptrsize>): Likewise.
22920 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22921 (vsx_div_v2di): Likewise.
22922 (vsx_udiv_v2di): Likewise.
22924 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22926 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22927 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22928 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22929 (rs6000_debug_reg_global): Delete print of LRA status.
22930 (rs6000_option_override_internal): Delete dead LRA related code.
22931 (rs6000_lra_p): Delete function.
22932 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22934 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22936 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22937 * config/riscv/rtems.h: New file.
22939 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22940 Sudakshina Das <sudi.das@arm.com>
22942 * config/aarch64/aarch64.md
22943 (define_split for and<mode>3nr_compare): Move
22944 non aarch64_logical_operand to a register.
22945 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22946 register immediate operand to a register.
22947 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22949 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22951 PR middle-end/81564
22952 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22954 2017-07-27 Richard Biener <rguenther@suse.de>
22956 PR tree-optimization/81573
22957 PR tree-optimization/81494
22958 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22959 multi defuse cycle case.
22961 2017-07-27 Richard Biener <rguenther@suse.de>
22963 PR tree-optimization/81571
22964 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22967 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22969 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22970 earlier and only if MASK_FPU is set. Adjust formatting.
22972 2017-07-27 Martin Liska <mliska@suse.cz>
22974 * opt-functions.awk: Add validation of value of Init.
22975 * optc-gen.awk: Pass new argument.
22977 2017-07-27 Martin Liska <mliska@suse.cz>
22979 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22980 Fix wrong condition.
22982 2017-07-27 Martin Liska <mliska@suse.cz>
22984 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22985 BBs and edges seen by autoFDO.
22987 2017-07-27 Richard Biener <rguenther@suse.de>
22989 PR tree-optimization/81502
22990 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22991 with incompatible but same sized type.
22992 (execute_update_addresses_taken): Likewise.
22994 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22996 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22997 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22999 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23002 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
23003 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
23004 Change s_operand to memory_operand.
23006 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
23008 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23009 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23010 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23011 Emit instructions rather than returning an expression. Handle TFmode
23012 and KFmode by casting to TImode.
23013 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23014 (rs6000_emit_le_vsx_store): Likewise.
23015 * config/rs6000/vsx.md (VSX_TI): New iterator.
23016 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23017 (*vsx_le_undo_permute_<mode>): Likewise.
23018 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23019 emit the split sequence.
23020 (*vsx_le_perm_store_<mode>): Likewise.
23022 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23024 PR tree-optimization/81555
23025 PR tree-optimization/81556
23026 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23027 if true, force CHANGED for the recursive invocation.
23028 (reassociate_bb): Remember original length of ops array, pass
23029 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23031 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23032 attributes for noipa attribute. For naked attribute use
23033 lookup_attribute first before lookup_attribute_spec.
23034 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23036 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23037 for functions with noipa attribute.
23038 (cgraph_externally_visible_p): Return true for functions with noipa
23040 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23041 for functions with noipa attribute.
23042 * doc/extend.texi: Document noipa function attribute.
23043 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23044 also for functions with noipa attribute.
23045 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23047 2017-07-26 Andrew Pinski <apinski@cavium.com>
23049 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23050 vec_unalign_load_cost and vec_unalign_store_cost.
23052 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23054 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23055 -mvsx-small-integer option.
23056 (ISA_3_0_MASKS_IEEE): Likewise.
23057 (OTHER_VSX_VECTOR_MASKS): Likewise.
23058 (POWERPC_MASKS): Likewise.
23059 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23060 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23061 code, only testing for DImode being allowed in non-VSX floating
23063 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23064 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23066 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23067 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23068 TARGET_P8_VECTOR test.
23069 (rs6000_secondary_reload_simple_move): Likewise.
23070 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23071 since TARGET_P9_VECTOR was already tested.
23072 (rs6000_opt_masks): Remove -mvsx-small-integer.
23073 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23074 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23076 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23077 test for TARGET_VEXTRACTUB was used, and that uses
23079 (p9 extract splitter): Likewise.
23080 (vsx_extract_<mode>_di_p9): Likewise.
23081 (vsx_extract_<mode>_store_p9): Likewise.
23082 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23083 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23084 the elimination of TARGET_VSX_SMALL_INTEGER.
23085 (vsx_extract_<mode>_p8): Likewise.
23086 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23087 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23088 (vsx_set_<mode>_p9): Likewise.
23089 (vsx_set_v4sf_p9): Likewise.
23090 (vsx_set_v4sf_p9_zero): Likewise.
23091 (vsx_insert_extract_v4sf_p9): Likewise.
23092 (vsx_insert_extract_v4sf_p9_2): Likewise.
23093 * config/rs6000/rs6000.md (sign extend splitter): Change
23094 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23095 (floatsi<mode>2_lfiwax_mem): Likewise.
23096 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23097 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23098 since a test for TARGET_P9_VECTOR was used.
23099 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23100 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23101 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23102 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23103 TARGET_P8_VECTOR test.
23104 (fix_trunc<mode>si2_stfiwx): Likewise.
23105 (fix_trunc<mode>si2_internal): Likewise.
23106 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23107 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23109 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23110 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23111 TARGET_P8_VECTOR test.
23112 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23113 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23114 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23116 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23117 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23118 since a test for TARGET_P9_VECTOR was used.
23119 (splitter for loading small constants): Likewise.
23121 2017-07-26 Andrew Pinski <apinski@cavium.com>
23123 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23126 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23129 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23130 (fp_valid_at): Likewise.
23132 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23134 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23135 (qdf24xx_addrcost_table): Likewise.
23136 (cortexa57_tunings): Update to use generic_branch_cost.
23137 (cortexa72_tunings): Likewise.
23138 (cortexa73_tunings): Likewise.
23139 (qdf24xx_tunings): Likewise.
23141 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23143 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23144 (thunderx2t99_branch_cost): Likewise.
23145 (cortexa35_tunings): Update to use generic_branch_cost.
23146 (cortexa53_tunings): Likewise.
23147 (cortexa57_tunings): Likewise.
23148 (cortexa72_tunings): Likewise.
23149 (cortexa73_tunings): Likewise.
23150 (thunderx2t99_tunings): Likewise.
23152 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23154 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23155 (sparc_option_override): Honour MASK_FSMULD.
23156 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23157 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23158 * config/sparc/sparc.opt (mfsmuld): New option.
23159 * doc/invoke.texi (mfsmuld): Document option.
23161 2017-07-26 Marek Polacek <polacek@redhat.com>
23163 PR middle-end/70992
23164 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23166 2017-07-26 Richard Biener <rguenther@suse.de>
23168 * gimple-match-head.c (do_valueize): Return OP if valueize
23170 (get_def): New helper to get at the def stmt of a SSA name
23171 if valueize allows.
23172 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23173 do_valueize to get at the def stmt.
23174 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23176 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23178 PR middle-end/46932
23179 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23181 2017-07-26 Martin Liska <mliska@suse.cz>
23184 * function.c (expand_function_start): Make expansion of
23185 nonlocal_goto_save_area after parm_birth_insn.
23187 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23189 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23190 from all CPU target flags enable members.
23192 2017-07-26 Richard Biener <rguenther@suse.de>
23194 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23195 (decision_tree::gen): Make 'type' const.
23196 (write_predicate): Likewise.
23198 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23200 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23201 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23202 (rs6000_option_override_internal): Likewise.
23203 (rs6000_expand_vector_set): Likewise.
23204 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23205 (TARGET_UPPER_REGS_SF): Likewise.
23206 (TARGET_UPPER_REGS_DI): Likewise.
23207 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23208 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23209 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23210 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23211 (Splitters for DI constants in Altivec registers): Likewise.
23212 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23213 (vsx_set_v4sf_p9): Likewise.
23214 (vsx_set_v4sf_p9_zero): Likewise.
23215 (vsx_insert_extract_v4sf_p9): Likewise.
23216 (vsx_insert_extract_v4sf_p9_2): Likewise.
23218 2017-07-25 Carl Love <cel@us.ibm.com>
23220 * doc/extend.texi: Update the built-in documentation file for the
23221 existing built-in functions
23222 vector signed char vec_cnttz (vector signed char);
23223 vector unsigned char vec_cnttz (vector unsigned char);
23224 vector signed short vec_cnttz (vector signed short);
23225 vector unsigned short vec_cnttz (vector unsigned short);
23226 vector signed int vec_cnttz (vector signed int);
23227 vector unsigned int vec_cnttz (vector unsigned int);
23228 vector signed long long vec_cnttz (vector signed long long);
23229 vector unsigned long long vec_cnttz (vector unsigned long long);
23231 2017-07-25 Andrew Pinski <apinski@cavium.com>
23233 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23234 accesses where the use is for the first operand of a BIT_INSERT.
23236 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23239 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23240 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23242 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23244 * config/i386/gstabs.h: Delete.
23245 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23247 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23249 * config/i386/i386.c (ix86_decompose_address): Do not check for
23250 register RTX when looking at index_reg or base_reg.
23251 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23253 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23255 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23256 to update EH info here.
23258 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23260 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23262 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23264 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23266 2017-07-25 Torsten Duwe <duwe@suse.de>
23268 * common.opt: Introduce -fpatchable-function-entry
23269 command line option, and its variables function_entry_patch_area_size
23270 and function_entry_patch_area_start.
23271 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23272 including a two-value parser.
23273 * target.def (print_patchable_function_entry): New target hook.
23274 * targhooks.h (default_print_patchable_function_entry): New function.
23275 * targhooks.c (default_print_patchable_function_entry): Likewise.
23276 * toplev.c (process_options): Switch off IPA-RA if
23277 patchable function entries are being generated.
23278 * varasm.c (assemble_start_function): Look at the
23279 patchable-function-entry command line switch and current
23280 function attributes and maybe generate NOP instructions by
23281 calling the print_patchable_function_entry hook.
23282 * doc/extend.texi: Document patchable_function_entry attribute.
23283 * doc/invoke.texi: Document -fpatchable_function_entry
23284 command line option.
23285 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23287 * doc/tm.texi: Re-generate.
23289 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23292 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23293 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23295 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23297 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23299 2017-07-25 Richard Biener <rguenther@suse.de>
23301 PR tree-optimization/81455
23302 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23303 not walk in cycles when looking for guards.
23305 2017-07-25 Richard Biener <rguenther@suse.de>
23307 PR tree-optimization/81529
23308 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23309 when optimizing backedge uses.
23311 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23313 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23315 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23316 to dl_section_ref. On AIX, append an expression to subtract
23317 the size of the section length to dl_section_ref.
23319 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23321 * configure.ac: If any of the config.* scripts fail, exit 1.
23322 * configure: Regenerate.
23324 2017-07-25 Richard Biener <rguenther@suse.de>
23326 PR middle-end/81546
23327 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23328 of immediate uses, be more verbose on errors.
23330 2017-07-25 Richard Biener <rguenther@suse.de>
23332 PR tree-optimization/81510
23333 * tree-vect-loop.c (vect_is_simple_reduction): When the
23334 reduction stmt is not inside the loop bail out.
23336 2017-07-25 Richard Biener <rguenther@suse.de>
23338 PR tree-optimization/81303
23339 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23340 profitability check against LOOP_VINFO_NITERSM1.
23342 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23344 * domwalk.c (cmp_bb_postorder): Simplify.
23345 (sort_bbs_postorder): New function. Use it...
23346 (dom_walker::walk): ...here to optimize common cases.
23348 2017-07-25 Martin Liska <mliska@suse.cz>
23351 * ipa-visibility.c (function_and_variable_visibility): Make the
23352 redirection just on target that supports aliasing.
23353 Fix GNU coding style.
23355 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23358 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23360 * config/arm/rtems.h: Likewise.
23361 * config/bfin/rtems.h: Likewise.
23362 * config/i386/rtemself.h: Likewise.
23363 * config/lm32/rtems.h: Likewise.
23364 * config/m32c/rtems.h: Likewise.
23365 * config/m68k/rtemself.h: Likewise.
23366 * config/microblaze/rtems.h: Likewise.
23367 * config/mips/rtems.h: Likewise.
23368 * config/moxie/rtems.h: Likewise.
23369 * config/nios2/rtems.h: Likewise.
23370 * config/powerpcspe/rtems.h: Likewise.
23371 * config/rs6000/rtems.h: Likewise.
23372 * config/rtems.h: Likewise.
23373 * config/sh/rtems.h: Likewise.
23374 * config/sh/rtemself.h: Likewise.
23375 * config/sparc/rtemself.h: Likewise.
23377 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23380 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23381 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23382 * tree-ssa-structalias.c (alias_get_name): Same.
23384 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23387 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23388 instructions if no du chain is found.
23390 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23392 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23394 2017-07-25 Richard Biener <rguenther@suse.de>
23396 PR middle-end/81505
23397 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23400 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23402 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23403 upper-regs options.
23404 (ISA_2_7_MASKS_SERVER): Likewise.
23405 (ISA_3_0_MASKS_IEEE): Likewise.
23406 (OTHER_P8_VECTOR_MASKS): Likewise.
23407 (OTHER_VSX_VECTOR_MASKS): Likewise.
23408 (POWERPC_MASKS): Likewise.
23409 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23410 duplicate list of options.
23411 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23412 explicit -mupper-regs options.
23413 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23414 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23415 alias for -mupper-regs-df.
23416 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23417 (rs6000_init_hard_regno_mode_ok): Likewise.
23418 (rs6000_option_override_internal): Likewise.
23419 (rs6000_opt_masks): Likewise.
23420 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23421 options in terms of whether -mvsx or -mpower8-vector was used.
23422 (TARGET_UPPER_REGS_DI): Likewise.
23423 (TARGET_UPPER_REGS_SF): Likewise.
23424 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23425 -mupper-regs-* options.
23427 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23429 * passes.c (emergency_dump_function): Print some empty lines and a
23430 header before the RTL dump.
23432 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23434 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23436 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23439 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23440 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23442 2017-07-24 Carl Love <cel@us.ibm.com>
23444 * config/rs6000/rs6000-c.c: Add support for built-in functions
23445 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23446 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23447 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23448 vec_extract_fp_from_shortl): Add defines for the two builtins.
23449 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23450 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23452 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23453 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23454 * doc/extend.texi: Update the built-in documentation file for the
23455 new built-in function.
23457 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23460 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23462 * doc/generic.texi: Likewise.
23463 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23464 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23466 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23468 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23469 (aarch64_mls_elt_merge<mode>): Likewise.
23471 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23473 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23474 having __cxa_atexit.
23476 2017-07-23 Michael Collison <michael.collison@arm.com>
23478 * config/arm/arm.c (arm_option_override): Deprecate
23479 use of -mstructure-size-boundary.
23480 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23481 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23483 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23486 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23487 Reduce cost estimate for direct moves.
23489 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23492 * config/i386/i386.c (ix86_option_override_internal): Disable
23493 BMI, BMI2 and TBM instructions for -m16.
23495 2017-07-21 Carl Love <cel@us.ibm.com>
23497 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23498 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23499 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23500 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23501 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23502 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23503 VMULOSW): New enum "unspec" values.
23504 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23505 altivec_vmulosw): New patterns.
23506 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23507 VMULOSW): Add definitions.
23509 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23511 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23512 (qdf24xx): Likewise.
23513 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23514 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23515 (AARCH64_FL_V8_1): Renumber.
23516 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23517 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23518 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23519 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23520 rdma to feature modifiers list.
23522 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23524 PR middle-end/56727
23525 * ipa-visibility (function_and_variable_visibility): Convert
23526 recursive PLT call to direct call if appropriate.
23528 2017-07-21 Andrew Pinski <apinski@cavium.com>
23530 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23531 operand 1 to see if the types precision matches.
23532 * fold-const.c (operand_equal_p): Likewise.
23534 2017-07-21 Richard Biener <rguenther@suse.de>
23536 PR tree-optimization/81303
23537 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23538 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23539 (vect_peeling_hash_get_lowest_cost): Adjust.
23540 (vect_enhance_data_refs_alignment): Likewise. Use
23541 vect_get_peeling_costs_all_drs to compute the penalty for no
23542 peeling to match up costs.
23544 2017-07-21 Richard Biener <rguenther@suse.de>
23546 PR tree-optimization/81500
23547 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23548 we didn't identify a reduction path.
23550 2017-07-21 Tom de Vries <tom@codesourcery.com>
23551 Cesar Philippidis <cesar@codesourcery.com>
23553 PR gcov-profile/81442
23554 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23557 2017-07-21 Tom de Vries <tom@codesourcery.com>
23560 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23562 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23563 nvptx_override_options_after_change.
23565 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23567 * dwarf2out.c (output_file_names): Avoid double testing for
23568 dwarf_version >= 5.
23570 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23572 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23574 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23576 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23578 (try_crossjump_to_edge): Do not punt on partitioned functions.
23580 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23582 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23583 Put all BBs reachable only via paths crossing cold region to cold
23585 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23587 2016-07-21 Richard Biener <rguenther@suse.de>
23589 PR tree-optimization/81303
23590 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23591 into account prologue and epilogue iterations when raising
23592 min_profitable_iters to sth at least covering one vector iteration.
23594 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23596 * config/arm/arm.c (arm_test_cpu_arch_dat):
23599 2017-07-20 Nathan Sidwell <nathan@acm.org>
23601 Remove TYPE_METHODS.
23602 * tree.h (TYPE_METHODS): Delete.
23603 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23604 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23605 (dbxout_type_methods): Scan TYPE_FIELDS.
23606 (dbxout_type): Don't check TYPE_METHODS here.
23607 * function.c (use_register_for_decl): Always ignore register for
23608 class types when not optimizing.
23609 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23610 * tree.c (free_lang_data_in_type): Stitch out member functions and
23611 templates from TYPE_FIELDS.
23612 (build_distinct_type_copy, verify_type_variant,
23613 verify_type): Member fns are on TYPE_FIELDS.
23614 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23615 * tree-pretty-print.c (dump_generic_node): Likewise.
23617 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23620 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23621 V2TImode and V4TImode.
23622 (ix86_expand_vector_extract): Likewise.
23623 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23624 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23625 (ssescalarmode): Handle V4TImode and V2TImode.
23626 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23627 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23628 (VEXTRACTI128_MODE): New mode iterator.
23629 (splitter for *vec_extractv?ti first element): New.
23630 (VEC_INIT_MODE): New mode iterator.
23631 (vec_init<mode>): Consolidate 3 expanders into one using
23632 VEC_INIT_MODE mode iterator.
23634 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23636 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23637 non_spilled_static_chain_regno_p.
23639 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23641 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23643 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23645 * bb-reorder.c (connect_traces): Allow copying of blocks within
23648 2017-07-20 Richard Biener <rguenther@suse.de>
23650 * gimple.h (gimple_phi_result): Add gphi * overload.
23651 (gimple_phi_result_ptr): Likewise.
23652 (gimple_phi_arg): Likewise. Adjust index assert to only
23653 allow actual argument accesses rather than all slots available
23655 (gimple_phi_arg_def): Add gphi * overload.
23656 * tree-phinodes.c (make_phi_node): Initialize only actual
23658 (resize_phi_node): Clear memory not covered by old node,
23659 do not initialize excess argument slots.
23660 (reserve_phi_args_for_new_edge): Initialize new argument slot
23663 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23665 PR tree-optimization/81388
23667 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23669 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23670 by removing computation of may_be_zero.
23672 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23673 Tom de Vries <tom@codesourcery.com>
23675 PR middle-end/81030
23676 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23677 when gimple level profile disagrees with what RTL expander did.
23679 2017-07-20 Richard Biener <rguenther@suse.de>
23681 PR tree-optimization/61171
23682 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23683 (vect_analyze_stmt): Add slp instance parameter.
23684 (vectorizable_reduction): Likewise.
23685 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23686 (vect_is_simple_reduction): Deal with chains not detected
23687 as SLP reduction chain, specifically not properly associated
23688 chains containing a mix of plus/minus.
23689 (get_reduction_op): Remove.
23690 (get_initial_defs_for_reduction): Simplify, pass in whether
23691 this is a reduction chain, pass in the SLP node for the PHIs.
23692 (vect_create_epilog_for_reduction): Get the SLP instance as
23694 (vectorizable_reduction): Get the SLP instance as arg.
23695 During analysis remember the SLP node with the PHIs in the
23696 instance. Simplify getting at the vectorized reduction PHIs.
23697 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23698 through SLP instance.
23699 (vect_slp_analyze_operations): Likewise.
23700 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23701 (vect_transform_stmt): Likewise.
23703 2017-07-20 Tom de Vries <tom@codesourcery.com>
23705 PR tree-optimization/81489
23706 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23707 read of phi arg location to before loop that modifies phi.
23709 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23711 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23714 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23716 PR middle-end/81331
23717 * except.c (execute): Fix ordering issue.
23719 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23721 PR rtl-optimization/81423
23722 * combine.c (make_compound_operation_int): Don't try to optimize
23723 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23725 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23727 PR rtl-optimization/81423
23728 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23729 with a constant that is -1 in the truncated to mode.
23731 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23733 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23734 (determine_unlikely_bbs): ... here.
23735 * predict.h (propagate_unlikely_bbs_forward): Declare.
23736 * cfgexpand.c (pass_expand::execute): Use it.
23737 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23739 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23740 propagate_unlikely_bbs_forward.
23742 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23744 PR middle-end/81331
23745 * except.c (maybe_add_nop_after_section_switch): New function.
23748 2017-07-19 Tom de Vries <tom@codesourcery.com>
23750 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23752 2017-07-19 Tom de Vries <tom@codesourcery.com>
23754 * gimple.h (gimple_phi_arg): Make assert more strict.
23756 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23758 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23759 * config/rs6000/mmintrin.h: New file.
23760 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23762 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23764 PR tree-optimization/81346
23765 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23767 2017-07-19 Tom de Vries <tom@codesourcery.com>
23769 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23771 2017-07-19 Tom de Vries <tom@codesourcery.com>
23773 * config/nvptx/nvptx-modes.def: Add V2DImode.
23774 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23775 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23776 (nvptx_output_mov_insn): Handle lack of mov.b128.
23777 (nvptx_print_operand): Handle 'H' and 'L' codes.
23778 (nvptx_vector_mode_supported): Allow V2DImode.
23779 (nvptx_preferred_simd_mode): New function.
23780 (nvptx_data_alignment): New function.
23781 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23782 nvptx_preferred_simd_mode.
23783 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23785 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23787 2017-07-19 Tom de Vries <tom@codesourcery.com>
23789 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23790 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23791 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23792 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23793 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23794 (mov<VECIM>_insn): New define_insn.
23795 (define_expand "mov<VECIM>): New define_expand.
23797 2017-07-19 Tom de Vries <tom@codesourcery.com>
23799 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
23801 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23803 PR tree-optimization/81346
23804 * fold-const.h (fold_div_compare, range_check_type): Declare.
23805 * fold-const.c (range_check_type): New function.
23806 (build_range_check): Use range_check_type.
23807 (fold_div_compare): No longer static, rewritten into
23808 a match.pd helper function.
23809 (fold_comparison): Don't call fold_div_compare here.
23810 * match.pd (X / C1 op C2): New optimization using fold_div_compare
23811 as helper function.
23813 2017-07-19 Nathan Sidwell <nathan@acm.org>
23815 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
23816 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
23817 * tree.c (find_decls_types_r, verify_type): Use
23818 TYPE_{MIN,MAX}_VALUE_RAW.
23819 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
23820 (hash_tree): Likewise.
23821 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
23823 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
23826 2017-07-18 Tom de Vries <tom@codesourcery.com>
23828 PR middle-end/81464
23829 * omp-expand.c (expand_omp_for_static_chunk): Handle
23830 equal-argument loop exit phi.
23832 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23835 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23836 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23837 operand 2 predicate.
23838 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23839 operand 2 predicate.
23840 (ror,rol -> rorx splitters): Use const_int_operand as
23841 operand 2 predicate.
23843 2017-06-18 Richard Biener <rguenther@suse.de>
23845 PR tree-optimization/81410
23846 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23847 the gap in the ! slp_perm SLP case after each group.
23849 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23851 PR middle-end/81463
23852 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
23855 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23857 PR middle-end/81462
23858 * predict.c (set_even_probabilities): Cleanup; do not affect
23859 probabilities that are already known.
23860 (combine_predictions_for_bb): Call even when count is set.
23862 2017-07-18 Nathan Sidwell <nathan@acm.org>
23864 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
23867 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23870 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
23871 optimization for loop niter analysis.
23873 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
23876 * config/avr/avr.c (avr_optimize_casesi): Don't use
23877 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
23879 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
23881 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
23882 body_cost_vec from _vect_peel_extended_info.
23883 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23884 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23887 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23889 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23891 2017-07-18 Richard Biener <rguenther@suse.de>
23893 PR tree-optimization/80620
23894 PR tree-optimization/81403
23895 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23896 info when re-using a VN table entry.
23898 2017-07-18 Richard Biener <rguenther@suse.de>
23900 PR tree-optimization/81418
23901 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23902 vectype_in. Verify that with lane-reducing reduction operations
23903 we have a single def-use cycle.
23905 2017-07-17 Carl Love <cel@us.ibm.com>
23907 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23909 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23910 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23911 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23912 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23913 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23914 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23915 VMULOSW): New enum "unspec" values.
23916 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23917 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23918 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23919 altivec_vmulosw): New patterns.
23920 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23921 VMULOSW): Add definitions.
23923 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23925 * config/alpha/alpha.c: Include predict.h.
23927 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23929 * tree-vrp.c (compare_assert_loc): Fix comparison function
23930 to return predictable results.
23932 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23934 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23936 (subdi3): Likewise.
23937 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23938 * doc/invoke.texi (mexpand-adddi): Update text.
23940 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23942 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23943 that also clobbers the CC register. The old expand code is moved
23945 (*arc_clzsi2): ... here.
23946 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23947 the CC register. The old expand code is moved to ...
23948 (arc_ctzsi2): ... here.
23950 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23952 * config/arc/arc.opt (mindexed-loads): Use initial value
23953 TARGET_INDEXED_LOADS_DEFAULT.
23954 (mauto-modify-reg): Use initial value
23955 TARGET_AUTO_MODIFY_REG_DEFAULT.
23956 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23957 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23958 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23959 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23961 2017-07-17 Martin Liska <mliska@suse.cz>
23964 * opts.c (finish_options): Do not allow -fgnu-tm
23965 w/ -fsanitize={kernel-,}address. Say sorry.
23967 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23970 * tree-loop-distribution.c (classify_partition): Only assert on
23971 numer of iterations.
23972 (merge_dep_scc_partitions): Delete prameter. Update function call.
23973 (distribute_loop): Remove code handling loop with unknown niters.
23974 (pass_loop_distribution::execute): Skip loop with unknown niters.
23976 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23979 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23980 function sort_partitions_by_post_order.
23982 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23984 PR tree-optimization/81374
23985 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23986 the max index of basic blocks, rather than number of basic blocks.
23988 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23990 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23992 (arc_legitimate_pic_operand_p): Likewise.
23993 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23995 (arc_needs_pcl_p): Likewise.
23996 (arc_legitimate_pc_offset_p): Likewise.
23997 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23998 function is also used in constrains.md.
23999 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
24000 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
24001 PLUS. Only return true/false in known cases, otherwise assert.
24002 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
24003 is already called in arc_legitimate_constant_p.
24004 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
24006 (LEGITIMATE_PIC_OPERAND_P): Use
24007 arc_raw_symbolic_reference_mentioned_p function.
24008 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24013 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24014 Andrew Burgess <andrew.burgess@embecosm.com>
24016 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24017 (arc_return_address_register): New function.
24018 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24019 (arc_handle_fndecl_attribute): Add naked attribute.
24020 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24021 (TARGET_WARN_FUNC_RETURN): Likewise.
24022 (arc_allocate_stack_slots_for_args): New function.
24023 (arc_warn_func_return): Likewise.
24024 (machine_function): Change type fn_type.
24025 (arc_compute_function_type): Consider new naked function type,
24026 change function return type.
24027 (arc_must_save_register): Adapt to handle new
24028 arc_compute_function_type's return type.
24029 (arc_expand_prologue): Likewise.
24030 (arc_expand_epilogue): Likewise.
24031 (arc_return_address_regs): Delete.
24032 (arc_return_address_register): New function.
24033 (arc_epilogue_uses): Use above function.
24034 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24035 (arc_function_type): Change encoding, add naked type.
24036 (ARC_INTERRUPT_P): Change to handle the new encoding.
24037 (ARC_FAST_INTERRUPT_P): Likewise.
24038 (ARC_NORMAL_P): Define.
24039 (ARC_NAKED_P): Likewise.
24040 (arc_compute_function_type): Delete prototype.
24041 * config/arc/arc.md (in_ret_delay_slot): Use
24042 arc_return_address_register function.
24043 (simple_return): Likewise.
24044 (p_return_i): Likewise.
24046 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24048 PR tree-optimization/81428
24049 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24050 can't be built for those types.
24052 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24054 Remove stuff dead since r239246.
24056 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24057 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24058 (avr_inform_devices): Remove dead stuff.
24060 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24062 * config/arm/arm_neon.h: Fix softp typo.
24064 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24066 PR tree-optimization/81365
24067 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24068 aggregate moves onto bb predecessor edges, make sure there are no
24069 loads that could alias the lhs in between the start of bb and the
24072 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24075 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24076 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24077 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24079 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24081 PR tree-optimization/81396
24082 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24083 (init_symbolic_number): Initialize it to 1.
24084 (perform_symbolic_merge): Add n_ops from both operands into the new
24086 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24087 without base_addr as useless if they need more than one operation.
24088 (bswap_replace): Handle !bswap case for NULL base_addr.
24090 2017-07-17 Tom de Vries <tom@codesourcery.com>
24093 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24096 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24098 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24099 conditional builtin define __FIX_LEON3FT_B2BST.
24101 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24103 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24104 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24107 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24109 PR rtl-optimization/81424
24110 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24111 to remove potential trapping from operands if -fnon-call-exceptions.
24113 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24115 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24116 profile_proability for scalling.
24117 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24119 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24121 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24123 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24125 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24126 fixpoint arithmetics.
24128 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24130 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24131 fixpoint arithmetics.
24133 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24135 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24136 fixpoint arithmetics.
24138 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24140 * profile-count.h (profile_probability::from_reg_br_prob_note,
24141 profile_probability::to_reg_br_prob_note): New functions.
24142 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24143 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24144 * predict.c (probability_reliable_p): Update.
24145 (edge_probability_reliable_p): Update.
24146 (br_prob_note_reliable_p): Update.
24147 (invert_br_probabilities): Update.
24148 (add_reg_br_prob_note): New function.
24149 (combine_predictions_for_insn): Update.
24150 * asan.c (asan_clear_shadow): Update.
24151 * cfgbuild.c (compute_outgoing_frequencies): Update.
24152 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24153 (update_br_prob_note): Update.
24154 (rtl_verify_edges): Update.
24155 (purge_dead_edges): Update.
24156 (fixup_reorder_chain): Update.
24157 * emit-rtl.c (try_split): Update.
24158 * ifcvt.c (cond_exec_process_insns): Update.
24159 (cond_exec_process_if_block): Update.
24160 (dead_or_predicable): Update.
24161 * internal-fn.c (expand_addsub_overflow): Update.
24162 (expand_neg_overflow): Update.
24163 (expand_mul_overflow): Update.
24164 * loop-doloop.c (doloop_modify): Update.
24165 * loop-unroll.c (compare_and_jump_seq): Update.
24166 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24167 * predict.h: Update.
24168 * reorg.c (mostly_true_jump): Update.
24170 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24171 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24172 * config/arc/arc.c: (emit_unlikely_jump): Update.
24173 * config/arm/arm.c: (emit_unlikely_jump): Update.
24174 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24175 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24176 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24177 (ix86_print_operand): Update.
24178 (ix86_split_fp_branch): Update.
24179 (predict_jump): Update.
24180 * config/ia64/ia64.c (ia64_print_operand): Update.
24181 * config/mmix/mmix.c (mmix_print_operand): Update.
24182 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24183 (rs6000_expand_split_stack_prologue): Update.
24184 * config/rs6000/rs6000.c: Update.
24185 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24186 (s390_expand_vec_movstr): Update.
24187 (s390_expand_cs_tdsi): Update.
24188 (s390_expand_split_stack_prologue): Update.
24189 * config/sh/sh.c (sh_print_operand): Update.
24190 (expand_cbranchsi4): Update.
24191 (expand_cbranchdi4): Update.
24192 * config/sparc/sparc.c (output_v9branch): Update.
24193 * config/spu/spu.c (get_branch_target): Update.
24194 (ea_load_store_inline): Update.
24195 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24196 * config/tilepro/tilepro.c: Update.
24198 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24200 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24201 (gimplify_save_expr): Likewise.
24203 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24205 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24207 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24209 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24211 * timevar.def (TV_IPA_FNSUMMARY): Define.
24213 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24215 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24216 to back store errata sensitive sequence from being generated.
24217 (sqrtdf2_fix): Likewise.
24219 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24221 * tree-ssa-threadupdate.c (compute_path_counts,
24222 update_joiner_offpath_counts): Use profile_probability.
24224 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24227 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24229 * config/arm/arm-c.c (arm_cpu_builtins): Define
24230 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24232 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24234 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24235 array entries to represent __ieee128 versions of the
24236 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24237 scalar_extract_sig, and scalar_insert_exp built-in functions.
24238 (altivec_resolve_overloaded_builtin): Add special case handling
24239 for the __builtin_scalar_insert_exp function, as represented by
24240 the P9V_BUILTIN_VEC_VSIEDP constant.
24241 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24242 exponent support for __ieee128 argument.
24243 (VSESQP): Add scalar extract signature support for __ieee128
24245 (VSTDCNQP): Add scalar test negative support for __ieee128
24247 (VSIEQP): Add scalar insert exponent support for __int128 argument
24248 with __ieee128 result.
24249 (VSIEQPF): Add scalar insert exponent support for __ieee128
24250 argument with __ieee128 result.
24251 (VSTDCQP): Add scalar test data class support for __ieee128
24253 (VSTDCNQP): Add overload support for scalar test negative with
24254 __ieee128 argument.
24255 (VSTDCQP): Add overload support for scalar test data class
24256 __ieee128 argument.
24257 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24258 UNSPEC_VSX_SXSIGDP.
24259 (UNSPEC_VSX_SIEXPQP): New constant.
24260 (xsxexpqp): New insn for VSX scalar extract exponent quad
24262 (xsxsigqp): New insn for VSX scalar extract significand quad
24264 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24265 precision with floating point argument.
24266 (xststdcqp): New expand for VSX scalar test data class quad
24268 (xststdcnegqp): New expand for VSX scalar test negative quad
24270 (xststdcqp): New insn to match expansions for VSX scalar test data
24271 class quad precision and VSX scalar test negative quad precision.
24272 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24273 special case operand checking to enforce that second operand of
24274 VSX scalar test data class with quad precision argument is a 7-bit
24276 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24277 prototypes and descriptions of __ieee128 versions of
24278 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24279 scalar_test_data_class, and scalar_test_neg built-in functions.
24281 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24283 PR tree-optimization/81162
24284 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24285 replace a negate with an add.
24287 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24289 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24291 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24293 * config/arm/arm-c.c (arm_cpu_builtins): Define
24294 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24296 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24298 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24299 (armv8-r): Set ARM Cortex-R52 as default CPU.
24300 * config/arm/arm-tables.opt: Regenerate.
24301 * config/arm/arm-tune.md: Regenerate.
24302 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24304 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24305 extension for -mcpu=cortex-r52.
24307 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24309 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24310 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24311 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24312 (fp-armv8): Define it as FP_ARMv8 only.
24313 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24314 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24316 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24317 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24318 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24319 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24320 than TARGET_FPU_ARMV8.
24321 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24322 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24323 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24325 * config/arm/neon.md (neon_vrint): Likewise.
24326 (neon_vcvt): Likewise.
24327 (neon_<fmaxmin_op><mode>): Likewise.
24328 (<fmaxmin><mode>3): Likewise.
24329 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24330 * config/arm/predicates.md (arm_cond_move_operator): Check against
24331 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24333 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24335 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24336 to top of function.
24338 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24340 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24341 loop in comment with memset.
24343 2017-07-14 Martin Liska <mliska@suse.cz>
24345 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24346 * dwarf2out.c (is_java): Remove the function.
24347 (output_pubname): Remove usage of the function.
24348 (lower_bound_default): Remove usage of DW_LANG_Java.
24349 (gen_compile_unit_die): Likewise.
24350 * gcc.c: Remove compiler defaults for .java and .zip files.
24351 * gimple-expr.c (remove_suffix): Change as there's no longer
24352 extension than 4-letter one.
24353 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24354 (gimplify_save_expr): Likewise.
24355 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24356 as it's possible even for other languages than Java.
24357 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24358 * lto-opts.c (lto_write_options): Remove reference to Java.
24359 * opts.c (strip_off_ending): Update file extension handling.
24360 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24361 * tree-eh.c (lower_resx): Likewise.
24362 * tree.c (free_lang_data_in_type): Remove dead code.
24363 (find_decls_types_r): Likewise.
24364 (build_common_builtin_nodes): Remove Java from a comment.
24365 (verify_type): Remove dead code.
24366 * varasm.c (assemble_external): Remove Java from a comment.
24368 2017-07-14 Martin Liska <mliska@suse.cz>
24370 * opts.c (finish_options): Add quotes.
24371 (common_handle_option): Likewise.
24373 2017-07-14 Martin Liska <mliska@suse.cz>
24375 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24376 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24377 Remove N_SO_PASCAL.
24378 * dwarf2out.c (lower_bound_default): Do not handle
24380 (gen_compile_unit_die): Likewise.
24381 * gcc.c: Remove default extension binding for GNU Pascal.
24382 * stmt.c: Remove Pascal language from a comment.
24383 * xcoffout.c: Likewise.
24385 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24388 * diagnostic-show-locus.c (fixit_cmp): New function.
24389 (layout::layout): Sort m_fixit_hints.
24390 (column_range::column_range): Assert that the values are valid.
24391 (struct char_span): New struct.
24392 (correction::overwrite): New method.
24393 (struct source_line): New struct.
24394 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24395 calls in terms of classes source_line and char_span, and
24396 correction::overwrite.
24397 (selftest::test_overlapped_fixit_printing_2): New function.
24398 (selftest::diagnostic_show_locus_c_tests): Call it.
24400 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24402 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24403 early if there is no lhs.
24405 2017-07-13 Martin Liska <mliska@suse.cz>
24407 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24408 (gen_reference_type_die): Likewise.
24409 * stor-layout.c: Remove Pascal-related comment.
24411 2017-07-13 Martin Liska <mliska@suse.cz>
24413 * opts.c (finish_options): Add quotes to error messages.
24414 (parse_sanitizer_options): Likewise.
24416 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24418 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24420 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24422 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24424 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24426 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24428 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24430 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24433 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24434 provides the hardware capability bits, define the macro
24435 __BUILTIN_CPU_SUPPORTS__.
24436 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24437 if GLIBC does not provide the hardware capability bits. Add a
24438 gcc_unreachable call if the built-in cpu function is neither
24439 __builtin_cpu_is nor __builtin_cpu_supports.
24440 (rs6000_get_function_versions_dispatcher): Change the warning
24441 that an old GLIBC is used which does not export the capability
24442 bits to be an error.
24443 * doc/extend.texi (target_clones attribute): Document the
24444 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24445 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24446 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24447 the macros defined by GCC if the newer GLIBC is available.
24449 2017-07-12 Jeff Law <law@redhat.com>
24451 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24452 remaining includes slightly.
24453 * config/riscv/riscv-builtins.c: Include profile-count.h.
24455 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24458 * config/avr/avr.c (avr_set_current_function): In diagnostic
24459 messages: Quote keywords and (parts of) identifiers.
24460 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24463 2017-07-12 Carl Love <cel@us.ibm.com>
24465 * config/rs6000/rs6000-c.c: Add support for built-in functions
24466 vector bool char vec_revb (vector bool char);
24467 vector bool short vec_revb (vector short char);
24468 vector bool int vec_revb (vector bool int);
24469 vector bool long long vec_revb (vector bool long long);
24470 * doc/extend.texi: Update the built-in documentation file for the
24471 new built-in functions.
24473 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24475 * config/s390/s390.md: Remove movcc splitter.
24477 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24479 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24480 load/store on condition.
24482 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24485 * config/avr/avr.c (avr_encode_section_info)
24486 [progmem && !TREE_READONLY]: Error if progmem object needs
24489 2017-07-11 Michael Collison <michael.collison@arm.com>
24491 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24494 2017-07-11 Carl Love <cel@us.ibm.com>
24496 * config/rs6000/rs6000-c.c: Add support for builtins
24497 vector unsigned int vec_parity_lsbb (vector signed int);
24498 vector unsigned int vec_parity_lsbb (vector unsigned int);
24499 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24500 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24501 vector unsigned long long vec_parity_lsbb (vector signed long long);
24502 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24503 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24504 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24505 * doc/extend.texi: Update the built-in documentation file for the
24506 new built-in functions.
24508 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24510 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24511 (layout::m_primary_loc): New field.
24512 (layout::layout): Initialize new field. Move location filtering
24513 logic from here to...
24514 (layout::maybe_add_location_range): ...this new method. Add
24515 support for filtering to just the lines already specified by other
24517 (layout::will_show_line_p): New method.
24518 (gcc_rich_location::add_location_if_nearby): New method.
24519 (selftest::test_add_location_if_nearby): New test function.
24520 (selftest::diagnostic_show_locus_c_tests): Call it.
24521 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24524 2017-07-11 Tom de Vries <tom@codesourcery.com>
24526 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24527 (bb_first_real_insn): New function.
24528 (nvptx_single): Add extra initialization of broadcasted condition
24531 2017-07-11 Nathan Sidwell <nathan@acm.org>
24533 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24535 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24537 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24538 Binutils doc as TEXI will mess them up.
24539 * doc/invoke.texi (AVR Options): Same here.
24541 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24543 * config/sparc/sparc.opt (mfix-ut700): New option.
24544 (mfix-gr712rc): Likewise.
24545 (sparc_fix_b2bst): New variable.
24546 * doc/invoke.texi (SPARC options): Document them.
24547 (ARM options): Fix warnings.
24548 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24549 instructions to prevent sequences that can trigger the store-store
24550 errata for certain LEON3FT processors.
24551 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24552 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24553 * config/sparc/sparc.md (fix_b2bst): New attribute.
24554 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24556 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24559 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24561 (*rsqrtsf2_sse): Ditto.
24563 (div<mode>3): Macroize insn from divdf3 and divsf3
24564 using MODEF mode iterator.
24566 2017-07-10 Martin Sebor <msebor@redhat.com>
24568 PR tree-optimization/80397
24569 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24570 instead of testing for equality to INTEGER_TYPE.
24572 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24574 * config.gcc: Remove uclibc from arc target spec.
24576 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24578 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24580 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24583 * lto-wrapper.c (remove_option): New function.
24584 (merge_and_complain): Merge PIC/PIE options more realistically.
24586 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24588 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24592 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24593 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24594 * config.in: Regenerate.
24595 * configure: Regenerate.
24596 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24597 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24598 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24599 (TARGET_GASISR_PROLOGUES): ...target mask.
24600 * common/config/avr/avr-common.c
24601 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24602 Set -mgas-isr-prologues.
24603 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24604 INSERT_PASS_BEFORE for it.
24605 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24606 * config/avr/avr.c (avr_option_override)
24607 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24608 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24609 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24610 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24611 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24613 (make_avr_pass_pre_proep): New function.
24614 (emit_push_sfr) <treg>: Add argument to function and use it
24615 instead of TMP_REG.
24616 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24617 and set machine->gasisr.yes.
24618 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24619 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24620 __gcc_isr.n_pushed to .L__stack_usage.
24621 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24622 (avr_asm_final_postscan_insn): ...this new static function.
24623 * config/avr/avr.h (machine_function)
24624 <is_no_gccisr, use_L__stack_usage>: New fields.
24625 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24626 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24627 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24628 (gasisr, *gasisr): New expander and insn.
24629 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24630 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24631 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24633 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24635 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24638 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24640 Move jump-tables out of .text again.
24643 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24644 (ASM_OUTPUT_ADDR_VEC): New function.
24645 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24646 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24647 INSN_ADDRESSes as asm comment.
24648 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24649 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24650 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24651 * config/avr/avr.md (*tablejump): Adjust comment.
24652 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24653 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24655 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24656 (avr_output_addr_vec): New proto.
24657 (avr_log_t) <insn_addresses>: New field.
24659 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24662 * config/i386/i386.c (ix86_function_arg_advance): Set
24663 outgoing_args_on_stack to true if there are outgoing arguments
24665 (ix86_function_arg): Likewise.
24666 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24667 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24668 * config/i386/i386.h (machine_function): Add
24669 outgoing_args_on_stack.
24671 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24673 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24674 supporting pthreds.
24675 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24677 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24679 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24680 (REAL_H): Remove $(MACHMODE_H).
24681 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24683 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24684 $(MACHMODE_H) and double-int.h.
24685 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24687 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24690 2017-07-07 Andrew Pinski <apinski@cavium.com>
24692 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24693 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24695 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24697 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24698 Add warning if GCC was not configured to link against a GLIBC that
24699 exports the hardware capability bits.
24700 (make_resolver_func): Make resolver function private and not a
24701 COMDAT function. Create the name with clone_function_name instead
24702 of make_unique_name.
24705 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24706 correct operand in doing the split.
24708 2017-07-07 Carl Love <cel@us.ibm.com>
24710 * config/rs6000/rs6000-c: Add support for built-in function
24711 vector unsigned short vec_pack_to_short_fp32 (vector float,
24713 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24714 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24715 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24716 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24717 (convert_4f32_8i16): Add define_expand.
24718 * doc/extend.texi: Update the built-in documentation file for the
24719 new built-in function.
24721 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24723 * config/sparc/m8.md: New file.
24724 * config/sparc/sparc.md: Include m8.md.
24726 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24728 * config/sparc/sparc.opt: New option -mvis4b.
24729 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24730 (sparc_option_override): Handle VIS4B.
24731 (enum sparc_builtins): Define
24732 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24733 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24734 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24735 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24736 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24737 (check_constant_argument): New function.
24738 (sparc_vis_init_builtins): Define builtins
24739 __builtin_vis_dictunpack{8,16,32},
24740 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24741 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24742 __builtin_vis_fpcmpde{8,16,32}shl and
24743 __builtin_vis_fpcmpur{8,16,32}shl.
24744 (sparc_expand_builtin): Check that the constant operands to
24745 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24746 constant and in range.
24747 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24749 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24750 (SPARC_IMM5_P): Likewise.
24751 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24752 (enabled): Handle vis4b.
24753 (UNSPEC_DICTUNPACK): New unspec.
24754 (UNSPEC_FPCMPSHL): Likewise.
24755 (UNSPEC_FPUCMPSHL): Likewise.
24756 (UNSPEC_FPCMPDESHL): Likewise.
24757 (UNSPEC_FPCMPURSHL): Likewise.
24758 (cpu_feature): New CPU feature `vis4b'.
24759 (dictunpack{8,16,32}): New insns.
24760 (FPCSMODE): New mode iterator.
24761 (fpcscond): New code iterator.
24762 (fpcsucond): Likewise.
24763 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24764 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24765 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24766 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24767 * config/sparc/constraints.md: Define constraints `q' for unsigned
24768 2-bit integer constants and `t' for unsigned 5-bit integer
24770 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24772 (imm5_operand_dictunpack16): Likewise.
24773 (imm5_operand_dictunpack32): Likewise.
24774 (imm2_operand): Likewise.
24775 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24776 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24777 ditunpack* and fpcmp*shl builtins.
24779 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24781 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24782 * config.in: Add HAVE_AS_SPARC6 define.
24783 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24785 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24787 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24788 (CPP_CPU_SPEC): Handle m8.
24789 (ASM_CPU_SPEC): Likewise.
24790 * config/sparc/sparc-opts.h (enum processor_type): Add
24792 * config/sparc/sparc.c (m8_costs): New struct.
24793 (sparc_option_override): Handle TARGET_CPU_m8.
24794 (sparc32_initialize_trampoline): Likewise.
24795 (sparc64_initialize_trampoline): Likewise.
24796 (sparc_issue_rate): Likewise.
24797 (sparc_register_move_cost): Likewise.
24798 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
24799 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
24800 (ASM_CPU64_DEFAULT_SPEC): Likewise.
24801 (CPP_CPU_SPEC): Handle M8.
24802 (ASM_CPU_SPEC): Likewise.
24803 (AS_M8_FLAG): Define.
24804 * config/sparc/sparc.md: Add m8 to the cpu attribute.
24805 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
24806 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
24808 * configure: Regenerate.
24809 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
24812 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24814 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
24816 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
24817 ("*movdi_insn_sp32"): Do not set v3pipe.
24818 ("*movsi_insn"): Likewise.
24819 ("*movdi_insn_sp64"): Likewise.
24820 ("*movsf_insn"): Likewise.
24821 ("*movdf_insn_sp32"): Likewise.
24822 ("*movdf_insn_sp64"): Likewise.
24823 ("*zero_extendsidi2_insn_sp64"): Likewise.
24824 ("*sign_extendsidi2_insn"): Likewise.
24825 ("*mov<VM32:mode>_insn"): Likewise.
24826 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24827 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24828 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24829 ("<vlop:code><VL:mode>3"): Likewise.
24830 ("*not_<vlop:code><VL:mode>3"): Likewise.
24831 ("*nand<VL:mode>_vis"): Likewise.
24832 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24833 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24834 ("one_cmpl<VL:mode>2"): Likewise.
24835 ("faligndata<VM64:mode>_vis"): Likewise.
24836 ("alignaddrsi_vis"): Likewise.
24837 ("alignaddrdi_vis"): Likweise.
24838 ("alignaddrlsi_vis"): Likewise.
24839 ("alignaddrldi_vis"): Likewise.
24840 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24841 ("bmaskdi_vis"): Likewise.
24842 ("bmasksi_vis"): Likewise.
24843 ("bshuffle<VM64:mode>_vis"): Likewise.
24844 ("cmask8<P:mode>_vis"): Likewise.
24845 ("cmask16<P:mode>_vis"): Likewise.
24846 ("cmask32<P:mode>_vis"): Likewise.
24847 ("pdistn<P:mode>_vis"): Likewise.
24848 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24850 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24852 * config/sparc/sparc.md ("subtype"): New insn attribute.
24853 ("*wrgsr_sp64"): Set insn subtype.
24854 ("*rdgsr_sp64"): Likewise.
24855 ("alignaddrsi_vis"): Likewise.
24856 ("alignaddrdi_vis"): Likewise.
24857 ("alignaddrlsi_vis"): Likewise.
24858 ("alignaddrldi_vis"): Likewise.
24859 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24860 ("fexpand_vis"): Likewise.
24861 ("fpmerge_vis"): Likewise.
24862 ("faligndata<VM64:mode>_vis"): Likewise.
24863 ("bshuffle<VM64:mode>_vis"): Likewise.
24864 ("cmask8<P:mode>_vis"): Likewise.
24865 ("cmask16<P:mode>_vis"): Likewise.
24866 ("cmask32<P:mode>_vis"): Likewise.
24867 ("fchksm16_vis"): Likewise.
24868 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
24869 ("fmean16_vis"): Likewise.
24870 ("fp<plusminus_insn>64_vis"): Likewise.
24871 ("<plusminus_insn>v8qi3"): Likewise.
24872 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24873 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
24874 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
24875 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
24876 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
24877 ("*movqi_insn"): Likewise.
24878 ("*movhi_insn"): Likewise.
24879 ("*movsi_insn"): Likewise.
24880 ("movsi_pic_gotdata_op"): Likewise.
24881 ("*movdi_insn_sp32"): Likewise.
24882 ("*movdi_insn_sp64"): Likewise.
24883 ("movdi_pic_gotdata_op"): Likewise.
24884 ("*movsf_insn"): Likewise.
24885 ("*movdf_insn_sp32"): Likewise.
24886 ("*movdf_insn_sp64"): Likewise.
24887 ("*zero_extendhisi2_insn"): Likewise.
24888 ("*zero_extendqihi2_insn"): Likewise.
24889 ("*zero_extendqisi2_insn"): Likewise.
24890 ("*zero_extendqidi2_insn"): Likewise.
24891 ("*zero_extendhidi2_insn"): Likewise.
24892 ("*zero_extendsidi2_insn_sp64"): Likewise.
24893 ("ldfsr"): Likewise.
24894 ("prefetch_64"): Likewise.
24895 ("prefetch_32"): Likewise.
24896 ("tie_ld32"): Likewise.
24897 ("tie_ld64"): Likewise.
24898 ("*tldo_ldub_sp32"): Likewise.
24899 ("*tldo_ldub1_sp32"): Likewise.
24900 ("*tldo_ldub2_sp32"): Likewise.
24901 ("*tldo_ldub_sp64"): Likewise.
24902 ("*tldo_ldub1_sp64"): Likewise.
24903 ("*tldo_ldub2_sp64"): Likewise.
24904 ("*tldo_ldub3_sp64"): Likewise.
24905 ("*tldo_lduh_sp32"): Likewise.
24906 ("*tldo_lduh1_sp32"): Likewise.
24907 ("*tldo_lduh_sp64"): Likewise.
24908 ("*tldo_lduh1_sp64"): Likewise.
24909 ("*tldo_lduh2_sp64"): Likewise.
24910 ("*tldo_lduw_sp32"): Likewise.
24911 ("*tldo_lduw_sp64"): Likewise.
24912 ("*tldo_lduw1_sp64"): Likewise.
24913 ("*tldo_ldx_sp64"): Likewise.
24914 ("*mov<VM32:mode>_insn"): Likewise.
24915 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24916 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24918 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24920 * config/sparc/sparc.md ("type"): New insn type viscmp.
24921 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24923 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24924 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24925 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24926 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24928 ("n7_vis_logical_11cycle"): Likewise.
24929 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24930 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24931 * config/sparc/niagara.md ("niag_vis"): Likewise.
24932 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24933 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24935 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24937 * config/sparc/sparc.md: New instruction type `bmask'.
24938 (bmaskdi_vis): Use the `bmask' type.
24939 (bmasksi_vis): Likewise.
24940 * config/sparc/ultra3.md (us3_array): Likewise.
24941 * config/sparc/niagara7.md (n7_array): Likewise.
24942 * config/sparc/niagara4.md (n4_array): Likewise.
24943 * config/sparc/niagara2.md (niag2_vis): Likewise.
24944 (niag3_vis): Likewise.
24945 * config/sparc/niagara.md (niag_vis): Likewise.
24947 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24949 * ipa-comdats.c: Remove optimize check from gate.
24950 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24951 for functions not optimized.
24952 (ipa_fn_summary_read): Skip optimize check.
24953 (ipa_fn_summary_write): Likewise.
24954 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24956 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24958 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24959 (check_callers): Give up on caller which is not optimized.
24960 (inline_small_functions): Likewise.
24961 (ipa_inline): Do not give up when not optimizing.
24962 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24963 away unoptimizes cdtors.
24964 (whole_program_function_and_variable_visibility): Do
24965 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24966 * ipa.c (process_references): Do not check optimize.
24967 (symbol_table::remove_unreachable_nodes): Update optimize check.
24968 (set_writeonly_bit): Update optimize check.
24969 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24970 (pass_ipa_single_use::gate): Remove.
24972 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24974 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24975 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24976 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24977 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24978 permute_load, permute_store, adjust_extract, adjust_splat,
24979 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24980 replace_swap_with_copy, dump_swap_insn_table,
24981 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24982 recombine_lvx_pattern, recombine_stvx_pattern,
24983 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24984 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24985 to file rs6000-p8swap.c.
24986 * config/rs6000/rs6000-p8swap.c: New file.
24987 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24988 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24989 and rs6000*-*-* targets.
24991 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24993 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24995 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24997 * lto-wrapper.c (merge_and_complain): Do not merge
24998 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24999 fsigned_zeros, ftrapping_math, fwrapv.
25000 (append_compiler_options): Do not track these options.
25001 (append_linker_options): Likewie
25003 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25005 * cgraphunit.c (cgraph_node::finalize_function): When
25006 !flag_toplevel_reorde set no_reorder flag.
25007 (varpool_node::finalize_decl): Likewise.
25008 (symbol_table::compile): Drop no toplevel reorder path.
25010 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25012 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25013 edges; zero probability is not better than uninitialized.
25015 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25017 * asan.h (asan_sanitize_allocas_p): Declare.
25018 * asan.c (asan_sanitize_allocas_p): New function.
25019 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25020 (handle_builtin_alloca): Likewise.
25021 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25022 if !asan_sanitize_allocas_p.
25023 * params.def (asan-instrument-allocas): Add new option.
25024 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25025 * opts.c (common_handle_option): Disable allocas sanitization for
25028 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25030 * asan.c: Include gimple-fold.h.
25031 (get_last_alloca_addr): New function.
25032 (handle_builtin_stackrestore): Likewise.
25033 (handle_builtin_alloca): Likewise.
25034 (asan_emit_allocas_unpoison): Likewise.
25035 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25036 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25037 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25038 (instrument_builtin_call): Pass gimple iterator to
25039 get_mem_refs_of_builtin_call.
25040 (last_alloca_addr): New global.
25041 * asan.h (asan_emit_allocas_unpoison): Declare.
25042 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25043 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25044 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25045 if function calls alloca.
25046 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25047 * gimple-fold.h (replace_call_with_value): Declare.
25048 * internal-fn.c: Include asan.h.
25049 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25050 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25052 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25054 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25055 (C_SELFTEST_FLAGS): New.
25056 (CPP_SELFTEST_FLAGS): New.
25057 (SELFTEST_DEPS): New, from deps of s-selftest.
25058 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25059 (CPP_SELFTEST_DEPS): New.
25060 (selftest): Add dependency on s-selftest-c++.
25061 (s-selftest): Rename to...
25062 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25063 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25064 than SELFTEST_FLAGS.
25065 (selftest-gdb): Rename to...
25066 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25068 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25069 (selftest-valgrind): Rename to...
25070 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25072 (selftest-valgrind): Reintroduce as an alias for
25073 selftest-c-valgrind.
25074 (s-selftest-c++): New.
25075 (selftest-c++-gdb): New.
25076 (selftest-c++-valgrind): New.
25078 2017-07-06 Olivier Hainque <hainque@adacore.com>
25080 * gcc.c (process_command): When deciding if undefined variables
25081 should be ignored when processing specs, accept "gcc -v" as well.
25083 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25085 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25086 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25088 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25090 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25091 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25092 * config/arm/arm-tables.opt: Regenerate.
25093 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25095 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25097 2017-07-06 Carl Love <cel@us.ibm.com>
25099 * ChangeLog: Clean up from mid air collision
25101 2017-07-06 Carl Love <cel@us.ibm.com>
25103 * config/rs6000/rs6000-c.c: Add support for built-in functions
25104 vector signed int vec_subc (vector signed int, vector signed int);
25105 vector signed __int128 vec_subc (vector signed __int128,
25106 vector signed __int128);
25107 vector unsigned __int128 vec_subc (vector unsigned __int128,
25108 vector unsigned __int128);
25109 vector signed int vec_sube (vector signed int, vector signed int,
25110 vector signed int);
25111 vector unsigned int vec_sube (vector unsigned int,
25112 vector unsigned int,
25113 vector unsigned int);
25114 vector signed __int128 vec_sube (vector signed __int128,
25115 vector signed __int128,
25116 vector signed__int128);
25117 vector unsigned __int128 vec_sube (vector unsigned __int128,
25118 vector unsigned __int128,
25119 vector unsigned __int128);
25120 vector signed int vec_subec (vector signed int, vector signed int,
25121 vector signed int);
25122 vector unsigned int vec_subec (vector unsigned int,
25123 vector unsigned int,
25124 vector unsigned int);
25125 vector signed __int128 vec_subec (vector signed __int128,
25126 vector signed __int128,
25127 vector signed__int128);
25128 vector unsigned __int128 vec_subec (vector unsigned __int128,
25129 vector unsigned __int128,
25130 vector unsigned __int128);
25131 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25132 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25133 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25134 BU_ALTIVEC_OVERLOAD_X definitions.
25135 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25136 * doc/extend.texi: Update the built-in documentation file for the new
25137 built-in functions.
25139 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25142 * diagnostic-show-locus.c (layout::layout): Use start and finish
25143 spelling location for the start and finish of each range.
25144 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25145 Add unused aspect param.
25146 * input.c (expand_location_1): Add "aspect" param, and use it
25147 to access the correct part of the location.
25148 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25150 (expand_location_to_spelling_point): Likewise.
25151 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25152 param, and pass it to expand_location_1.
25154 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25156 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25157 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25158 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25159 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25160 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25161 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25162 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25163 _mm_maskz_getmant_ss): New intrinsics.
25164 (__builtin_ia32_getexpss128_mask): Changed to ...
25165 __builtin_ia32_getexpss128_round ... this.
25166 (__builtin_ia32_getexpsd128_mask): Changed to ...
25167 __builtin_ia32_getexpsd128_round ... this.
25168 * config/i386/i386-builtin-types.def
25169 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25170 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25171 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25172 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25173 __builtin_ia32_getmantss_mask_round): New builtins.
25174 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25175 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25176 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25177 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25178 * config/i386/sse.md
25179 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25180 avx512f_sgetexp<mode><mask_scalar_name>
25181 <round_saeonly_scalar_name> ... this.
25182 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25183 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25184 vgetexp<ssescalarmodesuffix>
25185 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25186 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25187 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25188 avx512f_vgetmant<mode><mask_scalar_name>
25189 <round_saeonly_scalar_name> ... this.
25190 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25191 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25192 vgetmant<ssescalarmodesuffix>
25193 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25194 %0<mask_scalar_operand4>, %1, %2
25195 <round_saeonly_scalar_mask_op4>, %3} ... this.
25196 * config/i386/subst.md (mask_scalar_operand4,
25197 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25198 round_saeonly_scalar_nimm_predicate): New subst attributes.
25200 2017-07-06 Julia Koval <julia.koval@intel.com>
25202 * config/i386/i386.c (ix86_erase_embedded_rounding):
25203 Remove code for old rounding pattern.
25205 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25207 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25209 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25211 * doc/sourcebuild.texi (Test Directives, Variants of
25212 dg-require-support): Add documentation for dg-require-stack-check.
25214 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25216 * config/i386/subst.md (mask_scalar, round_scalar,
25217 round_saeonly_scalar): New meta-templates.
25218 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25219 round_scalar_mask_operand3, round_scalar_mask_op3,
25220 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25221 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25222 round_saeonly_scalar_constraint,
25223 round_saeonly_scalar_prefix): New subst attribute.
25224 * config/i386/sse.md
25225 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25226 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25227 <round_scalar_name> ... this.
25228 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25229 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25230 <round_scalar_name> ... this.
25231 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25232 <sse>_vm<code><mode>3<mask_scalar_name>
25233 <round_saeonly_scalar_name> ... this.
25234 (v<plusminus_mnemonic><ssescalarmodesuffix>
25235 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25236 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25237 v<plusminus_mnemonic><ssescalarmodesuffix>
25238 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25239 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25240 (v<multdiv_mnemonic><ssescalarmodesuffix>
25241 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25242 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25243 v<multdiv_mnemonic><ssescalarmodesuffix>
25244 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25245 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25246 (v<maxmin_float><ssescalarmodesuffix>
25247 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25248 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25249 v<maxmin_float><ssescalarmodesuffix>
25250 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25251 %0<mask_scalar_operand3>, %1, %<iptr>2
25252 <round_saeonly_scalar_mask_op3>} ... this.
25254 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25256 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25257 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25259 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25260 Alan Hayward <alan.hayward@arm.com>
25261 David Sherwood <david.sherwood@arm.com>
25263 * combine.c (simplify_if_then_else): Remove "enum" before
25265 * compare-elim.c (can_eliminate_compare): Likewise.
25266 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25268 (aarch64_lookup_simd_builtin_type): Likewise.
25269 (aarch64_simd_builtin_type): Likewise.
25270 (aarch64_init_simd_builtin_types): Likewise.
25271 (aarch64_simd_expand_args): Likewise.
25272 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25274 (aarch64_reverse_mask): Likewise.
25275 (aarch64_simd_emit_reg_reg_move): Likewise.
25276 (aarch64_gen_adjusted_ldpstp): Likewise.
25277 (aarch64_ccmp_mode_to_code): Likewise.
25278 (aarch64_operands_ok_for_ldpstp): Likewise.
25279 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25280 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25282 (aarch64_min_divisions_for_recip_mul): Likewise.
25283 (aarch64_reassociation_width): Likewise.
25284 (aarch64_get_condition_code_1): Likewise.
25285 (aarch64_simd_emit_reg_reg_move): Likewise.
25286 (aarch64_simd_attr_length_rglist): Likewise.
25287 (aarch64_reverse_mask): Likewise.
25288 (aarch64_operands_ok_for_ldpstp): Likewise.
25289 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25290 (aarch64_gen_adjusted_ldpstp): Likewise.
25291 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25293 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25294 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25295 (arm_lookup_simd_builtin_type): Likewise.
25296 (arm_simd_builtin_type): Likewise.
25297 (arm_init_simd_builtin_types): Likewise.
25298 (arm_expand_builtin_args): Likewise.
25299 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25300 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25301 (ft32_setup_incoming_varargs): Likewise.
25302 (ft32_function_arg): Likewise.
25303 (ft32_function_arg_advance): Likewise.
25304 (ft32_pass_by_reference): Likewise.
25305 (ft32_arg_partial_bytes): Likewise.
25306 (ft32_valid_pointer_mode): Likewise.
25307 (ft32_addr_space_pointer_mode): Likewise.
25308 (ft32_addr_space_legitimate_address_p): Likewise.
25309 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25311 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25312 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25313 (iamcu_alignment): Likewise.
25314 (canonicalize_vector_int_perm): Likewise.
25315 (ix86_noce_conversion_profitable_p): Likewise.
25316 (ix86_mpx_bound_mode): Likewise.
25317 (ix86_operands_ok_for_move_multiple): Likewise.
25318 * config/microblaze/microblaze-protos.h
25319 (microblaze_expand_conditional_branch_reg): Likewise.
25320 * config/microblaze/microblaze.c
25321 (microblaze_expand_conditional_branch_reg): Likewise.
25322 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25324 (rs6000_reassociation_width): Likewise.
25325 (rs6000_invalid_binary_op): Likewise.
25326 (fusion_p9_p): Likewise.
25327 (emit_fusion_p9_load): Likewise.
25328 (emit_fusion_p9_store): Likewise.
25329 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25331 (riscv_hard_regno_mode_ok_p): Likewise.
25332 (riscv_address_insns): Likewise.
25333 (riscv_split_symbol): Likewise.
25334 (riscv_legitimize_move): Likewise.
25335 (riscv_function_value): Likewise.
25336 (riscv_hard_regno_nregs): Likewise.
25337 (riscv_expand_builtin): Likewise.
25338 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25339 (riscv_build_integer): Likewise.
25340 (riscv_split_integer): Likewise.
25341 (riscv_legitimate_constant_p): Likewise.
25342 (riscv_cannot_force_const_mem): Likewise.
25343 (riscv_regno_mode_ok_for_base_p): Likewise.
25344 (riscv_valid_base_register_p): Likewise.
25345 (riscv_valid_offset_p): Likewise.
25346 (riscv_valid_lo_sum_p): Likewise.
25347 (riscv_classify_address): Likewise.
25348 (riscv_legitimate_address_p): Likewise.
25349 (riscv_address_insns): Likewise.
25350 (riscv_load_store_insns): Likewise.
25351 (riscv_force_binary): Likewise.
25352 (riscv_split_symbol): Likewise.
25353 (riscv_force_address): Likewise.
25354 (riscv_legitimize_address): Likewise.
25355 (riscv_move_integer): Likewise.
25356 (riscv_legitimize_const_move): Likewise.
25357 (riscv_legitimize_move): Likewise.
25358 (riscv_address_cost): Likewise.
25359 (riscv_subword): Likewise.
25360 (riscv_output_move): Likewise.
25361 (riscv_canonicalize_int_order_test): Likewise.
25362 (riscv_emit_int_order_test): Likewise.
25363 (riscv_function_arg_boundary): Likewise.
25364 (riscv_pass_mode_in_fpr_p): Likewise.
25365 (riscv_pass_fpr_single): Likewise.
25366 (riscv_pass_fpr_pair): Likewise.
25367 (riscv_get_arg_info): Likewise.
25368 (riscv_function_arg): Likewise.
25369 (riscv_function_arg_advance): Likewise.
25370 (riscv_arg_partial_bytes): Likewise.
25371 (riscv_function_value): Likewise.
25372 (riscv_pass_by_reference): Likewise.
25373 (riscv_setup_incoming_varargs): Likewise.
25374 (riscv_print_operand): Likewise.
25375 (riscv_elf_select_rtx_section): Likewise.
25376 (riscv_save_restore_reg): Likewise.
25377 (riscv_for_each_saved_reg): Likewise.
25378 (riscv_register_move_cost): Likewise.
25379 (riscv_hard_regno_mode_ok_p): Likewise.
25380 (riscv_hard_regno_nregs): Likewise.
25381 (riscv_class_max_nregs): Likewise.
25382 (riscv_memory_move_cost): Likewise.
25383 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25384 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25385 (rl78_addr_space_address_mode): Likewise.
25386 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25388 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25389 (rs6000_reassociation_width): Likewise.
25390 (rs6000_invalid_binary_op): Likewise.
25391 (fusion_p9_p): Likewise.
25392 (emit_fusion_p9_load): Likewise.
25393 (emit_fusion_p9_store): Likewise.
25394 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25395 (ok_for_simple_move_operands): Likewise.
25396 (ok_for_simple_move_strict_operands): Likewise.
25397 (ok_for_simple_arith_logic_operands): Likewise.
25398 (visium_legitimize_reload_address): Likewise.
25399 (visium_select_cc_mode): Likewise.
25400 (output_cbranch): Likewise.
25401 (visium_split_double_move): Likewise.
25402 (visium_expand_copysign): Likewise.
25403 (visium_expand_int_cstore): Likewise.
25404 (visium_expand_fp_cstore): Likewise.
25405 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25406 (visium_function_arg): Likewise.
25407 (visium_function_arg_advance): Likewise.
25408 (visium_libcall_value): Likewise.
25409 (visium_setup_incoming_varargs): Likewise.
25410 (visium_legitimate_constant_p): Likewise.
25411 (visium_legitimate_address_p): Likewise.
25412 (visium_legitimize_address): Likewise.
25413 (visium_secondary_reload): Likewise.
25414 (visium_register_move_cost): Likewise.
25415 (visium_memory_move_cost): Likewise.
25416 (prepare_move_operands): Likewise.
25417 (ok_for_simple_move_operands): Likewise.
25418 (ok_for_simple_move_strict_operands): Likewise.
25419 (ok_for_simple_arith_logic_operands): Likewise.
25420 (visium_function_value_1): Likewise.
25421 (rtx_ok_for_offset_p): Likewise.
25422 (visium_legitimize_reload_address): Likewise.
25423 (visium_split_double_move): Likewise.
25424 (visium_expand_copysign): Likewise.
25425 (visium_expand_int_cstore): Likewise.
25426 (visium_expand_fp_cstore): Likewise.
25427 (visium_split_cstore): Likewise.
25428 (visium_select_cc_mode): Likewise.
25429 (visium_split_cbranch): Likewise.
25430 (output_cbranch): Likewise.
25431 (visium_print_operand_address): Likewise.
25432 * expmed.c (flip_storage_order): Likewise.
25433 * expmed.h (emit_cstore): Likewise.
25434 (flip_storage_order): Likewise.
25435 * genrecog.c (validate_pattern): Likewise.
25436 * hsa-gen.c (gen_hsa_addr): Likewise.
25437 * internal-fn.c (expand_arith_overflow): Likewise.
25438 * ira-color.c (allocno_copy_cost_saving): Likewise.
25439 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25440 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25441 (process_invariant_for_inheritance): Likewise.
25442 * lra-eliminations.c (move_plus_up): Likewise.
25443 * omp-low.c (lower_oacc_reductions): Likewise.
25444 * simplify-rtx.c (simplify_subreg): Likewise.
25445 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25446 (TARGET_CHKP_BOUND_MODE): Likewise..
25447 * targhooks.c (default_chkp_bound_mode): Likewise.
25448 (default_setup_incoming_vararg_bounds): Likewise.
25449 * targhooks.h (default_chkp_bound_mode): Likewise.
25450 (default_setup_incoming_vararg_bounds): Likewise.
25451 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25452 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25453 (have_whole_vector_shift): Likewise.
25454 * tree-vect-stmts.c (vectorizable_load): Likewise.
25455 * doc/tm.texi: Regenerate.
25457 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25459 Graceful degrade if Binutils PR21472 is not available.
25462 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25463 .rodata in flash test fails.
25464 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25465 * confgure: Regenerate.
25466 * config.in: Regenerate.
25467 * config/avr/avr.c (avr_asm_named_section)
25468 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25469 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25470 (avr_asm_init_sections): Same.
25472 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25474 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25475 (fma<VH:mode>4_intrinsic): Likewise.
25476 (*fmsub<VCVTF:mode>4): Likewise.
25477 (*fmsub<VH:mode>4_intrinsic): Likewise.
25479 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25482 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25483 Don't depend on "optimize > 0".
25484 (out_movhi_r_mr, out_movqi_mr_r): Same.
25485 (out_movhi_mr_r, out_movqi_r_mr): Same.
25486 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25487 io_address_operand on "optimize > 0".
25489 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25491 * tree-loop-distribution.c: Add general explanantion on the pass.
25492 (generate_loops_for_partition): Mark distributed loop.
25493 (pg_add_dependence_edges): New parameter. Handle alias data
25494 dependence specially and record it in the parameter if asked.
25495 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25496 (init_partition_graph_vertices, add_partition_graph_edge): New.
25497 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25498 (free_partition_graph_vdata, build_partition_graph): New.
25499 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25500 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25501 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25502 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25503 (version_for_distribution_p, finalize_partitions): New.
25504 (distribute_loop): Handle alias data dependence specially. Factor
25505 out loop fusion code as functions and call these functions.
25507 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25509 * tree-loop-distribution.c (classify_partition): New parameter and
25510 better handle reduction statement.
25511 (rdg_build_partitions): Revise comment.
25512 (distribute_loop): Compute statements in all partitions and pass it
25513 to classify_partition.
25515 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25517 * tree-loop-distribution.c (enum partition_type): New.
25518 (struct partition): New field type.
25519 (partition_merge_into): Add parameter. Update partition type.
25520 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25521 (build_rdg_partition_for_vertex): Compute partition type.
25522 (rdg_build_partitions): Dump partition type.
25523 (distribute_loop): Update calls to partition_merge_into.
25525 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25527 * tree-loop-distribution.c (struct ddr_hasher): New.
25528 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25530 (classify_partition): Call get_data_dependence.
25531 (pg_add_dependence_edges): Ditto.
25532 (distribute_loop): Release data dependence hash table.
25534 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25536 * tree-loop-distribution.c (ref_base_address): Delete.
25537 (similar_memory_accesses): Rename ...
25538 (share_memory_accesses): ... to this. Check if partitions access
25539 the same memory reference.
25540 (distribute_loop): Call share_memory_accesses.
25542 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25544 * tree-loop-distribution.c (struct partition): New field recording
25545 its data reference.
25546 (partition_alloc, partition_free): Init and release data refs.
25547 (partition_merge_into): Merge data refs.
25548 (build_rdg_partition_for_vertex): Collect data refs for partition.
25549 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25551 (distribute_loop): Remve data refs from vertice data of partition
25554 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25556 * tree-loop-distribution.c (params.h): Include header file.
25557 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25558 (datarefs_vec): New global var.
25559 (create_rdg_vertices): Use datarefs_vec directly.
25560 (free_rdg): Don't free data references.
25561 (build_rdg): Update use. Don't free data references.
25562 (distribute_loop): Compute global variable for data references.
25563 Bail out if there are too many data references.
25565 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25567 * tree-loop-distribution.c (loop_nest): New global var.
25568 (build_rdg): Use loop directly, rather than loop nest.
25569 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25571 (distribute_loop): Compute global variable loop nest. Update use.
25573 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25575 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25576 (partition_merge_into): New parameter. Dump reason for fusion.
25577 (distribute_loop): Update use of partition_merge_into.
25579 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25581 * tree-loop-distribution.c (bb_top_order_index): New.
25582 (bb_top_order_index_size, bb_top_order_cmp): New.
25583 (stmts_from_loop): Use topological order.
25584 (pass_loop_distribution::execute): Compute and release topological
25585 order for basic blocks.
25587 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25589 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25592 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25594 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25595 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25596 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25597 * internal-fn.def (LOOP_DIST_ALIAS): New.
25598 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25599 (fold_loop_internal_call): ... this.
25600 (vect_loop_dist_alias_call): New function.
25601 (set_uid_loop_bbs): Call fold_loop_internal_call.
25602 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25605 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25608 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25609 Require dead FLAGS_REG at the beginning of a peephole.
25611 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25614 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25615 arguments in the call to __builtin_ia32_sbb_u32.
25616 (_subborrow_u64): Swap _X and _Y arguments in the call to
25617 __builtin_ia32_sbb_u64.
25619 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25622 * tree-vrp.c (compare_assert_loc): Turn into a function template
25623 with stable template parameter. Only test if a->e is NULL,
25624 !a->e == !b->e has been verified already. Use e == NULL or
25625 e != NULL instead of e or ! e tests. If stable is true, don't use
25626 iterative_hash_expr, on the other side allow a or b or both NULL
25627 and sort the NULLs last.
25628 (process_assert_insertions): Sort using compare_assert_loc<false>
25629 instead of compare_assert_loc, later sort using
25630 compare_assert_loc<true> before calling process_assert_insertions_for
25631 in a loop. Use break instead of continue once seen NULL pointer.
25633 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25635 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25636 Cortex-R7 and Cortex-R8 processors.
25638 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25640 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25641 uninitialized while src is not.
25643 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25645 * common/config/arm/arm-common.c: Adjust include path for
25647 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25648 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25649 (arm-cpu-data.h): Likewise.
25650 (arm-cpu-cdata.h): Likewise.
25651 * config/arm/arm-cpu.h: Delete.
25652 * config/arm/arm-cpu-cdata.h: Delete.
25653 * config/arm/arm-cpu-data.h: Delete.
25655 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25657 * config/arm/arm-cpus.in (cortex-a55): New.
25658 (cortex-a75): Likewise.
25659 (cortex-a75.cortex-a55): Likewise.
25660 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25662 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25663 * config/arm/arm-cpu-cdata.h: Regenerate.
25664 * config/arm/arm-cpu-data.h: Regenerate.
25665 * config/arm/arm-cpu.h: Regenerate.
25666 * config/arm/arm-tables.opt: Regenerate.
25667 * config/arm/arm-tune.md: Regenerate.
25669 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25671 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25673 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25675 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25678 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25680 PR tree-optimization/81292
25681 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25682 full_string_p, also call adjust_related_strinfos if the adjustment
25683 is simple, otherwise invalidate related strinfos.
25685 2017-07-04 Martin Liska <mliska@suse.cz>
25688 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25689 newly created variable as DECL_IGNORED_P.
25691 2017-07-04 Martin Liska <mliska@suse.cz>
25694 * ipa-inline.c (inline_small_functions):
25695 Use xstrdup_for_dump.
25697 2017-07-04 Tom de Vries <tom@codesourcery.com>
25699 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25701 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25704 * config/darwin.c (darwin_function_switched_text_sections):
25707 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25709 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25711 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25713 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25715 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25717 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25718 min_profitable_iters, and th as inclusive lower bounds.
25719 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25720 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25721 for min_profitable_iters and min_profitable_estimate.
25722 (vect_transform_loop): Treat th as an inclusive lower bound.
25723 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25725 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25728 * config/darwin.c (darwin_function_switched_text_sections):
25729 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25730 in two pieces, and suppress the use of buf.
25732 2017-07-03 Nathan Sidwell <nathan@acm.org>
25734 * hash-table.h (hash_table_mod1): Fix indentation.
25736 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25738 PR middle-end/81290
25739 * predict.c (force_edge_cold): Be more careful about propagation
25741 * profile-count.h (profile_probability::guessed,
25742 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25744 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25746 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25748 * doc/invoke.texi (rcpc architecture extension): Document it.
25750 2017-07-03 Richard Biener <rguenther@suse.de>
25752 PR tree-optimization/60510
25753 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25754 the scalar reduction PHI and use it.
25755 (vectorizable_reduction): Properly guard the single_defuse_cycle
25756 path for non-SLP reduction chains where we cannot use it.
25757 Rework reduc_def/index and vector type deduction. Rework
25758 vector operand gathering during reduction op code-gen.
25759 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25760 chains dissolve the chain and leave it to non-SLP reduction
25763 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25765 * tree-data-ref.h (dr_alignment): Declare.
25766 * tree-data-ref.c (dr_alignment): New function.
25767 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25768 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25770 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25772 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25774 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25775 and base_misalignment fields.
25776 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25777 * tree-data-ref.c: Include builtins.h.
25778 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25779 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25780 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25781 * tree-vect-data-refs.c: Include tree-cfg.h.
25782 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25783 fields instead of calculating an alignment here.
25784 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25785 innermost_loop_behavior fields.
25787 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25789 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25791 (DR_STEP_ALIGNMENT): New macro.
25792 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25793 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25794 (create_data_ref): Print it.
25795 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25796 to tell whether the step preserves vector (mis)alignment.
25797 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25798 Move the check for an integer step and generalise to all INTEGER_CST.
25799 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
25800 Print the outer step alignment.
25802 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25804 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
25805 with offset_alignment.
25806 (DR_ALIGNED_TO): Delete.
25807 (DR_OFFSET_ALIGNMENT): New macro.
25808 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
25809 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
25810 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
25811 (create_data_ref): Likewise.
25812 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25813 (vect_analyze_data_refs): Likewise.
25814 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
25815 creating dummy innermost behavior.
25817 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25819 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
25820 with a "innermost_loop_behavior *" and refeence tree.
25821 * tree-data-ref.c (dr_analyze_innermost): Likewise.
25822 (create_data_ref): Update call accordingly.
25823 * tree-predcom.c (find_looparound_phi): Likewise.
25825 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25827 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25828 fields with dr_wrt_vec_loop.
25829 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25830 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25831 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25832 (vect_dr_behavior): New function.
25833 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25834 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25835 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25836 track whether the step preserves the misalignment.
25837 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25838 Use vect_dr_behavior.
25839 (vect_setup_realignment): Update call accordingly.
25840 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25841 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25842 call to vect_create_addr_base_for_vector_ref.
25843 (vect_create_cond_for_align_checks): Likewise.
25844 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25845 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25846 (vect_recog_mask_conversion_pattern): Likewise.
25847 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25848 (new_stmt_vec_info): Remove redundant zeroing.
25850 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25852 * common/config/arm/arm-common.c (arm_be8_option): New function.
25853 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
25854 (ISA_ARMv6): Add isa_bit_be8.
25855 * config/arm/arm.h (arm_be8_option): Add prototype.
25856 (BE8_SPEC_FUNCTION): New define.
25857 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
25858 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
25859 (mlittle-endian): Similarly.
25860 (mbe8, mbe32): New options.
25861 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25862 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25864 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25866 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
25868 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25870 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
25871 (cleanup_tree_cfg_bb): Use it.
25872 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
25874 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
25876 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25879 * loop-doloop.c (add_test): Update profile.
25881 2017-07-03 Martin Liska <mliska@suse.cz>
25884 * sanopt.c (rewrite_usage_of_param): New function.
25885 (sanitize_rewrite_addressable_params): Likewise.
25886 (pass_sanopt::execute): Call rewrite_usage_of_param.
25888 2017-07-03 Richard Biener <rguenther@suse.de>
25890 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25891 back to using VIEW_CONVERT_EXPR.
25893 2017-07-03 Martin Liska <mliska@suse.cz>
25896 * doc/extend.texi: Document when a resolver function is
25897 generated for target_clones.
25899 2017-07-03 Martin Liska <mliska@suse.cz>
25901 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25902 and shadow memory of auto variables which are subject of
25903 use-after-scope sanitization.
25904 (asan_expand_mark_ifn): Add do set only when is_poison.
25906 2016-07-03 Richard Biener <rguenther@suse.de>
25908 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25910 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25911 (vectorizable_reduction): Perform reduction PHI creation when
25912 visiting a reduction PHI and adjust and simplify code generation
25913 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25914 (vect_transform_loop): Visit reduction PHIs.
25915 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25916 defs into the SLP tree.
25917 (vect_build_slp_tree): Reduction defs terminate the recursion.
25918 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25920 (vect_get_vec_defs_for_stmt_copy): Export.
25921 (vect_get_vec_defs): Likewise.
25922 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25924 (vect_get_vec_defs_for_stmt_copy): Declare.
25925 (vect_get_vec_defs): Likewise.
25927 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25929 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25930 parameter with a "loop" parameter and use it instead of the
25931 loop containing DR_STMT. Don't check simple_iv when doing
25932 BB analysis. Describe the two analysis modes in the comment.
25934 2017-07-03 Tom de Vries <tom@codesourcery.com>
25936 PR tree-optimization/69468
25937 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25938 (find_same_succ_bb): Handle ignore_edge_flags.
25940 2017-07-03 Tom de Vries <tom@codesourcery.com>
25942 PR tree-optimization/81192
25943 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25945 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25947 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25949 2017-07-03 Tom de Vries <tom@codesourcery.com>
25951 PR tree-optimization/81192
25952 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25953 BB_SAME_SUCC (bb) == NULL.
25955 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25957 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25960 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25962 * dumpfile.c: Include profile-count.h
25963 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25965 (insert_cond_bb): Update profile.
25966 * tree-cfg.h (insert_cond_bb): Update prototype.
25967 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25968 * tree-dump.c: Do not include tree-cfg.
25970 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25972 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25974 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25976 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25979 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25981 * tree-complex.c (expand_complex_div_wide): update profile.
25983 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25984 Alan Hayward <alan.hayward@arm.com>
25985 David Sherwood <david.sherwood@arm.com>
25987 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25988 (CORETYPES_H): New define.
25989 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25990 (insn-modes-inline.h, s-modes-inline-h): New rules.
25991 (generated_files): Add insn-modes-inline.h.
25992 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25993 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25994 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25995 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25996 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25997 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25998 (build/gencodes.o, build/genconditions.o): Likewise.
25999 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
26000 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
26001 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
26002 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
26003 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
26004 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
26005 * coretypes.h: Include everything up to real.h for generators.
26006 Include insn-modes.h first. Include wide-int-print.h after
26007 wide-int.h. Include insn-modes-inline.h and then machmode.h.
26008 * machmode.h: Don't include insn-modes.h here.
26009 * function-tests.c: Remove includes of signop.h, machmode.h,
26010 double-int.h and wide-int.h.
26012 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26014 * optc-save-gen.awk: Likewise.
26015 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26016 * godump.c: Remove include of wide-int-print.h.
26017 * pretty-print.h: Likewise.
26018 * wide-int-print.cc: Likewise.
26019 * wide-int.cc: Likewise.
26020 * hash-map-tests.c: Remove include of signop.h.
26021 * hash-set-tests.c: Likewise.
26022 * rtl-tests.c: Likewise.
26023 * mkconfig.sh: Remove include of machmode.h.
26024 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26026 (emit_insn_modes_inline_h): ...this new function. Emit the code
26027 into an insn-modes-inline.h header file, adding appropriate
26028 include guards and end comments.
26029 (emit_insn_modes_c_header): Remove include of machmode.h.
26030 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26032 (main): Handle -i flag and call emit_insn_modes_inline_h when
26035 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26037 * tree-ssa-strlen.c (strinfo): Rename the length field to
26038 nonzero_chars. Add a full_string_p field.
26039 (compare_nonzero_chars, zero_length_string_p): New functions.
26040 (get_addr_stridx): Add an offset_out parameter.
26041 Use compare_nonzero_chars.
26042 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26043 (new_strinfo): Update after above changes to strinfo.
26044 (set_endptr_and_length): Set full_string_p.
26045 (get_string_length): Update after above changes to strinfo.
26046 (unshare_strinfo): Update call to new_strinfo.
26047 (maybe_invalidate): Likewise.
26048 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26049 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26050 as a uhwi instead of an shwi. Update after above changes to
26051 strinfo and new_strinfo.
26052 (zero_length_string): Assert that chainsi contains full strings.
26053 Use zero_length_string_p. Update call to new_strinfo.
26054 (adjust_related_strinfos): Update after above changes to strinfo.
26055 Copy full_string_p from origsi.
26056 (adjust_last_stmt): Use zero_length_string_p.
26057 (handle_builtin_strlen): Update after above changes to strinfo and
26058 new_strinfo. Install the lhs as the string length if the previous
26059 entry didn't describe a full string.
26060 (handle_builtin_strchr): Update after above changes to strinfo
26062 (handle_builtin_strcpy): Likewise.
26063 (handle_builtin_strcat): Likewise.
26064 (handle_builtin_malloc): Likewise.
26065 (handle_pointer_plus): Likewise.
26066 (handle_builtin_memcpy): Likewise. Track nonzero characters
26067 that aren't necessarily followed by a nul terminator.
26068 (handle_char_store): Likewise.
26070 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26072 PR tree-optimization/80769
26073 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26074 for malloc and calloc. Document the new invariant that all related
26075 strinfos have delayed lengths or none do.
26076 (verify_related_strinfos): Move earlier in file.
26077 (set_endptr_and_length): New function, split out from...
26078 (get_string_length): ...here. Also set the lengths of related
26080 (zero_length_string): Assert that chainsi has known (rather than
26082 (adjust_related_strinfos): Likewise.
26084 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26086 PR tree-optimization/81136
26087 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26088 assert that two references with the same misalignment have the same
26089 compile-time misalignment if those compile-time misalignments
26092 2017-07-01 Andi Kleen <ak@linux.intel.com>
26094 * print-tree.c (print_node): Print all attributes.
26096 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26098 * cfg.c (scale_bbs_frequencies): New function.
26099 * cfg.h (scale_bbs_frequencies): Declare it.
26100 * cfgloopanal.c (single_likely_exit): Cleanup.
26101 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26103 (scale_loop_profile): Likewise.
26104 (loop_version): Likewise.
26105 (create_empty_loop_on_edge): Update.
26106 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26107 scale_loop_frequencies, scale_loop_profile, loopify,
26108 loop_version): Update prototypes.
26109 * modulo-sched.c (sms_schedule): Update.
26110 * predict.c (unlikely_executed_edge_p): Also check probability.
26111 (probably_never_executed_edge_p): Fix typo.
26112 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26113 * tree-parloops.c (gen_parallel_loop): Update.
26114 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26115 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26116 * tree-ssa-loop-split.c (split_loop): Update.
26117 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26118 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26119 (vect_loop_versioning): Update.
26120 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26122 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26124 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26126 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26128 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26129 to keep profile consistent.
26131 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26133 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26134 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26135 * profile-count.h (max_safe_multiplier): Make unsigned.
26136 (profile_count::guessed_zero): New.
26138 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26140 * bb-reorder.c (fix_up_crossing_landing_pad,
26141 fix_crossing_conditional_branches): Use make_single_succ_edge
26142 to keep profile consistent.
26144 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26146 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26149 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26152 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26153 the right scopes, make sure cond_jump isn't preserved between multiple
26154 iterations. Search for fallthru edge whenever there are 3+ edges and
26155 use find_fallthru_edge for it.
26157 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26159 Patch by Alexander Monakov <amonakov@ispras.ru>
26160 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26161 probabilities consistently.
26163 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26165 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26166 emit_cmp_and_jump_insns.
26168 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26171 * tree-inline.c (expand_call_inline): Combine profile statuses.
26173 2017-06-30 Andrew Pinski <apinski@cavium.com>
26175 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26176 fold_stmt returned true.
26178 2017-06-30 Nathan Sidwell <nathan@acm.org>
26180 * ggc.h (empty_string): Delete.
26181 * cfgexpand.c (expand_asm_stmt): Use plain "".
26182 * optabs.c (expand_asm_memory_barrier): Likewise.
26183 * stringpool.c (empty_string): Delete.
26184 (digit_vector, digit_string): Delete.
26185 (ggc_alloc_string): Use plain "", don't optimize single digit
26186 strings. Use ggc_alloc_atomic.
26188 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26190 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26191 comparison set and one other set, use the cost of the non-comparison
26194 2017-06-30 Nathan Sidwell <nathan@acm.org>
26196 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26199 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26201 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26202 loops. Remove now unneeded calls to gimple_switch_set_label() that
26203 just set removed labels to NULL_TREE.
26205 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26207 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26208 (set_range_info): ...here. Only call set_range_info_raw if domain
26210 (set_nonzero_bits): Call set_range_info_raw.
26211 * tree-ssanames.h (set_range_info_raw): New.
26213 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26216 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26217 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26218 of nonimmediate_operand and <store_mask_constraint> instead of m
26219 for the input operand. For V8FI iterator, always split if input
26220 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26221 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26222 <store_mask_predicate> instead of register_operand and
26223 <store_mask_constraint> instead of v for the input operand. Make
26224 sure both operands aren't MEMs for if not <mask_applied>.
26226 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26228 * lto-wrapper.c (copy_file) Close both file descriptors before
26231 2017-06-30 Martin Liska <mliska@suse.cz>
26234 * multiple_target.c (create_dispatcher_calls): Make ifunc
26235 also for function that don't have calls or are not referenced.
26237 2017-06-30 Richard Biener <rguenther@suse.de>
26239 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26240 analyze the first scalar stmt. Move vector type computation
26241 for the BB case here from ...
26242 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26243 live operation processing in the SLP case properly.
26245 2017-06-30 Richard Biener <rguenther@suse.de>
26247 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26249 2017-06-30 Martin Liska <mliska@suse.cz>
26252 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26253 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26255 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26257 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26260 2017-06-30 Richard Biener <rguenther@suse.de>
26262 PR tree-optimization/81249
26263 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26264 condition reduction result to original scalar type.
26266 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26268 * profile-count.h (enum profile_quality): Fix typos and whitespace
26271 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26273 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26274 type for branch probabilities.
26276 2017-06-29 Julian Brown <julian@codesourcery.com>
26277 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26279 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26280 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26281 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26282 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26284 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26286 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26287 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26288 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26289 CC usage from generic code to here.
26290 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26291 CC usage into the target macros.
26293 2017-06-29 Maya Rashish <coypu@sdf.org>
26295 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26298 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26300 * arm/arm-builtins.c: Include profile-count.h
26301 * except.c (sjlj_emit_function_enter): Use
26302 profile_probability::unlikely.
26304 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26306 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26307 and tocrel_offset be pointer args rather than implicitly using
26309 (legitimate_constant_pool_address_p, rs6000_emit_move,
26310 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26311 tocrel_offset and use in toc_relative_expr_p call.
26312 (print_operand, print_operand_address): Use static tocrel_base_oac
26313 and tocrel_offset_oac.
26314 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26317 2017-06-29 Maya Rashish <coypu@sdf.org>
26319 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26321 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26323 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26324 objects, take into account only the alignment of 'op0' and 'mode1' if
26327 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26329 * ccmp.c (ccmp_tree_comparison_p): New function.
26330 (ccmp_candidate_p): Update to use above function.
26331 (get_compare_parts): New function.
26332 (expand_ccmp_next): Update to use new functions.
26333 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26335 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26336 take mode as argument.
26337 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26338 * expr.c (expand_expr_real_1): Pass mode as argument.
26340 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26342 * combine.c (combine_instructions): Print insns to dump_file, together
26345 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26347 * asan.c (asan_emit_stack_protection): Update.
26348 (create_cond_insert_point): Update.
26349 * auto-profile.c (afdo_propagate_circuit): Update.
26350 * basic-block.h (struct edge_def): Turn probability to
26351 profile_probability.
26352 (EDGE_FREQUENCY): Update.
26353 * bb-reorder.c (find_traces_1_round): Update.
26354 (better_edge_p): Update.
26355 (sanitize_hot_paths): Update.
26356 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26357 (make_single_succ_edge): Update.
26358 (check_bb_profile): Update.
26359 (dump_edge_info): Update.
26360 (update_bb_profile_for_threading): Update.
26361 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26362 probabilitycount to 0.
26363 * cfgbuild.c (compute_outgoing_frequencies): Update.
26364 * cfgcleanup.c (try_forward_edges): Update.
26365 (outgoing_edges_match): Update.
26366 (try_crossjump_to_edge): Update.
26367 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26368 (expand_gimple_tailcall): Update.
26369 (construct_init_block): Use make_single_succ_edge.
26370 (construct_exit_block): Use make_single_succ_edge.
26371 * cfghooks.c (verify_flow_info): Update.
26372 (redirect_edge_succ_nodup): Update.
26373 (split_edge): Update.
26374 (account_profile_record): Update.
26375 * cfgloopanal.c (single_likely_exit): Update.
26376 * cfgloopmanip.c (scale_loop_profile): Update.
26377 (set_zero_probability): Remove.
26378 (duplicate_loop_to_header_edge): Update.
26379 * cfgloopmanip.h (loop_version): Update prototype.
26380 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26381 (force_nonfallthru_and_redirect): Update.
26382 (update_br_prob_note): Update.
26383 (rtl_verify_edges): Update.
26384 (purge_dead_edges): Update.
26385 (rtl_lv_add_condition_to_bb): Update.
26386 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26387 * cgraphunit.c (init_lowered_empty_function): Update.
26388 (cgraph_node::expand_thunk): Update.
26389 * cilk-common.c: Include profile-count.h
26390 * dojump.c (inv): Remove.
26391 (jumpifnot): Update.
26392 (jumpifnot_1): Update.
26393 (do_jump_1): Update.
26395 (do_jump_by_parts_greater_rtx): Update.
26396 (do_compare_rtx_and_jump): Update.
26397 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26398 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26399 * dwarf2cfi.c: Include profile-count.h
26400 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26401 (sjlj_emit_dispatch_table): Likewise.
26402 * explow.c: Include profile-count.h
26403 * expmed.c (emit_store_flag_force): Update.
26404 (do_cmp_and_jump): Update.
26405 * expr.c (compare_by_pieces_d::generate): Update.
26406 (compare_by_pieces_d::finish_mode): Update.
26407 (emit_block_move_via_loop): Update.
26408 (store_expr_with_bounds): Update.
26409 (store_constructor): Update.
26410 (expand_expr_real_2): Update.
26411 (expand_expr_real_1): Update.
26412 * expr.h (try_casesi, try_tablejump): Update prototypes.
26413 * gimple-pretty-print.c (dump_probability): Update.
26414 (dump_profile): New.
26415 (dump_gimple_label): Update.
26416 (dump_gimple_bb_header): Update.
26417 * graph.c (draw_cfg_node_succ_edges): Update.
26418 * hsa-gen.c (convert_switch_statements): Update.
26419 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26420 (find_if_case_1): Update.
26421 (find_if_case_2): Update.
26422 * internal-fn.c (expand_arith_overflow_result_store): Update.
26423 (expand_addsub_overflow): Update.
26424 (expand_neg_overflow): Update.
26425 (expand_mul_overflow): Update.
26426 (expand_vector_ubsan_overflow): Update.
26427 * ipa-cp.c (good_cloning_opportunity_p): Update.
26428 * ipa-split.c (split_function): Use make_single_succ_edge.
26429 * ipa-utils.c (ipa_merge_profiles): Update.
26430 * loop-doloop.c (add_test): Update.
26431 (doloop_modify): Update.
26432 * loop-unroll.c (compare_and_jump_seq): Update.
26433 (unroll_loop_runtime_iterations): Update.
26434 * lra-constraints.c (lra_inheritance): Update.
26435 * lto-streamer-in.c (input_cfg): Update.
26436 * lto-streamer-out.c (output_cfg): Update.
26437 * mcf.c (adjust_cfg_counts): Update.
26438 * modulo-sched.c (sms_schedule): Update.
26439 * omp-expand.c (expand_omp_for_init_counts): Update.
26440 (extract_omp_for_update_vars): Update.
26441 (expand_omp_ordered_sink): Update.
26442 (expand_omp_for_ordered_loops): Update.
26443 (expand_omp_for_generic): Update.
26444 (expand_omp_for_static_nochunk): Update.
26445 (expand_omp_for_static_chunk): Update.
26446 (expand_cilk_for): Update.
26447 (expand_omp_simd): Update.
26448 (expand_omp_taskloop_for_outer): Update.
26449 (expand_omp_taskloop_for_inner): Update.
26450 * omp-simd-clone.c (simd_clone_adjust): Update.
26451 * optabs.c (expand_doubleword_shift): Update.
26452 (expand_abs): Update.
26453 (emit_cmp_and_jump_insn_1): Update.
26454 (expand_compare_and_swap_loop): Update.
26455 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26456 * predict.c (predictable_edge_p): Update.
26457 (edge_probability_reliable_p): Update.
26458 (set_even_probabilities): Update.
26459 (combine_predictions_for_insn): Update.
26460 (combine_predictions_for_bb): Update.
26461 (propagate_freq): Update.
26462 (estimate_bb_frequencies): Update.
26463 (force_edge_cold): Update.
26464 * profile-count.c (profile_count::dump): Add missing space into dump.
26465 (profile_count::debug): Add newline.
26466 (profile_count::differs_from_p): Explicitly convert to unsigned.
26467 (profile_count::stream_in): Update.
26468 (profile_probability::dump): New member function.
26469 (profile_probability::debug): New member function.
26470 (profile_probability::differs_from_p): New member function.
26471 (profile_probability::differs_lot_from_p): New member function.
26472 (profile_probability::stream_in): New member function.
26473 (profile_probability::stream_out): New member function.
26474 * profile-count.h (profile_count_quality): Rename to ...
26475 (profile_quality): ... this one.
26476 (profile_probability): New.
26477 (profile_count): Update.
26478 * profile.c (compute_branch_probabilities): Update.
26479 * recog.c (peep2_attempt): Update.
26480 * sched-ebb.c (schedule_ebbs): Update.
26481 * sched-rgn.c (find_single_block_region): Update.
26482 (compute_dom_prob_ps): Update.
26483 (schedule_region): Update.
26484 * sel-sched-ir.c (compute_succs_info): Update.
26485 * stmt.c (struct case_node): Update.
26486 (do_jump_if_equal): Update.
26487 (get_outgoing_edge_probs): Update.
26488 (conditional_probability): Update.
26489 (emit_case_dispatch_table): Update.
26490 (expand_case): Update.
26491 (expand_sjlj_dispatch_table): Update.
26492 (emit_case_nodes): Update.
26493 * targhooks.c: Update.
26494 * tracer.c (better_p): Update.
26495 (find_best_successor): Update.
26496 * trans-mem.c (expand_transaction): Update.
26497 * tree-call-cdce.c: Update.
26498 * tree-cfg.c (gimple_split_edge): Upate.
26499 (move_sese_region_to_fn): Upate.
26500 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26501 * tree-eh.c (lower_resx): Upate.
26502 (cleanup_empty_eh_move_lp): Upate.
26503 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26504 * tree-inline.c (copy_edges_for_bb): Update.
26505 (copy_cfg_body): Update.
26506 * tree-parloops.c (gen_parallel_loop): Update.
26507 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26508 (gimple_gen_time_profiler): Update.
26509 * tree-ssa-dce.c (remove_dead_stmt): Update.
26510 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26511 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26512 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26513 (unloop_loops): Update.
26514 (try_peel_loop): Update.
26515 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26516 * tree-ssa-loop-split.c (connect_loops): Update.
26517 (split_loop): Update.
26518 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26519 (hoist_guard): Update.
26520 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26521 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26522 (value_replacement): Update.
26523 * tree-ssa-reassoc.c (branch_fixup): Update.
26524 * tree-ssa-tail-merge.c (replace_block_by): Update.
26525 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26526 (create_edge_and_update_destination_phis): Update.
26527 (compute_path_counts): Update.
26528 (recompute_probabilities): Update.
26529 (update_joiner_offpath_counts): Update.
26530 (freqs_to_counts_path): Update.
26531 (duplicate_thread_path): Update.
26532 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26533 (struct switch_conv_info): Update.
26534 (gen_inbound_check): Update.
26535 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26536 (vect_do_peeling): Update.
26537 (vect_loop_versioning): Update.
26538 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26539 (optimize_mask_stores): Update.
26540 * ubsan.c (ubsan_expand_null_ifn): Update.
26541 * value-prof.c (gimple_divmod_fixed_value): Update.
26542 (gimple_divmod_fixed_value_transform): Update.
26543 (gimple_mod_pow2): Update.
26544 (gimple_mod_pow2_value_transform): Update.
26545 (gimple_mod_subtract): Update.
26546 (gimple_mod_subtract_transform): Update.
26547 (gimple_ic): Update.
26548 (gimple_stringop_fixed_value): Update.
26549 (gimple_stringops_transform): Update.
26550 * value-prof.h: Update.
26552 2017-06-29 Carl Love <cel@us.ibm.com>
26554 * config/rs6000/rs6000-c.c: Add support for built-in functions
26555 vector signed int vec_signed (vector float);
26556 vector signed long long vec_signed (vector double);
26557 vector signed int vec_signed2 (vector double, vector double);
26558 vector signed int vec_signede (vector double);
26559 vector signed int vec_signedo (vector double);
26560 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26561 instruction generator.
26562 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26563 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26564 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26566 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26567 vunsignede_v2df): Add define_expands.
26568 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26569 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26570 VEC_UNSIGNEDO): Add definitions.
26571 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26572 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26573 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26574 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26575 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26576 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26577 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26578 vec_unsignede, vec_unsignedo): Add builtin defines.
26579 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26581 * doc/extend.texi: Update the built-in documentation file for the
26582 new built-in functions.
26584 2017-06-29 Richard Biener <rguenther@suse.de>
26586 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26587 reduction chains to LOOP_VINFO_REDUCTIONS.
26588 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26589 SLP reductions after processing reduction chains.
26591 2017-06-29 Nathan Sidwell <nathan@acm.org>
26593 * builtins.c (fold_builtin_FUNCTION): Use
26594 lang_hooks.decl_printable_name.
26596 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26598 PR middle-end/81194
26599 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26600 with only one label.
26601 * stmt.c (expand_case): Assert NCASES is greater than one.
26603 2017-06-29 Richard Biener <rguenther@suse.de>
26605 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26607 (group_case_labels): Likewise.
26608 (find_taken_edge): Push sanity checking on val to workers...
26609 (find_taken_edge_cond_expr): ... here
26610 (find_taken_edge_switch_expr): ... and here, handle cases
26611 with just a default label.
26612 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26613 (group_case_labels): Likewise.
26614 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26615 group_case_labels does anything cleanup the CFG again.
26617 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26619 PR tree-optimization/81196
26620 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26621 exit condition comparing two IVs.
26623 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26625 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26626 profile to the dummy entry at the end of the list of architectures.
26627 * config/arm/arm-cpu-cdata.h: Regenerated.
26629 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26630 Michael Collison <michael.collison@arm.com>
26633 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26635 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26636 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26637 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26638 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26639 of shift when the shift amount is masked with constant equal to
26640 the size of the mode.
26641 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26644 2017-06-29 Martin Liska <mliska@suse.cz>
26646 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26648 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26651 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26654 (edge_growth_cache_entry::edge_growth_cache_entry): New
26656 (reset_edge_growth_cache): Update to use constructor.
26658 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26660 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26661 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26662 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26664 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26666 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26667 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26669 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26671 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26672 (*-linux-uclibc*): Add t-uclibc tmake_file.
26673 * config/t-musl: New.
26674 * config/t-uclibc: New.
26676 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26678 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26680 (gen_comm_data): Emit architectural setting of arch_prof.
26681 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26683 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26684 (armv8-m.base, armv8-m.main): Likewise.
26685 * arm-protos.h (arm_build_target): Add profile field.
26686 (arch_option): Likewise.
26687 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26689 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26690 arm_active_target.profile.
26692 2017-06-28 Richard Biener <rguenther@suse.de>
26694 PR middle-end/81227
26695 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26696 TYPE_OVERFLOW_WRAPS.
26697 * match.pd (negate_expr_p): Likewise.
26698 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26699 fold_build2, not fold_binary.
26701 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26703 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26704 Convert memory address to Pmode.
26705 (aarch64_print_operand): Assert MEM operands are always Pmode.
26707 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26710 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26711 Remove redundant if.
26712 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26713 * config/arm/aarch-common-protos.h
26714 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26715 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26717 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26720 * multiple_target.c (create_dispatcher_calls): Set the default
26721 clone to be static, not public.
26723 2017-06-28 Richard Biener <rguenther@suse.de>
26725 * tree-vect-loop.c (vectorizable_reduction): Move special
26726 cond reduction IV var creation ...
26727 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26728 parameter. Use STMT_VINFO_VECTYPE.
26729 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26732 2017-06-28 Martin Liska <mliska@suse.cz>
26735 * ipa-visibility.c (non_local_p): Handle visibility.
26737 2017-06-28 Martin Liska <mliska@suse.cz>
26740 * common.opt: Add IntegerRange to various options.
26741 * opt-functions.awk (integer_range_info): New function.
26742 * optc-gen.awk: Add integer_range_info to cl_options struct.
26743 * opts-common.c (decode_cmdline_option): Handle
26744 CL_ERR_INT_RANGE_ARG.
26745 (cmdline_handle_error): Likewise.
26746 * opts.c (print_filtered_help): Show valid interval in
26747 when --help is provided.
26748 * opts.h (struct cl_option): Add range_min and range_max fields.
26749 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26751 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26753 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26754 (x * C EQ/NE y * C): New transformation.
26756 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26758 * genmultilib (combination_space): Accept '+' in option names.
26760 2017-06-28 Martin Liska <mliska@suse.cz>
26763 * asan.c (instrument_derefs): Bail out inner references
26764 that are hard register variables.
26766 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26769 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26770 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26772 2017-06-28 Richard Biener <rguenther@suse.de>
26774 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26775 (vect_get_slp_defs): Adjust.
26776 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26778 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26780 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26781 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26782 (vectorizable_reduction): Adjust.
26783 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26785 (vect_get_slp_defs): Likewise.
26786 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26787 (vectorizable_bswap): Adjust.
26788 (vectorizable_call): Likewise.
26789 (vectorizable_conversion): Likewise.
26790 (vectorizable_assignment): Likewise.
26791 (vectorizable_shift): Likewise.
26792 (vectorizable_operation): Likewise.
26793 (vectorizable_store): Likewise.
26794 (vectorizable_condition): Likewise.
26795 (vectorizable_comparison): Likewise.
26797 2017-06-28 Michael Collison <michael.collison@arm.com>
26800 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
26802 (arm_gen_movmemqi): Removed unused variable 'i'.
26803 Convert 'for' loop into 'while' loop.
26804 (arm_expand_prologue): Remove last unnecessary set of insn.
26805 (thumb_pop): Remove unused variable 'pushed_words'.
26806 (thumb_exit): Remove last unnecessary set of regs_to_pop.
26808 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26810 * config/s390/predicates.md: Use s390_rel_address_ok_p.
26811 * config/s390/s390-protos.h: Add prototype of
26812 s390_rel_address_ok_p.
26813 * config/s390/s390.c (s390_got_symbol): New function.
26814 (s390_rel_address_ok_p): New function.
26815 (legitimize_pic_address): Use s390_rel_address_ok_p.
26816 (s390_load_got): Use s390_got_symbol.
26817 (s390_option_override): Issue error if
26818 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
26819 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
26821 * config/s390/s390.opt: New option mpic-data-is-text-relative.
26823 2017-06-27 Andrew Pinski <apinski@cavium.com>
26825 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
26826 (X * copysign (1.0, X)): New pattern.
26827 (X * copysign (1.0, -X)): New pattern.
26828 (copysign (-1.0, CST)): New pattern.
26830 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26832 * genmultilib (combination_space): Remove variable.
26833 Validate reuse rules against regular expression for any sequence
26834 of multilib options in any order.
26836 2017-06-27 Michael Collison <michael.collison@arm.com>
26838 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26839 call aarch64_split_simd_combine.
26840 * (aarch64_combine_internal<mode>): Delete pattern.
26841 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26842 Allow register and subreg operands.
26844 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26846 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26847 specific need, just fallback on defaults.
26848 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26850 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26851 Olivier Hainque <hainque@adacore.com>
26853 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
26855 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
26856 targets. Pick a default if no particular attempt applied.
26857 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26860 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26862 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26863 (x86_64-wrs-vxworks7): Likewise.
26865 2017-06-27 Marek Polacek <polacek@redhat.com>
26868 * ubsan.c (instrument_null): Check get_base_address's result for null.
26870 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26872 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
26874 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26876 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
26877 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
26878 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
26879 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
26880 New function types.
26881 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
26882 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
26883 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26884 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26885 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26886 BUILT_IN_FEUPDATEENV): New builtins.
26887 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26888 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26889 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26890 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26892 (builtin_structptr_types): Adjust size.
26893 * tree.c (builtin_structptr_types): Add four entries.
26895 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26896 Olivier Hainque <hainque@adacore.com>
26898 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26899 (TLS_SYM): New local macro, forcing reference to __tls__ on
26900 link command lines for VxWorks 7 RTPs, triggering initialization
26902 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26903 OS features TLS support, true for RTPs on VxWorks 7.
26904 * config/vxworks.c (vxworks_override_options): Setup emutls
26907 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26909 * predict.c (test_prediction_value_range): Use -1U instead of -1
26910 to avoid narrowing conversion warning.
26911 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26912 to avoid narrowing conversion warning.
26913 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26915 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26917 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26919 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26920 64bit configurations.
26921 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26922 (SIZE_TYPE): Likewise.
26923 * config/vxworks.c (vxworks_emutls_var_fields): Use
26924 long_unsigned_type_node instead of unsigned_type_node as the offset
26925 field type, which is "pointer" mode in emutls.c.
26927 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26930 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26932 PR middle-end/81207
26933 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26934 gimple_vuse copying separately from gimple_vdef copying.
26936 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26938 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26941 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26942 Olivier Hainque <hainque@adacore.com>
26944 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26945 for all vxworks7 targets.
26946 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26947 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26948 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26949 variations for VX6/VX7 and 32/64bits later on in ...
26950 (VXWORKS_LIB_SPEC): Leverage new macros.
26951 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26952 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26954 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26956 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26957 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26959 2017-06-26 Carl Love <cel@us.ibm.com>
26961 * config/rs6000/rs6000-c.c: Add support for built-in functions
26962 vector bool char vec_reve (vector bool char);
26963 vector signed char vec_reve (vector signed char);
26964 vector unsigned char vec_reve (vector unsigned char);
26965 vector bool int vec_reve (vector bool int);
26966 vector signed int vec_reve (vector signed int);
26967 vector unsigned int vec_reve (vector unsigned int);
26968 vector bool long long vec_reve (vector bool long long);
26969 vector signed long long vec_reve (vector signed long long);
26970 vector unsigned long long vec_reve (vector unsigned long long);
26971 vector bool short vec_reve (vector bool short);
26972 vector signed short vec_reve (vector signed short);
26973 vector double vec_reve (vector double);
26974 vector float vec_reve (vector float);
26975 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26976 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26977 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26978 (altivec_vreve): New pattern.
26979 * config/rs6000/altivec.h (vec_reve): New define.
26980 * doc/extend.texi (vec_rev): Update the built-in documentation file
26981 for the new built-in functions.
26983 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26985 PR tree-optimization/71815
26986 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26988 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26990 (slsr_process_phi): Likewise.
26991 (replace_uncond_cands_and_profitable_phis): Don't replace a
26992 multiply candidate with a stride of 1 (copy or cast).
26993 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26995 (lowest_cost_path): Likewise.
26996 (total_savings): Likewise.
26998 2017-06-26 Richard Biener <rguenther@suse.de>
27001 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
27002 Use def_builtin_pure for all gather builtins.
27004 2017-06-26 Richard Biener <rguenther@suse.de>
27006 PR tree-optimization/81203
27007 * tree-tailcall.c (find_tail_calls): Do not move stmts into
27008 non-dominating BBs.
27010 2017-06-26 Marek Polacek <polacek@redhat.com>
27013 * doc/invoke.texi: Document -Wmultistatement-macros.
27015 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27017 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27018 arm_neon_ok_no_float_abi effective target.
27020 2017-06-26 Richard Biener <rguenther@suse.de>
27022 PR tree-optimization/80928
27023 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27024 (copy_bbs): Set BB_DUPLICATED flag early.
27025 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27027 (execute_on_shrinking_pred): Likewise.
27028 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27029 BB_DUPLICATED blocks.
27030 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27031 iterate over all PHIs considering removal of *gsi.
27033 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27035 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27038 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27040 * config/rs6000/rs6000-string.c: (expand_block_clear,
27041 do_load_for_compare, select_block_compare_mode,
27042 compute_current_alignment, expand_block_compare,
27043 expand_strncmp_align_check, expand_strn_compare,
27044 expand_block_move, rs6000_output_load_multiple)
27045 Move functions related to string/block move/compare
27046 to a separate file.
27047 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27048 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27049 for this function which is now used in two files.
27050 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27051 * config.gcc: Add rs6000-string.o to extra_objs for
27052 targets powerpc*-*-* and rs6000*-*-*.
27054 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27057 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27058 32-bit, since indexed is not valid for DImode.
27059 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27060 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27061 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27062 (define_peephole2 for Altivec d-form store): Likewise.
27065 * multiple_target.c (create_dispatcher_calls): Only create the
27066 dispatcher call if the function is the default clone of a
27067 versioned function.
27069 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27071 PR middle-end/80902
27072 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27073 a call, force the call to not be a tail call.
27075 2017-06-23 Jeff Law <law@redhat.com>
27077 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27080 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27082 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27083 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27084 handling for early expansion of vector shifts (sl,sr,sra,rl).
27085 (builtin_function_type): Add vector shift right instructions
27086 to the unsigned argument list.
27088 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27090 rtl-optimizatoin/79286
27091 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27092 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27093 trap. PIC register plus a const unspec without offset can never trap.
27095 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27097 * tree.h (builtin_structptr_type): New type.
27098 (builtin_structptr_types): Declare new array.
27099 * tree.c (builtin_structptr_types): New array.
27100 (free_lang_data, build_common_tree_nodes): Use it.
27102 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27105 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27108 2017-06-22 Matt Turner <mattst88@gmail.com>
27110 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27111 Lake models to skylake case. Assume skylake for unknown
27112 models with clflushopt.
27114 2017-06-22 Jeff Law <law@redhat.com>
27116 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27117 frame sizes that do not satisfy aarch64_uimm12_shift.
27119 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27121 * profile-count.h (apply_probability,
27122 apply_scale, probability_in): Fix checks for zero.
27124 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27126 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27127 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27129 2016-06-22 Richard Biener <rguenther@suse.de>
27131 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27132 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27133 REDUC_MAX_EXPR support.
27134 (vectorizable_reduction): Likewise.
27135 (vect_create_epilog_for_reduction): Likewise.
27137 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27139 * match.pd (A / (1 << B) -> A >> B): New.
27140 * generic-match-head.c: Include optabs-tree.h.
27141 * gimple-match-head.c: Likewise.
27142 * optabs-tree.h (target_supports_op_p): New.
27143 * optabs-tree.c (target_supports_op_p): New.
27145 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27147 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27148 $gcc_cv_ld --help output.
27149 (gcc_cv_ld_demangle): Likewise.
27150 (gcc_cv_ld_eh_frame_hdr): Likewise.
27151 (gcc_cv_ld_pie): Likewise.
27152 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27153 (gcc_cv_ld_buildid): Likewise.
27154 (gcc_cv_ld_sysroot): Likewise.
27155 (ld_bndplt_support): Likewise.
27156 (ld_pushpopstate_support): Likewise.
27157 * configure: Regenerate.
27158 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27160 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27163 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27164 operands indexes to avoid gap between operands and match_dups.
27166 2017-06-21 Andrew Pinski <apinski@cavium.com>
27168 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27169 Increment Arith_shift and Arith_shift_reg by 1.
27170 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27172 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27173 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27174 (aarch64_strip_extend): Add new argument and test for it.
27175 (aarch64_cheap_mult_shift_p): New function.
27176 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27177 add a cost if it is true.
27178 Update calls to aarch64_strip_extend.
27179 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27181 2017-06-21 Andrew Pinski <apinski@cavium.com>
27183 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27185 (thunderxt88): Likewise.
27186 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27187 (thunderx_prefetch_tune): New variable.
27188 (thunderx2t99_prefetch_tune): Update for the correct values.
27189 (thunderxt88_tunings): New variable.
27190 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27191 generic_prefetch_tune.
27192 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27194 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27196 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27197 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27198 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27199 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27200 (aarch64_atomic_cas<mode>, GPI): Likewise.
27202 2017-06-21 Martin Liska <mliska@suse.cz>
27204 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27205 statements on cold and hot labels.
27206 * predict.c (tree_estimate_probability_bb): Remove the
27207 prediction from this place.
27209 2017-06-21 Martin Liska <mliska@suse.cz>
27211 PR tree-optimization/79489
27212 * gimplify.c (maybe_add_early_return_predict_stmt): New
27214 (gimplify_return_expr): Call the function.
27215 * predict.c (tree_estimate_probability_bb): Remove handling
27217 * predict.def: Update comment about early return predictor.
27218 * gimple-predict.h (is_gimple_predict): New function.
27219 * predict.def: Change default value of early return to 66.
27220 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27222 * passes.def: Put pass_strip_predict_hints to the beginning of
27225 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27227 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27228 FUNCTION_DECL declarations.
27229 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27231 (dwaf2out_decl): Likewise.
27232 * godump.c (go_early_global_decl): Skip call to the real debug hook
27233 for FUNCTION_DECL declarations.
27234 * passes.c (rest_of_decl_compilation): Skip call to the
27235 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27236 -fdump-go-spec is passed.
27238 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27240 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27241 Reorder for compactness.
27242 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27243 (def_builtin_pure, def_builtin_pure2): New functions.
27244 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27246 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27248 * match.pd (nop_convert): New predicate.
27249 ((A +- CST1) +- CST2): Allow some NOP conversions.
27251 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27254 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27255 with ctors/dtors if GOVD_SHARED is set.
27257 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27259 * config/aarch64/aarch64.md (movti_aarch64):
27260 Emit mov rather than orr.
27261 (movtf_aarch64): Likewise.
27262 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27263 Emit mov rather than orr.
27265 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27267 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27268 Swap alternatives, make integer dup more expensive.
27270 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27272 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27273 Return true for non-tls symbols.
27275 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27277 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27278 (cortex-a75): Likewise.
27279 (cortex-a75.cortex-a55): Likewise.
27280 * config/aarch64/aarch64-tune.md: Regenerate.
27281 * doc/invoke.texi (-mtune): Document new values for -mtune.
27283 2017-06-21 Tom de Vries <tom@codesourcery.com>
27285 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27286 stack_size feature.
27287 (Effective-Target Keywords, Other attributes): Suggest using
27288 dg-add-options stack_size feature to get stack limit in stack_size
27289 effective target documentation.
27291 2017-06-21 Julian Brown <julian@codesourcery.com>
27292 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27294 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27295 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27296 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27298 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27299 attribute type list for neon_multiply.
27300 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27301 attribute type list for neon_multiply.
27302 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27303 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27304 attribute type list for neon_multiply.
27305 * config/arm/types.md (crypto_pmull): Add.
27306 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27307 attribute type list.
27309 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27311 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27314 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27316 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27317 to make sure not to dereference a NULL cost_classes_ptr pointer.
27319 2017-06-20 Carl Love <cel@us.ibm.com>
27321 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27322 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27323 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27324 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27325 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27326 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27327 VMULOSW): New enum "unspec" values.
27328 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27329 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27330 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27331 altivec_vmulosw): New patterns.
27332 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27333 VMULOSW): Add definitions.
27335 2017-06-20 Julia Koval <julia.koval@intel.com>
27337 * config/i386/i386.c: Fix rounding expand for new pattern.
27338 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27340 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27342 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27343 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27345 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27347 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27350 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27352 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27353 architecture, then by implementer ID.
27354 * config/aarch64/aarch64-tune.md: Regenerate.
27356 2017-06-20 Richard Biener <rguenther@suse.de>
27358 PR middle-end/81097
27359 * fold-const.c (split_tree): Fold to type before negating.
27361 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27363 * diagnostic-show-locus.c
27364 (selftest::test_fixit_deletion_affecting_newline): New function.
27365 (selftest::diagnostic_show_locus_c_tests): Call it.
27367 2017-06-20 Andreas Schwab <schwab@suse.de>
27370 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27373 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27375 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27376 __ARM_FEATURE_COPROC according to support.
27378 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27380 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27381 Rewritten to avoid overflow for > 32-bit pointers.
27384 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27385 by removing enum keyword.
27386 (ubsan_type_descriptor): Likewise. Formatting fix.
27389 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27390 splitter): Require TARGET_SSE2 in the condition.
27392 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27395 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27396 for doing vector set of SFmode on ISA 3.0.
27397 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27398 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27400 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27401 SFmode value into a V4SF variable that was extracted from another
27402 V4SF variable without converting the element to double precision
27403 and back to single precision vector format.
27404 (vsx_insert_extract_v4sf_p9_2): Likewise.
27406 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27408 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27409 in UWHI to avoid undefined overflow.
27412 * ubsan.h (enum ubsan_encode_value_phase): New.
27413 (ubsan_encode_value): Change second argument to
27414 enum ubsan_encode_value_phase with default value of
27415 UBSAN_ENCODE_VALUE_GENERIC.
27416 * ubsan.c (ubsan_encode_value): Change second argument to
27417 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27418 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27419 create_tmp_var_raw instead of create_tmp_var and use a
27421 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27422 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27423 ubsan_encode_value callers.
27426 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27427 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27428 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27430 2017-06-19 Richard Biener <rguenther@suse.de>
27432 PR middle-end/81118
27433 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27434 estimates if we changed anything.
27436 2017-06-19 Richard Biener <rguenther@suse.de>
27438 PR tree-optimization/80887
27439 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27440 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27441 simplified lookups, then reset mprts_hook.
27442 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27444 (try_to_simplify): Likewise.
27446 2017-06-19 Martin Liska <mliska@suse.cz>
27449 * gimplify.c (gimplify_switch_expr):
27450 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27452 2017-06-19 Martin Liska <mliska@suse.cz>
27454 * doc/install.texi: Document that PGO runs in 4 stages.
27456 2017-06-19 Martin Liska <mliska@suse.cz>
27459 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27460 to dispatcher function name.
27461 * multiple_target.c (replace_function_decl): New function.
27462 (create_dispatcher_calls): Redirect both edges and references.
27464 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27466 * profile-count.c (profile_count::dump): Dump quality.
27467 (profile_count::differs_from_p): Update for unsigned val.
27468 * profile-count.h (profile_count_quality): New enum.
27469 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27471 2017-06-19 Richard Biener <rguenther@suse.de>
27473 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27474 struct function as arg.
27475 (estimate_numbers_of_iterations): Export overload with loop arg.
27476 (free_numbers_of_iterations_estimates_loop): Use an overload of
27477 free_numbers_of_iterations_estimates instead.
27478 * tree-cfg.c (remove_bb): Adjust.
27479 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27480 * tree-parloops.c (gen_parallel_loop): Likewise.
27481 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27483 (tree_unroll_loops_completely): Likewise.
27484 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27485 Use an overload instead and export.
27486 (estimated_loop_iterations): Adjust.
27487 (max_loop_iterations): Likewise.
27488 (likely_max_loop_iterations): Likewise.
27489 (estimate_numbers_of_iterations): Take struct function as arg
27491 (loop_exits_before_overflow): Adjust.
27492 (free_numbers_of_iterations_estimates_loop): Use an overload.
27493 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27494 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27496 2017-06-19 Richard Biener <rguenther@suse.de>
27499 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27500 RANGE_EXPR conservatively.
27502 2017-06-16 Carl Love <cel@us.ibm.com>
27504 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27505 definitions for vec_float, vec_float2, vec_floato,
27506 vec_floate built-ins.
27507 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27508 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27510 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27511 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27512 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27513 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27514 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27515 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27516 vec_floato): Add builtin defines.
27517 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27518 Update the built-in documentation file for the new built-in
27521 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27523 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27524 (mthumb): Mark as the negative of -marm.
27526 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27528 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27530 (ARM Options, -mtune): Document that this accepts the same
27531 extension options as -mcpu.
27532 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27534 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27536 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27537 permitted extensions.
27539 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27541 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27542 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27543 (armv8-m.main): Add option +nodsp.
27544 * config/arm/arm-cpu-cdata.h: Regenerated.
27546 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27548 * config/arm/t-fuchsia: New file.
27549 * config.gcc (arm*-*-fuchsia*): Use it.
27551 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27553 * config/arm/t-symbian: Rewrite for new option infrastructure.
27555 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27557 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27558 (MULTILIB_REQUIRED): Likewise.
27560 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27562 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27563 (MULTILIB_RESUE): Likewise.
27564 (MULTILIB_MATCHES): Likewise.
27565 (MULTLIB_REQUIRED): Likewise.
27567 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27569 * config/arm/t-rtems: Rewrite for new option framework.
27571 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27573 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27574 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27575 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27576 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27577 * config/arm/t-multilib: ... here.
27578 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27579 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27580 armv7-a and armv8*-a when A-profile libraries have not been built.
27581 * config/arm/t-rmprofile: Rewrite.
27583 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27585 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27586 with a backslash. Remove the backslash after substituting unescaped
27588 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27590 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27592 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27593 the driver to pass through the default setting of -mfloat-abi.
27594 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27596 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27597 (all_feat_combs): New rule.
27598 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27600 * config/arm/t-aprofile: Rewrite.
27602 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27604 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27605 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27606 fpu is not specified by the user/command-line.
27607 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27608 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27609 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27610 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27611 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27612 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27614 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27616 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27617 * config/arm/t-arm-elf: Rewritten.
27619 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27621 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27622 have some floating-point instructions.
27623 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27624 (TARGET_MAYBE_HARD_FLOAT): New macro.
27625 * config/arm/arm-builtins.c (arm_init_builtins): Use
27626 TARGET_MAYBE_HARD_FLOAT.
27627 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27629 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27631 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27632 (configargs.h): Include it.
27633 (arm_print_hint_for_fpu_option): New function.
27634 (arm_parse_fpu_option): New function.
27635 (candidate_extension): New class.
27636 (arm_canon_for_multilib): New function.
27637 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27638 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27639 (ARCH_CANONICAL_SPECS): New macro.
27640 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27642 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27644 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27645 are set after handling multilib fragments. Set target_cpu_default2
27648 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27650 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27652 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27653 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27655 (check_arch): Likewise.
27656 * config/arm/arm.c (arm_configure_build_target): Handle
27657 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27658 options in the default.
27660 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27662 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27663 when an option is an alias of another.
27664 * config/arm/parsecpu.awk (optalias): New parser token.
27665 (gen_comm_data): Mark non-alias options as such. Emit entries
27666 for extension aliases.
27667 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27668 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27669 (armv6kz, armv6zk, armv6t2): Likewise.
27670 (armv7): Make vfpv3-d16 an alias.
27671 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27673 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27674 Sort in canonical order.
27675 (armv8-a): Sort in canonical order.
27676 (armv8.1-a, armv8.2-a): Likewise.
27677 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27679 (cortex-a9): Sort in canonical order.
27680 * config/arm/arm.c (selftests.h): Include it.
27681 (arm_test_cpu_arch_data): New function.
27682 (arm_run_self_tests): New function.
27683 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27684 (targetm): Move declaration to the end of the file.
27685 * arm-cpu-cdata.h: Regenerated.
27687 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27689 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27690 call to target_mode_check describing the type of option passed.
27691 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27692 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27693 arm_parse_cpu_option_name to match parameters against list of
27695 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27696 arm_arch_core_flags data structure.
27697 * config/arm/arm-cpu_cdata.h: Regenerated.
27699 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27701 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27703 (arm_print_hint_for_cpu_option): Likewise.
27704 (arm_print_hint_for_arch_option): Likewise.
27705 (arm_parse_cpu_option_name): Likewise.
27706 (arm_parse_arch_option_name): Likewise.
27707 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27708 of entries in the all_fpus list.
27709 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27710 (arm_parse_cpu_option_name): Declare.
27711 (arm_parse_arch_option_name): Declare.
27712 (arm_parse_option_features): Declare.
27713 (arm_intialize_isa): Declare.
27714 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27716 (gen_comm_data): ... here. Make definitions non-static.
27717 * config/arm/arm-cpu-data.h: Regenerated.
27718 * config/arm/arm-cpu-cdata.h: Regenerated.
27720 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27722 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27723 (cpu_arch_extension): New structure.
27724 (cpu_arch_option, arch_option, cpu_option): New structures.
27725 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27726 architecture types.
27727 (gen_data): Generate new format data tables.
27728 * config/arm/arm.c (cpu_tune): New structure.
27729 (cpu_option, processors): Delete.
27730 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27731 (arm_print_hint_for_cpu_option): ... this and ...
27732 (arm_print_hint_for_arch_option): ... this.
27733 (arm_parse_arch_cpu_name): Delete. Replace with ...
27734 (arm_parse_cpu_option_name): ... this and ...
27735 (arm_parse_arch_option_name): ... this.
27736 (arm_unrecognized_feature): Change type of target parameter to
27738 (arm_parse_arch_cpu_features): Delete. Replace with ...
27739 (arm_parse_option_features): ... this.
27740 (arm_configure_build_target): Rework to use new configuration data
27742 (arm_print_tune_info): Rework for new configuration data tables.
27743 * config/arm/arm-cpu-data.h: Regenerated.
27744 * config/arm/arm-cpu.h: Regenerated.
27746 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27748 * Makefile.in (OBJS): Move sbitmap.o from here ...
27749 (OBJS-libcommon): ... to here.
27751 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27753 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27754 (ISA_ALL_CRYPTO): New macro.
27755 (ISA_ALL_SIMD): New macro
27756 (ISA_ALL_FP): New macro.
27757 * config/arm/arm.c (fpu_bitlist): Update initializer.
27758 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27760 (arm9e): Add fpu. Add option for nofp
27761 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27762 (arm926ej-s, arm1026ej-s): Likewise.
27763 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27764 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27765 neon-fp16, neon-vfpv4, nofp and nosimd.
27766 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27767 (cortex-a8): Add fpu. Add option for nofp.
27768 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27769 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27770 (cortex-r4f): Add fpu.
27771 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27772 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27774 (cortex-r8): Likewise.
27775 (cortex-m4): Add fpu. Add option for nofp.
27776 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27777 (cortex-a17.cortex-a7): Likewise.
27778 (cortex-a32): Add fpu. Add options for crypto and nofp.
27779 (cortex-a35, cortex-a53): Likewise.
27780 (cortex-a57): Add fpu. Add option for crypto.
27781 (cortex-a72, cortex-a73): Likewise.
27782 (exynos-m1): Likewise.
27783 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27784 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27785 (cortex-m33): Add fpu. Add option for nofp.
27786 * config/arm/arm-cpu-cdata.h: Regenerated
27787 * config/arm/arm-cpu-data.h: Regenerated.
27789 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27791 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27792 (armv5te, armv5tej): Likewise.
27793 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27794 (armv7): Add options fp and vfpv3-d16.
27795 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27796 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27798 (armv7ve): Likewise.
27799 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
27800 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
27801 (armv8-a): Add nocrypto option.
27802 (armv8.1-a, armv8.2-a): Likewise.
27803 (armv8-m.main): add options fp, fp.dp and nofp.
27805 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27807 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
27809 (armv8-a+crc): Delete.
27810 (armv8.1-a): Add options simd, crypto and nofp.
27811 (armv8.2-a): Add options fp16, simd, crypto and nofp.
27812 (armv8.2-a+fp16): Delete.
27813 (armv8-m.main): Add option dsp.
27814 (armv8-m.main+dsp): Delete.
27815 (cortex-a8): Add fpu. Add nofp option.
27816 (cortex-a9): Add fpu. Add nofp and nosimd options.
27817 * config/arm/parsecpu.awk (gen_data): Generate option tables and
27818 link to main cpu and architecture data structures.
27819 (gen_comm_data): Only put isa attributes from the main architecture
27821 (option): New statement for architecture and CPU entries.
27822 * arm.c (struct cpu_option): New structure.
27823 (struct processors): Add entry for options.
27824 (arm_unrecognized_feature): New function.
27825 (arm_parse_arch_cpu_name): Ignore any characters after the first
27827 (arm_parse_arch_cpu_feature): New function.
27828 (arm_configure_build_target): Separate out any CPU and architecture
27829 features and parse separately. Don't error out if -mfpu=auto is
27830 used with only an architecture string.
27831 (arm_print_asm_arch_directives): New function.
27832 (arm_file_start): Call it.
27833 * config/arm/arm-cpu-cdata.h: Regenerated.
27834 * config/arm/arm-cpu-data.h: Likewise.
27835 * config/arm/arm-tables.opt: Likewise.
27837 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27839 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27840 assembler when it is not -mfpu=auto.
27842 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27844 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27845 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27846 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27847 (ASM_CPU_SPEC): Rewrite.
27848 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27849 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27850 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27851 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
27852 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
27853 copied string is NUL-terminated. Also strip any characters prefixed
27855 (arm_rewrite_selected_arch): New function.
27856 (arm_rewrite_march): New function.
27858 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27860 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27861 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27862 (march, mcpu, mtune): Convert to string-based options.
27863 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27864 (arm_parse_arch_cpu_name): New function.
27865 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
27866 identify selected architecture or CPU.
27867 (arm_option_save): New function.
27868 (TARGET_OPTION_SAVE): Redefine.
27869 (arm_option_restore): Restore string options.
27870 (arm_option_print): Print string options.
27872 2017-06-16 Martin Sebor <msebor@redhat.com>
27874 PR tree-optimization/80933
27875 PR tree-optimization/80934
27876 * builtins.c (fold_builtin_3): Do not handle bcmp here.
27877 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
27878 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
27879 (gimple_fold_builtin): Call them.
27881 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27883 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27884 as unlikely; update profile.
27886 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27888 * predict.c (force_edge_cold): Handle declaring edges impossible
27891 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27893 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27895 (try_unroll_loop_completely): Fix reporting.
27897 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27899 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27901 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27904 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27905 if given a non-constant argument for an intrinsic which requires a
27908 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27910 * profile.c (compare_freqs): New function.
27911 (branch_prob): Sort edge list.
27912 (find_spanning_tree): Assume that the list is priority sorted.
27914 2017-06-16 Richard Biener <rguenther@suse.de>
27916 PR tree-optimization/81090
27917 * passes.def (pass_record_bounds): Remove.
27918 * tree-pass.h (make_pass_record_bounds): Likewise.
27919 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27920 make_pass_record_bounds): Likewise.
27921 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27922 not free niter estimates at the beginning but at the end.
27923 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27925 2017-06-16 Richard Biener <rguenther@suse.de>
27927 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27928 initializer to workaround ICE in host GCC 4.8.
27930 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27932 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27934 (clone_inlined_nodes): Update.
27936 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27938 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27939 prefetch settings, and enable prefetching by default at -O3.
27941 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27943 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27944 Set flag_prefetch_loop_arrays according to tuning data.
27946 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27948 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27949 New tune structure.
27950 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27951 [Unrelated to main purpose of the patch] Place the pointer field last
27952 to enable type checking errors when tune structure are wrongly merged.
27953 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27954 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27955 (thunderx2t99_prefetch_tune): New tune constants.
27956 (tune_params *_tunings): Update all tunings (no functional change).
27957 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27958 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27959 from tunings structures.
27961 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27964 * ubsan.c (instrument_null): Add T argument, use it instead
27965 of computing it based on IS_LHS.
27966 (instrument_object_size): Likewise.
27967 (pass_ubsan::execute): Adjust instrument_null and
27968 instrument_object_size callers to pass gimple_get_lhs or
27969 gimple_assign_rhs1 result to it. Use instrument_null instead of
27970 calling get_base_address and instrument_mem_ref. Handle
27971 aggregate call arguments for object-size sanitization.
27973 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27975 PR tree-optimization/81089
27976 * tree-vrp.c (is_masked_range_test): Validate operands of
27979 2017-06-15 Martin Sebor <msebor@redhat.com>
27982 * dumpfile.c (dump_register): Avoid calling memset to initialize
27983 a class with a default ctor.
27984 * gcc.c (struct compiler): Remove const qualification.
27985 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27986 * hash-table.h: Ditto.
27987 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27989 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27990 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27992 * params.h (struct param_info): Make struct members non-const.
27993 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27994 with default initialization.
27995 * vec.h (vec_copy_construct, vec_default_construct): New helper
27997 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27998 with vec_copy_construct.
27999 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
28000 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
28001 * doc/invoke.texi (-Wclass-memaccess): Document.
28003 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28005 * emit-rtl.h (is_leaf): Update comment about local
28006 register allocator.
28008 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28011 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28012 for a variable to have a section before checking if the section has a
28014 Set section to.persistent if persistent attribute is set.
28015 Warn if .persistent attribute is used on an automatic variable.
28017 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28019 PR rtl-optimization/80474
28020 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28022 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28024 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28027 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28029 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28030 (casesi): Emit bounds checking as RTL.
28031 (casesi_internal_mips16_<mode>): Remove bounds checking.
28033 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28035 * config/xtensa/xtensa.c (xtensa_option_override): Append
28036 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28037 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28038 xtensa_doloop_hooks): Define unconditionally.
28039 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28041 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28042 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28043 for it in xtensa_option_override.
28044 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28045 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28047 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28049 * doc/cppopts.texi: Document '-' special value to -MF.
28051 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28053 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28054 (cortex_a53_fconst): Likewise.
28055 (cortex_a53_fpmul): Likewise.
28056 (cortex_a53_f_load_64): Likewise.
28057 (cortex_a53_f_load_many): Likewise.
28058 (cortex_a53_advsimd_alu): Likewise.
28059 (cortex_a53_advsimd_alu_q): Likewise.
28060 (cortex_a53_advsimd_mul): Likewise.
28061 (cortex_a53_advsimd_mul_q): Likewise.
28062 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28063 Add missing fmul, r2f_cvt and fconst cases.
28065 2017-06-14 Richard Biener <rguenther@suse.de>
28067 PR middle-end/81088
28068 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28070 (fold_binary_loc): When associating do not treat pre-existing
28071 TREE_OVERFLOW on literal constants as a reason to allow
28072 TREE_OVERFLOW on associated literal constants.
28074 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28076 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28077 (MASK_FEATURES): New macro.
28078 * config/sparc/sparc.c (sparc_option_override): Remove the special
28079 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28081 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28083 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28084 a division of 0 if non-call exceptions are enabled.
28086 2017-06-14 Andrew Pinski <apinski@cavium.com>
28087 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28090 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28091 Improve vector initialization code gen for only variable case.
28093 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28095 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28097 2017-06-14 Richard Biener <rguenther@suse.de>
28099 PR tree-optimization/81083
28100 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28103 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28105 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28106 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28107 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28108 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28109 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28110 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28112 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28114 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28115 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28117 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28119 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28121 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28123 * config/rs6000/t-rtems: Don't handle SPE.
28125 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28127 * config/rs6000/t-linux: Don't handle SPE.
28129 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28131 * config/rs6000/eabispe.h: Delete file.
28133 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28135 * config/rs6000/t-spe: Delete file.
28137 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28139 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28140 (rs6000_legitimate_offset_address_p): Return false for anything in
28141 V2SImode or V2SFmode.
28143 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28145 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28146 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28147 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28149 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28150 (rs6000_legitimate_offset_address_p): Ditto.
28151 (rs6000_emit_move): Ditto.
28152 (rs6000_init_builtins): Remove V4HI_type_node.
28154 2017-06-13 Martin Liska <mliska@suse.cz>
28157 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28158 (gate_asan): Likewise.
28159 * asan.h (asan_no_sanitize_address_p): Remove the function.
28160 (sanitize_flags_p): New function.
28161 * builtins.def: Fix coding style.
28162 * common.opt: Use renamed enum value.
28163 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28164 * doc/extend.texi: Document no_sanitize attribute.
28165 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28166 to SANITIZE_UNDEFINED_NONDEFAULT.
28167 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28168 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28169 Use sanitize_flags_p.
28170 * gimplify.c (gimplify_function_tree): Likewise.
28171 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28172 * opts.c (parse_no_sanitize_attribute): New function.
28173 (common_handle_option): Use renamed enum value.
28174 * opts.h (parse_no_sanitize_attribute): Declare.
28175 * tree.c (sanitize_flags_p): New function.
28176 * tree.h: Declared here.
28177 * tsan.c: Use sanitize_flags_p.
28178 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28179 (instrument_mem_ref): Likewise.
28180 (instrument_bool_enum_load): Likewise.
28181 (do_ubsan_in_current_function): Remove the function.
28182 (pass_ubsan::execute): Use sanitize_flags_p.
28183 * ubsan.h: Remove do_ubsan_in_current_function
28184 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28185 (dump_function_to_file): Use it here.
28187 2017-06-13 Martin Jambor <mjambor@suse.cz>
28189 PR tree-optimization/80803
28190 PR tree-optimization/81063
28191 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28192 (propagate_subaccesses_across_link): Enqueue subtree whenever
28193 necessary instead of relying on the caller.
28195 2017-06-13 Martin Jambor <mjambor@suse.cz>
28197 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28198 that have a first_link.
28199 (sort_and_splice_var_accesses): Do not check first_link before
28201 (subtree_mark_written_and_enqueue): Likewise.
28202 (propagate_all_subaccesses): Likewise and do not stop at first
28203 parent with a first_link.
28205 2017-06-13 Martin Jambor <mjambor@suse.cz>
28207 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28210 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28212 * match.pd: New pattern.
28214 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28216 * tree-vrp.c (is_masked_range_test): New function.
28217 (register_edge_assert_for): Determine ranges for
28220 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28222 PR tree-optimization/67328
28223 * fold-const.c (maskable_range_p): New function.
28224 (build_range_check): Generate bittests if possible.
28226 2017-06-13 Martin Liska <mliska@suse.cz>
28228 * gimple-pretty-print.c (dump_probability): Add new argument.
28229 (dump_edge_probability): Dump both probability and count.
28230 (dump_gimple_label): Likewise.
28231 (dump_gimple_bb_header): Likewise.
28233 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28236 * config/avr/avr-devices.c: Fix indentation.
28237 * config/avr/gen-avr-mmcu-specs.c: Dito.
28239 2017-06-13 Richard Biener <rguenther@suse.de>
28241 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28242 instead get vector type from stmt_info.
28243 (vectorizable_reduction): Adjust. Remove dead code.
28245 2017-06-13 Richard Biener <rguenther@suse.de>
28247 PR middle-end/81065
28248 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28249 case of C * (x * C2 + C3).
28250 (fold_addr_of_array_ref_difference): Properly fold index difference.
28252 2017-06-12 David S. Miller <davem@davemloft.net>
28255 * config/sparc/sparc.md (return expander): Emit frame blockage if
28256 function uses alloca.
28258 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28260 * combine.c (make_field_assignment): Check len rather than the mode
28261 precision when calling force_to_mode.
28263 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28265 Support multilibs and devices that see flash in RAM address range.
28268 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28269 (avr_mcu_t) <flash_pm_offset>: New field.
28270 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28271 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28272 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28273 (AVR_TINY_PM_OFFSET): Remove macro.
28274 * config/avr/avr.opt (-mshort-calls): New option.
28275 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28276 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28277 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28278 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28279 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28280 instead of avr_arch->have_jmp_call.
28281 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28282 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28283 avr_arch->flash_pm_offset to define.
28284 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28285 new field flash_pm_offset. Add entry for avrxmega3.
28286 (avr_texinfo): Add entry for avrxmega3.
28287 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28288 attiny212, attiny214,
28289 attiny412, attiny414, attiny416, attiny417,
28290 attiny814, attiny816, attiny817,
28291 attiny1614, attiny1616, attiny1617,
28292 attiny3214, attiny3216, attiny3217.
28293 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28294 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28295 (avr_print_operand_address) [AVR_TINY]: Same.
28296 (avr_asm_init_sections) <readonly_data_section>: Only patch
28297 callback if avr_arch->flash_pm_offset = 0.
28298 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28299 for rodata if avr_arch->flash_pm_offset != 0.
28300 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28301 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28302 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28303 (m_options): Append opt_rcall.
28304 (m_dirnames): Append dir_rcall.
28305 * config/avr/t-multilib: Regenerate.
28307 * configure.ac [target=avr]: Check whether avrxmega3 default
28308 linker description file works as needed.
28309 * configure: Regenerate.
28310 * doc/avr-mmcu.texi: Regenerate.
28311 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28312 <__AVR_ARCH__>: Document avrxmega3 and 103.
28313 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28314 <__AVR_SHORT_CALLS__>: Document it.
28315 <__AVR_PM_BASE_ADDRESS__>: Document it.
28316 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28317 (AVR Variable Attributes) <progmem>: Document this is
28318 not needed for avrxmega3.
28319 (AVR Named Address Spaces) <__flash>: Dito.
28321 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28323 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28325 2017-06-12 Doug Rupp <rupp@adacore.com>
28327 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28328 Append vxworks-stdint.h to the tm_file list.
28329 * config/vxworks-stdint.h: New file.
28331 2017-06-12 Martin Liska <mliska@suse.cz>
28333 PR tree-optimization/81041
28334 * tree-profile.c (gimple_gen_ic_func_profiler):
28335 Create an extra BB in profile-generate
28336 (gimple_gen_time_profiler): Likewise.
28338 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28340 PR tree-optimization/81003
28341 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28342 (update_range_test): Use it instead of force_gimple_operand_gsi.
28344 2017-06-12 Richard Biener <rguenther@suse.de>
28346 PR tree-optimization/81053
28347 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28348 with backedge value not defined in loop. Simplify def stmt
28351 2017-06-11 Tom de Vries <tom@codesourcery.com>
28354 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28356 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28357 nvptx_cannot_force_const_mem.
28359 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28361 * opts.c (finish_options): Move test for flag_split_stack after
28362 it has been initialized.
28364 2017-06-11 Jason Merrill <jason@redhat.com>
28366 * tree.h (id_equal): New.
28367 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28368 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28369 instead of strcmp of IDENTIFIER_POINTER.
28371 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28373 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28374 (mark_all_inlined_calls_cdtor): Fix formating.
28375 (inline_transform): Rescale profile before inlining.
28377 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28379 * cgraph.h (cgraph_edge::clone): Update prototype.
28380 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28381 (cgraph_node::create_clone): Update.
28382 (cgraph_node::create_version_clone): Update.
28383 * tree-inline.c (copy_bb): Update.
28384 (expand_call_inline): Update.
28386 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28388 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28389 factored out from ...
28390 (rs6000_emit_prologue): ... here.
28392 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28394 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28395 factored out from ...
28396 (rs6000_emit_prologue): ... here.
28398 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28400 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28402 (handle_missing_profiles): Fix computation of tp_first_run.
28403 (counts_to_freqs): Do not touch freqs when count is 0.
28405 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28407 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28410 2017-06-10 Tom de Vries <tom@codesourcery.com>
28412 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28413 attributes): Document signal effective target.
28415 2017-06-10 Tom de Vries <tom@codesourcery.com>
28417 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28418 Document effective target stack_size.
28420 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28422 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28423 to the edit_context if they can be auto-applied.
28425 2017-06-9 Ian Lance Taylor <iant@golang.org>
28427 * opts.c (finish_options): If -fsplit-stack, disable implicit
28428 -forder-blocks-and-partition.
28429 * doc/invoke.texi (Optimize Options): Document that when using
28430 -fsplit-stack -forder-blocks-and-partition is not implicitly
28433 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28435 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28436 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28437 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28438 * builtins.def (abort, trap, unreachable): Declare cold.
28439 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28440 * tree-core.h (ECF_COLD): New.
28441 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28442 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28444 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28446 * predict.c (unlikely_executed_stmt_p): Cleanup.
28448 2017-06-09 Richard Biener <rguenther@suse.de>
28450 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28451 model if the ref is always written to.
28453 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28455 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28457 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28459 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28462 2017-06-09 Tom de Vries <tom@codesourcery.com>
28465 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28466 "target cannot support label values" when encountering LABEL_REF.
28468 2017-06-09 Martin Liska <mliska@suse.cz>
28470 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28471 (gimple_gen_ic_func_profiler): Emit direct comparison
28472 of __gcov_indirect_call_callee with NULL.
28473 (gimple_gen_time_profiler): Change probability from
28474 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28476 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28478 * profile.c (edge_gcov_counts): Turn to pointer.
28479 (compute_branch_probabilities, compute_branch_probabilities): Update.
28480 (branch_prob): Do not clear edge_gcov_count.
28481 * profile.h (edge_gcov_counts): Turn to pointer.
28482 (edge_gcov_count): Update.
28484 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28486 * gimple.h (gimple_check_failed): Mark cold.
28488 2017-06-09 Richard Biener <rguenther@suse.de>
28490 PR tree-optimization/66623
28491 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28492 refactor check_reduction into two parts, properly computing
28493 whether we have to check reduction validity for outer loop
28496 2017-06-09 Richard Biener <rguenther@suse.de>
28498 PR tree-optimization/79483
28499 * graphite-scop-detection.c (order): New global.
28500 (get_order): Compute bb to order mapping that satisfies code
28501 generation constraints.
28502 (cmp_pbbs): New helper.
28503 (build_scops): Start domwalk at entry block, sort generated
28506 2017-06-09 Richard Biener <rguenther@suse.de>
28508 PR middle-end/81007
28509 * ipa-polymorphic-call.c
28510 (ipa_polymorphic_call_context::restrict_to_inner_class):
28511 Skip FIELD_DECLs with error_mark_node type.
28512 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28515 2017-06-09 Martin Liska <mliska@suse.cz>
28517 * predict.c (struct branch_predictor): New struct.
28518 (test_prediction_value_range): New test.
28519 (predict_c_tests): New function.
28520 * selftest-run-tests.c (selftest::run_tests): Run the function.
28521 * selftest.h: Declare new tests.
28523 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28526 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28527 gen_add3_insn did not fail.
28528 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28529 r0, construct that number in a temporary reg and add that reg to r0.
28530 If asked to put the result in r0 as well, fail.
28532 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28534 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28535 for early expansion of vec_eqv.
28537 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28539 PR middle-end/81005
28540 * ubsan.c (instrument_null): Avoid pointless code temporary.
28541 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28543 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28547 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28550 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28553 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28555 * predict.c (unlikely_executed_edge_p): Move ahead.
28556 (probably_never_executed_edge_p): Use it.
28558 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28560 PR middle-end/79988
28561 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28562 gimple_call_builtin_p call.
28564 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28566 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28567 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28568 rtl_check_failed_type2, rtl_check_failed_code1,
28569 rtl_check_failed_code2, rtl_check_failed_code_mode,
28570 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28571 rtvec_check_failed_bounds, rtl_check_failed_flag,
28572 _fatal_insn_not_found, _fatal_insn): Likewise.
28573 * tree.h (tree_contains_struct_check_failed,
28574 tree_check_failed, tree_not_check_failed,
28575 tree_class_check_failed, tree_range_check_failed,
28576 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28577 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28578 tree_operand_check_failed, omp_clause_check_failed,
28579 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28582 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28584 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28585 flag_branch_probabilities.
28586 * ipa-inline.c (edge_badness): Likewise.
28587 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28588 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28589 * predict.c (maybe_hot_frequency_p): Likewise.
28590 (probably_never_executed): Likewise.
28591 * sched-ebb.c (schedule_ebbs): Likewise.
28592 * sched-rgn.c (find_single_block_region): Likewise.
28593 * tracer.c (tail_duplicate): Likewise.
28595 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28597 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28598 longer requires x_flag_profile_use.
28600 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28602 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28603 instead of flag_reorder_blocks_and_partition.
28604 * dbxout.c (dbxout_function_end): Likewise.
28605 * dwarf2out.c (gen_subprogram_die): Likewise.
28606 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28607 * hw-doloop.c (reorg_loops): Likewise.
28608 * varasm.c (assemble_start_function,
28609 assemble_end_function): Likewise.
28610 (decide_function_section): Do not check for
28611 flag_reorder_blocks_and_partition.
28613 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28615 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28617 (chkp_get_hard_register_fake_addr_expr): Ditto.
28618 (chkp_build_addr_expr): Add check for hard reg case.
28619 (chkp_parse_array_and_component_ref): Ditto.
28620 (chkp_find_bounds_1): Ditto.
28621 (chkp_process_stmt): Don't generate bounds store for
28624 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28626 * predict.c (maybe_hot_bb_p): Do not check profile status.
28627 (maybe_hot_edge_p): Likewise.
28628 (probably_never_executed): Check for zero counts even if profile
28630 (unlikely_executed_edge_p): New function.
28631 (unlikely_executed_stmt_p): New function.
28632 (unlikely_executed_bb_p): New function.
28633 (set_even_probabilities): Use unlikely predicates.
28634 (combine_predictions_for_bb): Likewise.
28635 (predict_paths_for_bb): Likewise.
28636 (predict_paths_leading_to_edge): Likewise.
28637 (determine_unlikely_bbs): New function.
28638 (estimate_bb_frequencies): Use it.
28639 (compute_function_frequency): Use zero counts even if profile is
28641 * profile-count.h: Fix typo.
28643 2017-08-08 Julia Koval <julia.koval@intel.com>
28645 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28646 _mm512_mask_cvtsepi16_storeu_epi8,
28647 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28648 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28649 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28650 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28651 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28652 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28653 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28654 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28655 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28656 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28657 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28658 __builtin_ia32_pmovuswb256mem_mask,
28659 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28660 __builtin_ia32_pmovwb512mem_mask): New builtins.
28662 2017-08-08 Julia Koval <julia.koval@intel.com>
28664 PR target/73350,80862
28665 * config/i386/subst.md (round): Fix round pattern.
28666 * config/i386/i386.c (ix86_erase_embedded_rounding):
28667 Fix erasing rounding for the fixed pattern.
28669 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28671 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28673 2017-06-08 Martin Liska <mliska@suse.cz>
28675 PR gcov-profile/80911
28676 * gcov.c (block_info::block_info): New constructor.
28678 2017-06-07 Carl Love <cel@us.ibm.com>
28680 * config/rs6000/rs6000-c: The return type of the following
28681 built-in functions was implemented as int not long long. Fix sign
28682 of return value for the unsigned version of vec_mulo and vec_mule.
28683 vector unsigned long long vec_bperm (vector unsigned long long,
28684 vector unsigned char)
28685 vector signed long long vec_mule (vector signed int,
28687 vector unsigned long long vec_mule (vector unsigned int,
28688 vector unsigned int)
28689 vector signed long long vec_mulo (vector signed int,
28691 vector unsigned long long vec_mulo (vector unsigned int,
28692 vector unsigned int)
28693 * doc/extend.texi: Fix the documentation for the built-in
28696 2017-06-07 Carl Love <cel@us.ibm.com>
28699 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28702 2017-06-07 Carl Love <cel@us.ibm.com>
28704 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28705 support, Generate doublehv for signed int/float for BE case only.
28707 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28709 * doc/invoke.texi (mcx16): Rewrite.
28711 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28713 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28714 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28715 *mov<mode>_softfloat, and an anonymous splitter): Use
28716 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28718 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28720 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28722 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28723 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28724 (rs6000_debug_reg_global): Adjust.
28725 (rs6000_init_hard_regno_mode_ok): Adjust.
28726 (rs6000_dbx_register_number): Adjust.
28727 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28728 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28729 Remove SPE_ACC and SPEFSCR.
28730 (REG_ALLOC_ORDER): Ditto.
28731 (FRAME_POINTER_REGNUM): Change to 111.
28732 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28733 (REG_CLASS_NAMES): Ditto.
28734 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28735 (REGISTER_NAMES): Ditto.
28736 (ADDITIONAL_REG_NAMES): Ditto.
28737 (rs6000_reg_names): Ditto.
28738 * config/rs6000/rs6000.md: Renumber some register number
28741 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28743 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28745 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28746 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28748 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28749 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28750 Delete the SPE high registers.
28751 (REG_ALLOC_ORDER): Ditto.
28752 (enum reg_class): Remove SPE_HIGH_REGS.
28753 (REG_CLASS_NAMES): Ditto.
28754 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28755 (REGISTER_NAMES): Ditto.
28756 (rs6000_reg_names): Ditto.
28757 * doc/tm.texi.in: Remove SPE as example.
28758 * doc/tm.texi: Regenerate.
28760 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28762 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28763 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28764 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28765 * config/rs6000/rs6000.md (type): Remove "brinc".
28767 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28769 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28770 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28771 * config/rs6000/linuxspe.h: Delete file.
28772 * config/rs6000/rs6000.md: Don't include spe.md.
28773 * config/rs6000/spe.h: Delete file.
28774 * config/rs6000/spe.md: Delete file.
28775 * config/rs6000/t-rs6000: Remove spe.md.
28777 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28779 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28780 (reg_or_none500mem_operand): Delete.
28781 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28782 instead of reg_or_none500mem_operand.
28784 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28786 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28787 handling of SPE flags.
28788 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28790 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28792 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28794 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28795 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28796 paired_divv2sf3): Similar.
28797 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28798 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
28799 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
28801 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
28802 Rename the paired_* instruction patterns.
28803 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
28805 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
28806 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
28807 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
28808 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
28809 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
28810 PAIRED_VECTOR_MODE.
28811 (struct machine_function): Delete field spe_insn_chain_scanned_p.
28812 (spe_func_has_64bit_regs_p): Delete.
28813 (spe_expand_predicate_builtin): Delete.
28814 (spe_expand_evsel_builtin): Delete.
28815 (TARGET_DWARF_REGISTER_SPAN): Do not define.
28816 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
28817 (invalid_e500_subreg): Delete.
28818 (rs6000_legitimize_address): Always force_reg op2 as well, for
28819 paired single memory accesses.
28820 (rs6000_member_type_forces_blk): Delete.
28821 (rs6000_spe_function_arg): Delete.
28822 (rs6000_expand_unop_builtin): Delete SPE handling.
28823 (rs6000_expand_binop_builtin): Ditto.
28824 (spe_expand_stv_builtin): Delete.
28825 (bdesc_2arg_spe): Delete.
28826 (spe_expand_builtin): Delete.
28827 (spe_expand_predicate_builtin): Delete.
28828 (spe_expand_evsel_builtin): Delete.
28829 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28830 (spe_init_builtins): Delete.
28831 (spe_func_has_64bit_regs_p): Delete.
28832 (savres_routine_name): Delete "info" parameter. Adjust callers.
28833 (rs6000_emit_stack_reset): Ditto.
28834 (rs6000_dwarf_register_span): Delete.
28835 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28836 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28837 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28839 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28841 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28842 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28843 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28844 mulv2sf3, divv2sf3): Delete expanders.
28846 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28848 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28850 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28852 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
28853 * config/rs6000/rs6000.c: Ditto.
28855 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28857 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
28858 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
28859 comparison_operator.
28861 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28863 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28864 * config/rs6000/rs6000.opt: Ditto.
28865 * config/rs6000/t-rtems: Ditto.
28867 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28869 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
28870 TARGET_E500_SINGLE by 0, simplify.
28871 * config/rs6000/rs6000.c: Ditto.
28872 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
28873 (spe_build_register_parallel): Delete.
28874 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
28875 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
28876 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
28877 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
28878 (E500_CONVERT): Delete.
28879 * config/rs6000/spe.md: Remove many patterns and all define_constants.
28881 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28883 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28884 * config/rs6000/dfp.md: Ditto.
28885 (negdd2, *negdd2_fpr): Merge.
28886 (absdd2, *absdd2_fpr): Merge.
28887 (negtd2, *negtd2_fpr): Merge.
28888 (abstd2, *abstd2_fpr): Merge.
28889 * config/rs6000/e500.h: Delete file.
28890 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28891 TARGET_FPRS by 1 and simplify.
28892 * config/rs6000/rs6000-c.c: Ditto.
28893 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28894 TARGET_DF_SPE by 0.
28895 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28897 * config/rs6000/rs6000.md: Ditto.
28898 (floatdidf2, *floatdidf2_fpr): Merge.
28899 (move_from_CR_gt_bit): Delete.
28900 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28901 (E500_CR_IOR_COMPARE): Delete.
28902 (All patterns that require !TARGET_FPRS): Delete.
28903 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28905 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28907 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28909 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28911 * graphds.c (add_edge): Intitialize edge's attached data.
28912 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28913 pointer parameter. Call pointed function on each edge during
28914 graph traversing. Skip traversing the edge when the function
28916 (graphds_dfs, graphds_scc): Ditto.
28917 (for_each_edge): New parameter. Pass the new parameter to callback
28919 * graphds.h (skip_edge_callback): New function pointer type.
28920 (graphds_dfs, graphds_scc): New function pointer parameter.
28921 (graphds_edge_callback, for_each_edge): New parameter.
28923 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28925 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28926 out code checking if runtime alias check is possible to below ...
28927 Call the new function.
28928 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28929 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28931 2017-06-07 Marek Polacek <polacek@redhat.com>
28934 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28935 TYPE_OVERFLOW_WRAPS check.
28937 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28939 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28940 if versioning is required.
28941 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28942 peeling with the check for versioning.
28944 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28946 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28947 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28948 Set true to new parameter if new ssa variable is defined.
28949 (vect_gen_vector_loop_niters): Refactor. Set range information
28950 for the new vector loop bound variable.
28951 (vect_do_peeling): Ditto.
28953 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28955 * tree-affine.c (ssa.h): Include header file.
28956 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28957 has wrapping overflow behavior.
28959 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28961 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28963 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28965 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28966 (tree_to_aff_combination): ... here.
28968 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28970 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28971 reg_pressure model function.
28972 (ivopts_global_cost_for_size): Delete.
28973 (determine_set_costs, iv_ca_recount_cost): Call new model function
28974 ivopts_estimate_reg_pressure.
28976 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28978 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28979 expensive than udiv. Remove floating point cases from mod.
28981 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28983 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28984 Increase idiv cost.
28986 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28988 * config/aarch64/aarch64.md
28989 (copysignsf3): Fix mask generation.
28991 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28993 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28995 (class dump_manager): Add register_dumps method.
28996 * dumpfile.c: Include langhooks.h.
28997 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28998 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28999 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
29000 (dump_manager::dump_register): Start with 512 entries instead of 32.
29001 (dump_manager::register_dumps): New method.
29002 * toplev.c (general_init): Instead of invoking register_dumps
29003 langhook, invoke register_dumps method on the dump manager.
29004 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
29007 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29009 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29010 Say that # requires an associated define_split to exist, and that
29011 the define_split must be suitable for use after register allocation.
29013 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29015 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29016 (compute_outgoing_frequencies): Also initialize zero counts.
29017 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29018 around loops; preserve more of profile when nothing changes.
29020 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29022 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29024 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29025 * config/arm/arm-cpu-cdata.h: Regenerate.
29026 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29027 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29028 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29029 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29030 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29032 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29034 * config/arm/t-rmprofile: Likewise.
29035 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29037 2017-06-06 David S. Miller <davem@davemloft.net>
29040 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29041 blockage if function uses alloca.
29043 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29045 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29046 New "uid" fields to hold pretty-print IDs of group and ref.
29047 Memory references are now identified as <group_id>:<ref_id>
29048 instead of using [random] addresses.
29049 (dump_mem_details): Simplify, no functional change.
29050 (dump_mem_ref): Simplify and make output more concise.
29051 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29052 (find_or_create_group): Initialize group uid.
29053 (record_ref): Initialize ref uid. Improve debug output.
29054 (prune_group_by_reuse, should_issue_prefetch_p,)
29055 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29056 (mark_nontemporal_store, determine_loop_nest_reuse):
29057 Improve debug output.
29059 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29061 * dbgcnt.def (prefetch): New debug counter.
29062 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29063 (schedule_prefetches): Stop issueing prefetches if debug counter
29066 2017-06-06 Tom de Vries <tom@codesourcery.com>
29068 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29069 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29071 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29073 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29074 Use aarch64_reg_or_zero predicate for operand 4.
29075 (aarch64_compare_and_swap<mode> define_insn_and_split):
29076 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29077 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29079 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29081 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29082 (arm_compute_save_core_reg_mask): This.
29083 (thumb1_compute_save_reg_mask): Rename into ...
29084 (thumb1_compute_save_core_reg_mask): This.
29085 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29086 (arm_compute_frame_layout): Likewise.
29088 2017-06-06 Richard Biener <rguenther@suse.de>
29090 PR tree-optimization/80974
29091 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29092 keep or clear leaders SSA info.
29094 2017-06-06 Tom de Vries <tom@codesourcery.com>
29096 * config/nvptx/nvptx.c (split_mode_p): New function.
29097 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29099 2017-06-06 Tom de Vries <tom@codesourcery.com>
29101 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29103 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29106 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29109 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29111 * shrink-wrap.c (handle_simple_exit): Update profile.
29112 (try_shrink_wrapping): Upate profile.
29114 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29116 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29117 (tree_guess_outgoing_edge_probabilities): New.
29118 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29119 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29121 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29123 * ipa-split.c (split_function): Initialize return bb profile.
29125 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29127 * profile.c (compute_branch_probabilities): Also initialize
29128 EXIT_BLOCK profile.
29130 2017-06-06 Richard Biener <rguenther@suse.de>
29132 PR tree-optimization/80928
29133 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29134 (vect_analyze_loop_operations): Properly guard analysis for
29136 (vect_transform_loop): Likewise.
29137 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29138 (vect_model_induction_cost): Do not cost for pure SLP.
29139 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29140 of induction in inner loop vectorization.
29141 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29142 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29143 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29145 (vect_analyze_slp_cost_1): Cost induction.
29146 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29147 (vect_get_slp_vect_defs): Likewise.
29148 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29149 (vect_transform_stmt): Handle SLP reductions.
29150 * tree-vectorizer.h (vectorizable_induction): Adjust.
29152 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29154 * config/rs6000/rs6000.c (make_resolver_func): Update
29155 init_lowered_empty_function call.
29157 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29159 * doc/invoke.texi: Document the -fprofile-abs-path option.
29160 * common.opt (fprofile-abs-path): New option.
29161 * gcov-io.h (gcov_write_filename): Declare.
29162 * gcov-io.c (gcov_write_filename): New function.
29163 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29164 * profile.c (output_location): Likewise.
29166 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29168 * shring-wrap.c: Revert accidental commit.
29170 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29172 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29174 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29176 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29178 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29179 profile in callgraph edge.
29180 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29181 (apply_scale): Likewise.
29182 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29183 Also scale profile when inlining function with zero profile.
29184 (initialize_cfun): Update exit block profile even when it is zero.
29185 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29186 when profile is read.
29188 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29190 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29191 (CLONE_*): New constants to define the processors we can generate
29192 code for with the target_clone attribute.
29193 (rs6000_clone_map): New array to identify which clone processors
29194 the current program is running on.
29195 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29196 target_clone attribute.
29197 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29198 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29199 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29200 (cpu_expand_builtin): Add support for target_clone attribute.
29201 (rs6000_valid_attribute_p): Allow "default" attribute.
29202 (get_decl_name): New debug function to simplify printing the
29203 current function name in debugging statements.
29204 (rs6000_clone_priority): New functions to support the target_clone
29205 attribute, and be able to generate code to switch between ISA 2.05
29206 through ISA 3.0 (power6 through power9).
29207 (rs6000_compare_version_priority): Likewise.
29208 (rs6000_get_function_versions_dispatcher): Likewise.
29209 (make_resolver_func): Likewise.
29210 (add_condition_to_bb): Likewise.
29211 (dispatch_function_versions): Likewise.
29212 (rs6000_generate_version_dispatcher_body): Likewise.
29213 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29214 (fusion_gpr_load_p): Fix a spacing issue.
29215 * doc/extend.texi (Common Function Attributes): Document that the
29216 PowerPC supports the target_clone attribute.
29218 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29220 * config/arm/arm.h: explain F symbol found in description of ARM
29221 register allocation in its legend.
29223 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29225 * config/mips/frame-header-opt.c: Include profile-count.h.
29226 * config/riscv/riscv.c: Include profile-count.h
29228 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29230 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29232 (sm_set_flag_if_changed): Add bbs field.
29233 (execute_sm_if_changed_flag_set): Pass BBS.
29234 (execute_sm): Update.
29236 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29238 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29241 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29243 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29244 (peephole2): New peephole2 to emit the above.
29245 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29247 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29249 * config/aarch64/aarch64.c (define_peephole2 above
29250 *sub_<shift>_<mode>): New peephole.
29252 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29254 * config/i386/i386.c (make_resolver_func): Update.
29255 * Makefile.in: Add profile-count.h and profile-count.o
29256 * auto-profile.c (afdo_indirect_call): Update to new API.
29257 (afdo_set_bb_count): Update.
29258 (afdo_propagate_edge): Update.
29259 (afdo_propagate_circuit): Update.
29260 (afdo_calculate_branch_prob): Update.
29261 (afdo_annotate_cfg): Update.
29262 * basic-block.h: Include profile-count.h
29263 (struct edge_def): Turn count to profile_count.
29264 (struct basic_block_def): Likewie.
29265 (REG_BR_PROB_BASE): Move to profile-count.h
29266 (RDIV): Move to profile-count.h
29267 * bb-reorder.c (max_entry_count): Turn to profile_count.
29268 (find_traces): Update.
29269 (rotate_loop):Update.
29270 (connect_traces):Update.
29271 (sanitize_hot_paths):Update.
29272 * bt-load.c (migrate_btr_defs): Update.
29273 * cfg.c (RDIV): Remove.
29274 (init_flow): Use alloc_block.
29275 (alloc_block): Uninitialize count.
29276 (unchecked_make_edge): Uninitialize count.
29277 (check_bb_profile): Update.
29278 (dump_edge_info): Update.
29279 (dump_bb_info): Update.
29280 (update_bb_profile_for_threading): Update.
29281 (scale_bbs_frequencies_int): Update.
29282 (scale_bbs_frequencies_gcov_type): Update.
29283 (scale_bbs_frequencies_profile_count): New.
29284 * cfg.h (update_bb_profile_for_threading): Update.
29285 (scale_bbs_frequencies_profile_count): Declare.
29286 * cfgbuild.c (compute_outgoing_frequencies): Update.
29287 (find_many_sub_basic_blocks): Update.
29288 * cfgcleanup.c (try_forward_edges): Update.
29289 (try_crossjump_to_edge): Update.
29290 * cfgexpand.c (expand_gimple_tailcall): Update.
29291 (construct_exit_block): Update.
29292 * cfghooks.c (verify_flow_info): Update.
29293 (dump_bb_for_graph): Update.
29294 (split_edge): Update.
29295 (make_forwarder_block): Update.
29296 (duplicate_block): Update.
29297 (account_profile_record): Update.
29298 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29299 (get_estimated_loop_iterations): Update.
29300 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29301 (single_likely_exit): Update.
29302 * cfgloopmanip.c (scale_loop_profile): Update.
29304 (set_zero_probability): Update.
29305 (lv_adjust_loop_entry_edge): Update.
29306 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29307 (purge_dead_edges): Update.
29308 (rtl_account_profile_record): Update.
29309 * cgraph.c (cgraph_node::create): Uninitialize count.
29310 (symbol_table::create_edge): Uninitialize count.
29311 (cgraph_update_edges_for_call_stmt_node): Update.
29312 (cgraph_edge::dump_edge_flags): Update.
29313 (cgraph_node::dump): Update.
29314 (cgraph_edge::maybe_hot_p): Update.
29315 * cgraph.h: Include profile-count.h
29316 (create_clone), create_edge, create_indirect_edge): Update.
29317 (cgraph_node): Turn count to profile_count.
29318 (cgraph_edge0: Likewise.
29319 (make_speculative, clone): Update.
29320 (create_edge): Update.
29321 (init_lowered_empty_function): Update.
29322 * cgraphclones.c (cgraph_edge::clone): Update.
29323 (duplicate_thunk_for_node): Update.
29324 (cgraph_node::create_clone): Update.
29325 * cgraphunit.c (cgraph_node::analyze): Update.
29326 (cgraph_node::expand_thunk): Update.
29327 * final.c (dump_basic_block_info): Update.
29328 * gimple-streamer-in.c (input_bb): Update.
29329 * gimple-streamer-out.c (output_bb): Update.
29330 * graphite.c (print_global_statistics): Update.
29331 (print_graphite_scop_statistics): Update.
29332 * hsa-brig.c: Include basic-block.h.
29333 * hsa-dump.c: Include basic-block.h.
29334 * hsa-gen.c (T sum_slice): Update.
29335 (convert_switch_statements):Update.
29336 * hsa-regalloc.c: Include basic-block.h.
29337 * ipa-chkp.c (chkp_produce_thunks): Update.
29338 * ipa-cp.c (struct caller_statistics): Update.
29339 (init_caller_stats): Update.
29340 (gather_caller_stats): Update.
29341 (ipcp_cloning_candidate_p): Update.
29342 (good_cloning_opportunity_p): Update.
29343 (get_info_about_necessary_edges): Update.
29344 (dump_profile_updates): Update.
29345 (update_profiling_info): Update.
29346 (update_specialized_profile): Update.
29347 (perhaps_add_new_callers): Update.
29348 (decide_about_value): Update.
29349 (ipa_cp_c_finalize): Update.
29350 * ipa-devirt.c (struct odr_type_warn_count): Update.
29351 (struct decl_warn_count): Update.
29352 (struct final_warning_record): Update.
29353 (possible_polymorphic_call_targets): Update.
29354 (ipa_devirt): Update.
29355 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29356 * ipa-icf.c (sem_function::merge): Update.
29357 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29358 * ipa-inline.c (compute_uninlined_call_time): Update.
29359 (compute_inlined_call_time): Update.
29360 (want_inline_small_function_p): Update.
29361 (want_inline_self_recursive_call_p): Update.
29362 (edge_badness): Update.
29363 (lookup_recursive_calls): Update.
29364 (recursive_inlining): Update.
29365 (inline_small_functions): Update.
29366 (dump_overall_stats): Update.
29367 (dump_inline_stats): Update.
29368 * ipa-profile.c (ipa_profile_generate_summary): Update.
29369 (ipa_propagate_frequency): Update.
29370 (ipa_profile): Update.
29371 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29372 * ipa-utils.c (ipa_merge_profiles): Update.
29373 * loop-doloop.c (doloop_modify): Update.
29374 * loop-unroll.c (report_unroll): Update.
29375 (unroll_loop_runtime_iterations): Update.
29376 * lto-cgraph.c (lto_output_edge): Update.
29377 (lto_output_node): Update.
29378 (input_node): Update.
29379 (input_edge): Update.
29380 (merge_profile_summaries): Update.
29381 * lto-streamer-in.c (input_cfg): Update.
29382 * lto-streamer-out.c (output_cfg): Update.
29383 * mcf.c (create_fixup_graph): Update.
29384 (adjust_cfg_counts): Update.
29385 (sum_edge_counts): Update.
29386 * modulo-sched.c (sms_schedule): Update.
29387 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29388 * predict.c (maybe_hot_count_p): Update.
29389 (probably_never_executed): Update.
29390 (dump_prediction): Update.
29391 (combine_predictions_for_bb): Update.
29392 (propagate_freq): Update.
29393 (handle_missing_profiles): Update.
29394 (counts_to_freqs): Update.
29395 (rebuild_frequencies): Update.
29396 (force_edge_cold): Update.
29397 * predict.h: Include profile-count.h
29398 (maybe_hot_count_p, counts_to_freqs): UPdate.
29399 * print-rtl-function.c: Do not include cfg.h
29400 * print-rtl.c: Include basic-block.h
29401 * profile-count.c: New file.
29402 * profile-count.h: New file.
29403 * profile.c (is_edge_inconsistent): Update.
29404 (correct_negative_edge_counts): Update.
29405 (is_inconsistent): Update.
29406 (set_bb_counts): Update.
29407 (read_profile_edge_counts): Update.
29408 (compute_frequency_overlap): Update.
29409 (compute_branch_probabilities): Update; Initialize and deinitialize
29411 (branch_prob): Update.
29412 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29413 (edge_gcov_count): New.
29414 (bb_gcov_count): New.
29415 * shrink-wrap.c (try_shrink_wrapping): Update.
29416 * tracer.c (better_p): Update.
29417 * trans-mem.c (expand_transaction): Update.
29418 (ipa_tm_insert_irr_call): Update.
29419 (ipa_tm_insert_gettmclone_call): Update.
29420 * tree-call-cdce.c: Update.
29421 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29422 (gimple_duplicate_sese_tail): Update.
29423 (gimple_account_profile_record): Update.
29424 (execute_fixup_cfg): Update.
29425 * tree-inline.c (copy_bb): Update.
29426 (copy_edges_for_bb): Update.
29427 (initialize_cfun): Update.
29428 (freqs_to_counts): Update.
29429 (copy_cfg_body): Update.
29430 (expand_call_inline): Update.
29431 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29432 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29433 (try_unroll_loop_completely): Update.
29434 (try_peel_loop): Update.
29435 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29436 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29437 * tree-ssa-loop-split.c (connect_loops): Update.
29438 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29439 * tree-ssa-reassoc.c (branch_fixup): Update.
29440 * tree-ssa-tail-merge.c (replace_block_by): Update.
29441 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29442 (compute_path_counts): Update.
29443 (update_profile): Update.
29444 (recompute_probabilities): Update.
29445 (update_joiner_offpath_counts): Update.
29446 (estimated_freqs_path): Update.
29447 (freqs_to_counts_path): Update.
29448 (clear_counts_path): Update.
29449 (ssa_fix_duplicate_block_edges): Update.
29450 (duplicate_thread_path): Update.
29451 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29452 (struct switch_conv_info): Update.
29453 * tree-tailcall.c (decrease_profile): Update.
29454 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29455 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29456 * value-prof.c (check_counter): Update.
29457 (gimple_divmod_fixed_value): Update.
29458 (gimple_mod_pow2): Update.
29459 (gimple_mod_subtract): Update.
29460 (gimple_ic_transform): Update.
29461 (gimple_stringop_fixed_value): Update.
29462 * value-prof.h (gimple_ic): Update.
29464 2017-06-02 Carl Love <cel@us.ibm.com>
29466 * config/rs6000/rs6000-c: Add support for built-in functions
29467 vector double vec_doublee (vector signed int);
29468 vector double vec_doublee (vector unsigned int);
29469 vector double vec_doublee (vector float);
29470 vector double vec_doubleh (vector signed int);
29471 vector double vec_doubleh (vector unsigned int);
29472 vector double vec_doubleh (vector float);
29473 vector double vec_doublel (vector signed int);
29474 vector double vec_doublel (vector unsigned int);
29475 vector double vec_doublel (vector float);
29476 vector double vec_doubleo (vector signed int);
29477 vector double vec_doubleo (vector unsigned int);
29478 vector double vec_doubleo (vector float);.
29479 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29480 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29482 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29483 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29484 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29486 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29487 vec_doublel, vec_doubleh.
29488 * doc/extend.texi: Update the built-in documentation file for the
29489 new built-in functions.
29491 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29494 * ipa-inline-analysis.c (free_growth_caches): Set
29495 edge_removal_hook_holder to NULL after removing it.
29497 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29499 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29500 comparision with zero.
29502 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29503 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29504 for early expansion of vec_min and vec_max builtins.
29505 (builtin_function_type): Add min/max unsigned variants to those
29506 identified as having unsigned arguments.
29508 2017-06-02 Olivier Hainque <hainque@adacore.com>
29510 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29512 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29514 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29515 Use VALL_F16 iterator rather than VALL.
29517 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29519 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29520 Emit CBNZ inside loop when doing a strong exchange and comparing
29521 against zero. Generate the CC flags after the loop.
29523 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29525 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29526 (dl_section_ref): New.
29527 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29528 On AIX, append an expression to subtract the size of the
29529 section length to dl_section_ref.
29531 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29533 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29534 for early expansion of vector absolute builtins.
29536 2017-06-02 Richard Biener <rguenther@suse.de>
29538 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29539 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29541 2017-06-02 Richard Biener <rguenther@suse.de>
29543 PR tree-optimization/80948
29544 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29545 stmt order as well.
29547 2017-06-02 Richard Biener <rguenther@suse.de>
29549 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29551 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29552 for inductions as relevant.
29554 2017-06-02 Richard Biener <rguenther@suse.de>
29556 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29557 (vectorizable_induction): ... this. Remove dead code.
29559 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29561 * builtins. (expand_builtin_alloca): Remove second parameter and
29562 infer its value from the first parameter instead.
29563 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29565 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29567 PR rtl-optimization/80903
29568 * loop-doloop.c (add_test): Unshare sequence.
29570 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29572 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29574 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29576 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29578 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29579 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29580 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29581 (xlogue_layout::get_stub_rtx): Make static.
29582 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29583 (xlogue_layout::compute_stub_managed_regs): Rename to...
29584 (xlogue_layout::count_stub_managed_regs): ...this.
29585 (xlogue_layout::is_stub_managed_reg): New function.
29586 (xlogue_layout::m_stub_names): Rename to...
29587 (xlogue_layout::s_stub_names): ...this, make static.
29588 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29589 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29590 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29591 xlogue_layout::s_stub_names): Instantiate statics.
29592 (stub_managed_regs): Remove.
29593 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29594 (disable_call_ms2sysv_xlogues): Rename to...
29595 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29596 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29598 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29599 change after reload_completed.
29600 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29602 (ix86_expand_prologue): Likewise.
29603 (ix86_expand_epilogue): Likewise.
29604 (ix86_expand_split_stack_prologue): Likewise.
29605 (ix86_compute_frame_layout): Remove frame parameter ...
29606 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29607 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29609 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29610 (ix86_frame): Move from here ...
29611 * config/i386/i386.h (ix86_frame): ... to here.
29612 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29613 complete ix86_frame data structure instead. Remove some_ld_name.
29615 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29617 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29618 symbols that hold a DECL_VALUE_EXPR.
29620 2017-06-01 Martin Jambor <mjambor@suse.cz>
29622 PR tree-optimization/80898
29623 * tree-sra.c (process_subtree_disqualification): Removed.
29624 (disqualify_candidate): Do not acll
29625 process_subtree_disqualification.
29626 (subtree_mark_written_and_enqueue): New function.
29627 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29628 RHS has been disqualified and re-queue LHS if necessary. Apart
29629 from that, ignore disqualified RHS.
29631 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29633 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29634 address fetch for z10 or later.
29636 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29638 * config/arc/arc.md (tst_movb): Add guard when splitting.
29640 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29642 * config/arc/arc.c (arc_can_eliminate): Test against
29643 arc_frame_pointer_needed.
29645 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29647 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29648 to prevent store reordering.
29649 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29650 (type): Add block type.
29651 (stack_tie): Define special instruction to be used in
29654 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29656 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29657 constraint. It is not valid for the pattern.
29658 (noncommutative_binary_comparison): Likewise.
29660 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29662 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29665 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29667 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29668 be used by the reg-alloc.
29670 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29672 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29673 reg-alloc when having mul64 or mul32x16 instructions.
29674 (mulsidi3): Likewise.
29675 (umulsidi3): Likewise.
29676 (mulsi32x16): New pattern.
29677 (mulsi64): Likewise.
29678 (mulsidi64): Likewise.
29679 (umulsidi64): Likewise.
29680 (MUL32x16_REG): Define.
29681 (mul64_600): Use MUL32x16_REG.
29682 (mac64_600): Likewise.
29683 (umul64_600): Likewise.
29684 (umac64_600): Likewise.
29686 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29688 * config/arc/arc.md (mulsi3_700): Make it commutative.
29690 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29692 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29694 (*sign_extendsidi2_insn): Likewise for movstosw.
29696 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29698 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29699 the type of the input discriminant value. Convert the
29700 discriminant value of signedness vary.
29702 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29704 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29705 Add to -Wall section.
29707 2017-06-01 Richard Biener <rguenther@suse.de>
29709 PR middle-end/66313
29710 * fold-const.c (fold_plusminus_mult_expr): If the factored
29711 factor may be zero use a wrapping type for the inner operation.
29712 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29713 and handle moved defs.
29714 (process_assignment): Properly guard the unary op case. Return a
29715 tri-state indicating that moving the stmt before the call may allow
29716 to continue. Pass through to_move.
29717 (find_tail_calls): Handle moving unrelated defs before
29720 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29723 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29724 splitter result in the canonical way.
29726 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29728 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29729 also for 32bit target. Update insn attributes.
29730 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29732 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29734 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29735 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29736 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29737 (_mm_maskz_min_ss): New intrinsics.
29739 2017-05-31 Martin Liska <mliska@suse.cz>
29741 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29742 Change comment style to one we normally use.
29743 (vectorizable_reduction): Likewise.
29744 (vectorizable_induction): Likewise.
29745 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29746 (vectorizable_call): Likewise.
29747 (vectorizable_simd_clone_call): Likewise.
29748 (vectorizable_conversion): Likewise.
29749 (vectorizable_assignment): Likewise.
29750 (vectorizable_shift): Likewise.
29751 (vectorizable_operation): Likewise.
29752 (vectorizable_store): Likewise.
29753 (vectorizable_load): Likewise.
29754 * tree-vectorizer.h: Likewise.
29756 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29758 * passes.c (emergency_dump_function): New.
29759 * tree-pass.h (emergency_dump_function): Declare.
29760 * plugin.c (plugins_internal_error_function): Remove.
29761 * plugin.h (plugins_internal_error_function): Remove declaration.
29762 * toplev.c (internal_error_function): New static function. Use it...
29763 (general_init): ...here.
29765 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29767 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29768 (arc_rtx_costs): Add costs for new patterns.
29769 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29770 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29772 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29774 * tree-ssa-strlen.c (get_next_strinfo): New function.
29775 (get_stridx_plus_constant): Use it.
29776 (zero_length_string): Likewise.
29777 (adjust_related_strinfos): Likewise.
29778 (adjust_last_stmt): Likewise.
29780 2017-05-31 Richard Biener <rguenther@suse.de>
29783 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29784 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29786 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29788 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29789 loop_vinfo argument and use of dependence distance vectors.
29790 Check instead whether the two references differ only in their
29791 initial value and assume that they have the same alignment if the
29792 difference is a multiple of the vector alignment.
29793 (vect_analyze_data_refs_alignment): Update call accordingly.
29795 2017-05-31 Martin Liska <mliska@suse.cz>
29798 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
29800 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29802 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
29803 (create_intersect_range_checks): Move from ...
29804 * tree-data-ref.c (create_intersect_range_checks_index)
29805 (create_intersect_range_checks): ... to here.
29806 (create_runtime_alias_checks): New function factored from ...
29807 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
29808 here. Call above function.
29809 * tree-data-ref.h (create_runtime_alias_checks): New function.
29811 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29813 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
29814 segment length for dr_b and compute it in wide_int.
29816 2017-05-31 Richard Biener <rguenther@suse.de>
29818 PR tree-optimization/80906
29819 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
29820 and pass through iv_map.
29821 (copy_bb_and_scalar_dependences): Adjust.
29822 (translate_pending_phi_nodes): Likewise.
29823 (copy_loop_close_phi_args): Handle code-generating IVs instead
29826 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29828 * diagnostic-color.c (color_dict): Add "type-diff".
29829 (parse_gcc_colors): Update comment.
29830 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29831 -fdiagnostics-show-template-tree and -fno-elide-type.
29832 (GCC_COLORS): Add type-diff to example.
29834 (-fdiagnostics-show-template-tree): New.
29835 (-fno-elide-type): New.
29836 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29837 the pp_format_decoder callback. Call any m_format_postprocessor's
29839 (pretty_printer::pretty_printer): Initialize
29840 m_format_postprocessor.
29841 (pretty_printer::~pretty_printer): Delete any
29842 m_format_postprocessor.
29843 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29844 (class format_postprocessor): New class.
29845 (struct pretty_printer::format_decoder): Document the new parameters.
29846 (struct pretty_printer::m_format_postprocessor): New field.
29847 * tree-diagnostic.c (default_tree_printer): Update for new
29848 bool and const char ** params.
29849 * tree-diagnostic.h (default_tree_printer): Likewise.
29851 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
29853 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
29854 (lwa_operand): Delete rs6000_gen_cell_microcode test.
29855 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29856 rs6000_gen_cell_microcode code.
29857 (rs6000_final_prescan_insn): Delete.
29858 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
29859 "warn-cell-microcode" entries.
29860 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29861 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29862 throughout. Change cc_reg_not_micro_cr0_operand to
29863 cc_reg_not_cr0_operand throughout.
29864 (*extendhi<mode>2_noload): Delete.
29865 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
29866 (mwarn-cell-microcode): Delete.
29867 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
29868 -mgen-cell-microcode and -mwarn-cell-microcode.
29870 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
29873 * config/i386/constraints.md (Yd): New constraint.
29875 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
29876 and (?Yd, r) alternatives. Update insn attributes.
29877 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
29878 and (?*Yd, r) alternatives. Update insn attributes.
29879 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
29881 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
29883 * gimplify.c (gimplify_modify_expr): Don't create a
29884 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29887 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29889 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29891 2017-05-30 Richard Biener <rguenther@suse.de>
29893 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29894 and reduc_def fields.
29895 (STMT_VINFO_REDUC_TYPE): New define.
29896 (STMT_VINFO_REDUC_DEF): Likewise.
29897 (vect_force_simple_reduction): Adjust prototype.
29898 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29899 (vect_is_simple_reduction): Remove check_reduction argument.
29900 (vect_force_simple_reduction): Adjust and set
29901 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29902 (vectorizable_reduction): Do not re-do reduction analysis
29903 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29904 * tree-parloops.c (gather_scalar_reductions): Adjust.
29906 2017-05-30 Richard Biener <rguenther@suse.de>
29908 PR middle-end/80901
29909 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29912 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29914 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29915 Introduce unknown_misalignment parameter and remove vf.
29916 (vect_peeling_hash_get_lowest_cost):
29917 Pass unknown_misalignment parameter.
29918 (vect_enhance_data_refs_alignment):
29919 Fix unsupportable data ref treatment.
29921 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29923 * tree-vect-data-refs.c (vect_get_data_access_cost):
29924 Workaround for SLP handling.
29925 (vect_enhance_data_refs_alignment):
29926 Compute costs for doing no peeling at all, compare to the best
29927 peeling costs so far and avoid peeling if cheaper.
29929 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29931 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29932 Return peeling info and set costs to zero for unlimited cost
29934 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29935 with unknown misalignment. Compute and costs for unknown
29936 misalignment, compare them to the costs for known misalignment
29937 and choose the cheapest for peeling.
29939 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29941 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29942 (vect_get_peeling_costs_all_drs): Create function.
29943 (vect_peeling_hash_get_lowest_cost):
29944 Use vect_get_peeling_costs_all_drs.
29945 (vect_peeling_supportable): Create function.
29946 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29948 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29950 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29951 DR_HAS_NEGATIVE_STEP.
29952 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29953 (vect_enhance_data_refs_alignment): Use.
29954 (vect_duplicate_ssa_name_ptr_info): Use.
29955 * tree-vectorizer.h (dr_misalignment): Use.
29956 (known_alignment_for_access_p): Use.
29958 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29961 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29963 (has_section_name): New function.
29965 2017-05-30 Martin Liska <mliska@suse.cz>
29968 * auto-profile.c (get_function_decl_from_block): Fix
29971 2017-05-30 Richard Biener <rguenther@suse.de>
29973 PR middle-end/80876
29974 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29976 2017-05-30 Martin Liska <mliska@suse.cz>
29978 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29979 * dumpfile.h (struct dump_file_info): Remove ctors.
29981 2017-05-30 Martin Liska <mliska@suse.cz>
29983 * predict.def: Fix GNU coding style.
29985 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29987 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29988 Mark 'to' argument with ATTRIBUTE_UNUSED.
29990 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29992 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29993 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29994 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29997 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29999 * doc/install.texi (Options specification): Restore entry of
30000 --enable-sjlj-exceptions.
30002 2017-05-27 Michael Eager <eager@eagercon.com>
30005 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
30007 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30009 * config/microblaze/microblaze.h
30010 (FIXED_REGISTERS): Update in macro.
30011 (CALL_USED_REGISTERS): Update in macro.
30013 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30015 * doc/install.texi: Add links to macOS binary distributions.
30017 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30021 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30023 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30025 2017-05-26 Martin Liska <mliska@suse.cz>
30027 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30029 2017-05-26 Martin Liska <mliska@suse.cz>
30031 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30032 always leading ';; '.
30033 (dump_bb_info): Likewise.
30034 (brief_dump_cfg): Likewise.
30035 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30036 * dumpfile.c: Remove usage of TDF_VERBOSE.
30037 * dumpfile.h (enum dump_kind): Likewise.
30038 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30039 * print-tree.c (debug_verbose): Remove.
30040 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30041 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30042 * tree-diagnostic.c (default_tree_printer): Replace
30043 TDF_DIAGNOSTIC with TDF_SLIM.
30045 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30047 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30048 in parameter loop, rather than loop_vinfo.
30049 (create_intersect_range_checks): Ditto.
30050 (vect_create_cond_for_alias_checks): Update call to above functions.
30052 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30054 PR tree-optimization/80815
30055 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30056 for merging runtime alias checks. Handle negative DR_STEPs.
30058 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30060 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30062 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30063 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30064 out code pruning runtime alias checks.
30065 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30066 factored out from above.
30067 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30069 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30071 (prune_runtime_alias_test_list): New decalaration.
30073 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30075 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30076 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30077 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30078 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30079 (operator==, comp_dr_with_seg_len_pair): Ditto.
30080 (vect_prune_runtime_alias_test_list): Ditto.
30082 2017-05-26 Martin Liska <mliska@suse.cz>
30085 * params.def: Bound partial-inlining-entry-probability param.
30087 2017-05-26 Marek Polacek <polacek@redhat.com>
30090 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30093 2017-05-26 Richard Biener <rguenther@suse.de>
30095 PR tree-optimization/80842
30096 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30099 2017-05-26 Richard Biener <rguenther@suse.de>
30101 PR tree-optimization/80844
30102 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30104 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30106 * doc/md.texi (Machine Constraints): Update x86 family
30107 machine constraints section to match 'config/i386/constraints.md'.
30109 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30111 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30113 2017-05-25 Nathan Sidwell <nathan@acm.org>
30115 * doc/invoke.texi (--enable-languages): Update documentation.
30117 2017-05-25 Martin Liska <mliska@suse.cz>
30119 * dumpfile.c: Add TDF_FOLDING.
30120 * dumpfile.h (enum dump_kind): Likewise.
30121 * genmatch.c (dt_simplify::gen_1): Use it.
30123 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30125 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30127 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30129 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30130 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30132 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30134 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30135 * match.pd (X == C): Rewrite it here.
30136 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30137 with_certain_nonzero_bits2): New predicates.
30138 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30140 2017-05-24 Nathan Sidwell <nathan@acm.org>
30142 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30145 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30148 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30150 * config/powerpcspe: New port. Files are copied from the rs6000
30151 port, with "rs6000" in filenames replaced by "powerpcspe".
30153 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30155 PR rtl-optimization/80754
30156 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30158 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30160 * config/sparc/sparc.md (length): Return the correct value for -mflat
30161 sibcalls to match output_sibcall.
30163 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30167 * config/rs6000/rs6000.c (struct machine_function): Add new field
30169 (rs6000_get_separate_components): Init that field, use it.
30170 (rs6000_components_for_bb): Use the field.
30172 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30174 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30176 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30178 PR middle-end/80823
30179 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30181 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30184 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30185 address against address_operand predicate.
30186 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30188 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30190 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30191 parameters passed indirectly.
30193 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30195 * config/i386/i386.md (*movdi_internal): Remove SSE4
30196 alternative 18 (?r, *v). Update insn attributes.
30197 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30198 Update insn attributes.
30199 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30200 Update insn attributes.
30201 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30202 alternative 1 (r, v). Remove isa attribute.
30203 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30204 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30205 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30207 2017-05-23 Tom de Vries <tom@codesourcery.com>
30209 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30212 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30214 * cgraphunit.c (symbol_table::process_new_functions): Update.
30215 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30216 (inline_generate_summary): Rename to ...
30217 (ipa_fn_summary_generate): ... this one.
30218 (inline_read_summary): Rename to ...
30219 (ipa_fn_summary_read): ... this one.
30220 (inline_write_summary): Rename to ...
30221 (ipa_fn_summary_write): ... this one.
30222 (inline_free_summary): Rename to ...
30223 (ipa_free_fn_summary): ... this one.
30224 (pass_data_local_fn_summary, pass_local_fn_summary,
30225 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30226 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30227 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30228 make_pass_ipa_fn_summary): New.
30229 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30230 inline_write_summary, inline_free_summary): Remove.
30231 (ipa_free_fn_summary) : New.
30232 * ipa-inline.c (ipa_inline): Update.
30233 (pass_ipa_inline): Do not generate summaries.
30234 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30236 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30237 and add pass_ipa_fn_summary.
30238 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30240 (make_pass_inline_parameters): Remove.
30242 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30244 * omp-low.c (struct omp_context): Remove "default_kind" member.
30247 * omp-offload.c (execute_oacc_device_lower): Remove the
30248 parallelism dimensions function attributes for unparallelized
30249 OpenACC kernels constructs.
30251 2017-05-23 Martin Liska <mliska@suse.cz>
30253 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30255 (cgraph_edge::make_speculative): Likewise.
30256 (cgraph_edge::resolve_speculation): Likewise.
30257 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30258 (cgraph_node::dump): Likewise.
30259 * cgraph.h: Likewise.
30260 * cgraphunit.c (analyze_functions): Likewise.
30261 (symbol_table::compile): Likewise.
30262 * ipa-cp.c (print_all_lattices): Likewise.
30263 (determine_versionability): Likewise.
30264 (initialize_node_lattices): Likewise.
30265 (ipcp_verify_propagated_values): Likewise.
30266 (estimate_local_effects): Likewise.
30267 (update_profiling_info): Likewise.
30268 (create_specialized_node): Likewise.
30269 (perhaps_add_new_callers): Likewise.
30270 (decide_about_value): Likewise.
30271 (decide_whether_version_node): Likewise.
30272 (identify_dead_nodes): Likewise.
30273 (ipcp_store_bits_results): Likewise.
30274 * ipa-devirt.c (dump_targets): Likewise.
30275 (ipa_devirt): Likewise.
30276 * ipa-icf.c (sem_item::dump): Likewise.
30277 (sem_function::equals): Likewise.
30278 (sem_variable::equals): Likewise.
30279 (sem_item_optimizer::read_section): Likewise.
30280 (sem_item_optimizer::execute): Likewise.
30281 (congruence_class::dump): Likewise.
30282 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30283 (dump_inline_summary): Likewise.
30284 (estimate_node_size_and_time): Likewise.
30285 (inline_analyze_function): Likewise.
30286 * ipa-inline-transform.c (inline_call): Likewise.
30287 * ipa-inline.c (report_inline_failed_reason): Likewise.
30288 (want_early_inline_function_p): Likewise.
30289 (edge_badness): Likewise.
30290 (update_edge_key): Likewise.
30291 (inline_small_functions): Likewise.
30292 * ipa-profile.c (ipa_profile): Likewise.
30293 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30294 (ipa_make_edge_direct_to_target): Likewise.
30295 (remove_described_reference): Likewise.
30296 (ipa_impossible_devirt_target): Likewise.
30297 (propagate_controlled_uses): Likewise.
30298 (ipa_print_node_params): Likewise.
30299 (ipcp_transform_function): Likewise.
30300 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30301 (propagate_pure_const): Likewise.
30302 * ipa-reference.c (generate_summary): Likewise.
30303 (read_write_all_from_decl): Likewise.
30304 (propagate): Likewise.
30305 (ipa_reference_read_optimization_summary): Likewise.
30306 * ipa-utils.c (ipa_merge_profiles): Likewise.
30307 * ipa.c (walk_polymorphic_call_targets): Likewise.
30308 (symbol_table::remove_unreachable_nodes): Likewise.
30309 (ipa_single_use): Likewise.
30310 * passes.c (execute_todo): Likewise.
30311 * predict.c (drop_profile): Likewise.
30312 * symtab.c (symtab_node::get_dump_name): New function.
30313 (symtab_node::dump_name): Likewise.
30314 (symtab_node::dump_asm_name): Likewise.
30315 (symtab_node::dump_references): Likewise.
30316 (symtab_node::dump_referring): Likewise.
30317 (symtab_node::dump_base): Likewise.
30318 (symtab_node::debug_symtab): Likewise.
30319 * tree-sra.c (convert_callers_for_node): Likewise.
30320 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30321 * value-prof.c (init_node_map): Likewise.
30323 2017-05-23 Martin Liska <mliska@suse.cz>
30325 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30326 and symtab_node::debug_symtab to symbol_table::debug.
30327 * cgraphunit.c (analyze_functions): Use the renamed function.
30328 (symbol_table::compile): Likewise.
30329 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30330 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30331 * passes.c (execute_todo): Likewise.
30332 * symtab.c (symbol_table::dump): New function.
30333 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30335 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30337 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30338 that nonconst implies exec.
30340 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30342 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30343 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30344 (inline_edge_summary_vec): Turn into ...
30345 (ipa_call_summaries): ... this one.
30346 (redirect_to_unreachable, edge_set_predicate,
30347 evaluate_properties_for_edge, inline_summary_alloc,
30348 reset_ipa_call_summary, reset_inline_summary,
30349 inline_summary_t::duplicate): Update.
30350 (inline_edge_duplication_hook): Turn to ...
30351 (ipa_call_summary_t::duplicate): ... this one.
30352 (inline_edge_removal_hook): Turn to ...
30353 (ipa_call_summary_t::remove): ... this one.
30354 (dump_inline_edge_summary): Turn to ...
30355 (dump_ipa_call_summary): ... this one.
30356 (estimate_function_body_sizes): Update.
30357 (inline_update_callee_summaries): Update.
30358 (remap_edge_change_prob): Update.
30359 (remap_edge_summaries): Update.
30360 (inline_merge_summary): Update.
30361 (do_estimate_edge_time): Update.
30362 (inline_generate_summary): Update.
30363 (inline_read_section): Update.
30364 (inline_read_summary): Update.
30365 (inline_free_summary): Update.
30366 * ipa-inline.c (can_inline_edge_p): Update.
30367 (compute_inlined_call_time): Update.
30368 (want_inline_small_function_p): Update.
30369 (edge_badness): Update.
30370 (early_inliner): Update.
30371 * ipa-inline.h (inline_edge_summary): Turn to ...
30372 (ipa_call_summary): ... this one.
30373 (ipa_call_summary_t): New class.
30374 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30375 (ipa_call_summaries): New.
30376 (inline_edge_summary): Remove.
30377 (estimate_edge_growth): Update.
30378 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30379 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30380 * ipa-split.c (execute_split_functions): Update.
30381 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30383 2017-05-23 Tom de Vries <tom@codesourcery.com>
30385 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30386 attributes): Document rdrand effective target.
30388 2017-05-23 Tom de Vries <tom@codesourcery.com>
30390 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30391 attributes): Sort alphabetically.
30393 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30395 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30397 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30400 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30401 V2DF/V2DI splat into two separate patterns, one that handles
30402 registers, and the other that only handles memory. Drop support
30403 for splatting from a GPR on ISA 2.07 and then splitting the
30404 splat into direct move and splat.
30405 (vsx_splat_<mode>_reg): Likewise.
30406 (vsx_splat_<mode>_mem): Likewise.
30408 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30410 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30412 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30414 PR middle-end/80809
30415 * omp-low.c (finish_taskreg_remap): New function.
30416 (finish_taskreg_scan): If unit size of ctx->record_type
30417 is non-constant, unshare the size expression and replace
30418 decls in it with possible outer var refs.
30420 PR middle-end/80809
30421 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30422 GOVD_SHARED rather than GOVD_PRIVATE with it.
30423 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30424 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30426 PR middle-end/80853
30427 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30428 as last argument to build_outer_var_ref for pointer bases of array
30429 section reductions.
30431 2017-05-19 Martin Sebor <msebor@redhat.com>
30433 * print-tree.c (print_node): Print DECL_READ_P flag.
30435 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30437 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30438 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30439 * cgraph.c: Likewise.
30440 * cgraphunit.c: Likewise.
30441 * gengtype.c: Likewise.
30442 * ipa-cp.c: Likewise.
30443 * ipa-devirt.c: Likewise.
30444 * ipa-icf.c: Likewise.
30445 * ipa-predicate.c: Likewise.
30446 * ipa-profile.c: Likewise.
30447 * ipa-prop.c: Likewise.
30448 * ipa-split.c: Likewise.
30450 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30451 edge_predicate_pool, dump_inline_hints,
30452 inline_summary::account_size_time, redirect_to_unreachable,
30453 edge_set_predicate, set_hint_predicate,
30454 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30455 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30456 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30457 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30458 ipa_call_summary_t::remove, initialize_growth_caches,
30459 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30460 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30461 mark_modified, unmodified_parm_1, unmodified_parm,
30462 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30463 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30464 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30465 will_be_nonconstant_predicate, record_modified_bb_info,
30466 get_minimal_bb, record_modified, param_change_prob,
30467 phi_result_unknown_predicate, predicate_for_phi_result,
30468 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30469 estimate_function_body_sizes, compute_inline_parameters,
30470 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30471 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30472 inline_update_callee_summaries, remap_edge_change_prob,
30473 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30474 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30475 inline_analyze_function, inline_summary_t::insert,
30476 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30477 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30478 inline_free_summary): Move to ipa-fnsummary.h
30479 (predicate_t): Remove.
30480 * ipa-fnsummary.c: New file.
30481 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30482 (enum inline_hints_vals, inline_hints, agg_position_info,
30483 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30484 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30485 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30486 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30487 inline_read_summary, inline_write_summary, inline_free_summary,
30488 inline_analyze_function, initialize_inline_failed,
30489 inline_merge_summary, inline_update_overall_summary,
30490 compute_inline_parameters): Move to ipa-fnsummary.h
30491 * ipa-fnsummary.h: New file.
30492 * ipa-inline-transform.h: Include ipa-inline.h.
30493 * ipa-inline.c: LIkewise.
30495 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30497 * ipa-inline.c (edge_badness): Use inlined_time instead of
30498 inline_summaries->get.
30500 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30502 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30504 2017-05-22 Nathan Sidwell <nathan@acm.org>
30506 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30507 (fdump-lang): Document 'raw' option.
30508 * dumpfile.h (TDI_tu): Delete.
30509 * dumpfile.c (dump_files): Remove translation-unit.
30510 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30512 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30514 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30515 command option from $(AWK) call.
30516 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30517 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30518 [FORMAT]: Remove handling of variable.
30519 * config/avr/t-multilib: Regenerate.
30521 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30523 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30525 (dump_inline_summary): Do not print self_time.
30526 (estimate_function_body_sizes): Do not set self_time.
30527 (compute_inline_parameters): Likewise.
30528 (inline_read_section, inline_write_summary): Do not stream self_time.
30529 * ipa-inline.h (inline_summary): Drop self_time.
30531 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30533 * ipa-inline-analysis.c (account_size_time): Rename to ...
30534 (inline_summary::account_size_time): ... this one.
30535 (reset_ipa_call_summary): Turn to ...
30536 (ipa_call_summary::reset): ... this one.
30537 (reset_inline_summary): Turn to ...
30538 (inline_summary::reset): ... this one.
30539 (inline_summary_t::remove): Update.
30540 (inline_summary_t::duplicate): Update.
30541 (ipa_call_summary_t::remove): Update.
30542 (dump_inline_summary): Update.
30543 (estimate_function_body_sizes): Update.
30544 (compute_inline_parameters): Update.
30545 (estimate_node_size_and_time): Update.
30546 (inline_merge_summary): Update.
30547 (inline_update_overall_summary): Update.
30548 (inline_read_section): Update.
30549 (inline_write_summary): Update.
30550 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30551 add account_size_time and reset member functions.
30552 (ipa_call_summary): Add reset function.
30553 * ipa-predicate.h (predicate::operator &): Constify.
30555 2017-05-22 Richard Biener <rguenther@suse.de>
30557 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30559 2017-05-19 Jason Merrill <jason@redhat.com>
30561 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30563 2017-05-19 Marek Polacek <polacek@redhat.com>
30566 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30567 TYPE_OVERFLOW_WRAPS checks.
30569 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30571 * tree-core.h (enum omp_clause_default_kind): Add
30572 "OMP_CLAUSE_DEFAULT_PRESENT".
30573 * tree-pretty-print.c (dump_omp_clause): Handle it.
30574 * gimplify.c (enum gimplify_omp_var_data): Add
30575 "GOVD_MAP_FORCE_PRESENT".
30576 (gimplify_adjust_omp_clauses_1): Map it to
30577 "GOMP_MAP_FORCE_PRESENT".
30578 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30580 * gimplify.c (oacc_default_clause): Clarify.
30582 2017-05-19 Nathan Sidwell <nathan@acm.org>
30584 LANG_HOOK_REGISTER_DUMPS
30585 * toplev.c (general_init): Call register dump lang hook.
30586 * doc/invoke.texi: Document -fdump-lang option family.
30587 * dumpfile.c (dump_files): Remove class dump here.
30588 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30589 * dumpfile.h (tree_dump_index): Remove TDI_class.
30590 * langhooks-def.h (lhd_register_dumps): Declare.
30591 (LANG_HOOKS_REGISTER_DUMPS): Define.
30592 (LANG_HOOKS_INITIALIZER): Add it.
30593 * langhooks.c (lhd_register_dumps): Define.
30594 * langhooks.h (struct lang_hooks): Add register_dumps.
30596 2017-05-19 Nathan Sidwell <nathan@acm.org>
30598 * context.h (context::set_passes): New.
30599 * context.c (context::context): Do not create pass manager.
30600 * toplev.c (general_init): Create pass manager here.
30602 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30604 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30605 use this splitter if two add or or instructions would also work for
30606 the constant we want to generate.
30608 2017-05-19 Richard Biener <rguenther@suse.de>
30611 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30612 predicate evaluation.
30614 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30616 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30618 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30621 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30623 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30624 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30625 (inline_edge_summary_vec): Turn into ...
30626 (ipa_call_summaries): ... this one.
30627 (redirect_to_unreachable, edge_set_predicate,
30628 evaluate_properties_for_edge, inline_summary_alloc,
30629 reset_ipa_call_summary, reset_inline_summary,
30630 inline_summary_t::duplicate): Update.
30631 (inline_edge_duplication_hook): Turn to ...
30632 (ipa_call_summary_t::duplicate): ... this one.
30633 (inline_edge_removal_hook): Turn to ...
30634 (ipa_call_summary_t::remove): ... this one.
30635 (dump_inline_edge_summary): Turn to ...
30636 (dump_ipa_call_summary): ... this one.
30637 (estimate_function_body_sizes): Update.
30638 (inline_update_callee_summaries): Update.
30639 (remap_edge_change_prob): Update.
30640 (remap_edge_summaries): Update.
30641 (inline_merge_summary): Update.
30642 (do_estimate_edge_time): Update.
30643 (inline_generate_summary): Update.
30644 (inline_read_section): Update.
30645 (inline_read_summary): Update.
30646 (inline_free_summary): Update.
30647 * ipa-inline.c (can_inline_edge_p): Update.
30648 (compute_inlined_call_time): Update.
30649 (want_inline_small_function_p): Update.
30650 (edge_badness): Update.
30651 (early_inliner): Update.
30652 * ipa-inline.h (inline_edge_summary): Turn to ...
30653 (ipa_call_summary): ... this one.
30654 (ipa_call_summary_t): New class.
30655 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30656 (ipa_call_summaries): New.
30657 (inline_edge_summary): Remove.
30658 (estimate_edge_growth): Update.
30659 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30660 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30661 * ipa-split.c (execute_split_functions): Update.
30662 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30664 2017-05-19 Richard Biener <rguenther@suse.de>
30666 PR middle-end/80764
30667 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30669 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30671 * config/rs6000/rs6000.c (struct machine_function): Add field
30672 fpr_is_wrapped_separately.
30673 (rs6000_get_separate_components): Use 64 components. Handle the
30674 new FPR components.
30675 (rs6000_components_for_bb): Handle the FPR components.
30676 (rs6000_emit_prologue_components): Handle the FPR components.
30677 (rs6000_emit_epilogue_components): Handle the FPR components.
30678 (rs6000_set_handled_components): Handle the FPR components.
30679 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30680 that are already separately shrink-wrapped.
30681 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30682 that are already separately shrink-wrapped.
30684 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30687 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30690 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30691 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30692 cases where the register allocator uses a move and an offsettable
30693 memory operation to/from a FPR register on ISA 2.06/2.07.
30694 (define_peephole2 for Altivec d-form store): Likewise.
30696 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30699 * config/i386/mmx.md (*mov<mode>_internal): Enable
30700 alternatives 11, 12, 13 and 14 also for 32bit targets.
30701 Remove alternatives 15, 16, 17 and 18.
30702 * config/i386/sse.md (vec_concatv2di): Change
30703 alternative (!x, *y) to (x, ?!*Yn).
30705 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30707 * dumpfile.h (enum dump_kind): Remove stray comma.
30709 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30711 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30712 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30713 predicate::num_conditions
30714 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30715 (CHANGED): turn into predicate::changed.
30716 (agg_position_info): Move to ipa-predicate.h
30717 (add_condition, predicate::add_clause, predicate::operator &=,
30718 predicate::or_with, predicate::evaluate, predicate::probability,
30719 dump_condition, dump_clause, predicate::dump,
30720 predicate::remap_after_duplication, predicate::remap_after_inlining,
30721 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30722 (evaluate_conditions_for_known_args): Update.
30723 (set_cond_stmt_execution_predicate): Update.
30724 * ipa-inline.h: Include ipa-predicate.h
30725 (condition, inline_param_summary, conditions, agg_position_info,
30726 predicate): Move to ipa-predicate.h
30727 * ipa-predicate.c: New file.
30728 * ipa-predicate.h: New file.
30730 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30732 * final.c (leaf_function_p): Check we are not in a sequence.
30734 2017-05-18 Martin Liska <mliska@suse.cz>
30736 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30737 * dumpfile.c (dump_register): Use new enum dump_kind.
30738 (get_dump_file_name): Likewise.
30739 (dump_enable_all): Likewise.
30740 (dump_switch_p_1): Likewise.
30741 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30742 * dumpfile.h (enum dump_kind): New enum type.
30743 (struct dump_file_info): Create constructor and
30744 format fields and comments.
30745 * passes.c (pass_manager::register_one_dump_file):
30747 * statistics.c (statistics_early_init): Likewise.
30748 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30749 TDF_TREE with TDF_SLIM.
30750 (gather_memory_references_ref): Likewise.
30752 2017-05-18 Martin Liska <mliska@suse.cz>
30754 * vec.h (struct vnull): Use it.
30756 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30758 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30759 (true_predicate, false_predicate, true_predicate_p,
30760 false_predicate_p): Remove.
30761 (single_cond_predicate, not_inlined_predicate): Turn to member function
30763 (add_condition): Update.
30764 (add_clause): Turn to...
30765 (predicate::add_clause): ... this one; update; allow passing NULL
30767 (and_predicates): Turn to ...
30768 (predicate::operator &=): ... this one.
30769 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30770 (or_predicates): Turn to ...
30771 (predicate::or_with): ... this one.
30772 (evaluate_predicate): Turn to ...
30773 (predicate::evaluate): ... this one.
30774 (predicate_probability): Turn to ...
30775 (predicate::probability): ... this one.
30776 (dump_condition): Update.
30777 (dump_predicate): Turn to ...
30778 (predicate::dump): ... this one.
30779 (account_size_time): Update.
30780 (edge_set_predicate): Update.
30781 (set_hint_predicate): UPdate.
30782 (evaluate_conditions_for_known_args): Update.
30783 (evaluate_properties_for_edge): Update.
30784 (remap_predicate_after_duplication): Turn to...
30785 (predicate::remap_after_duplication): ... this one.
30786 (remap_hint_predicate_after_duplication): Update.
30787 (inline_summary_t::duplicate): UPdate.
30788 (dump_inline_edge_summary): Update.
30789 (dump_inline_summary): Update.
30790 (set_cond_stmt_execution_predicate): Update.
30791 (set_switch_stmt_execution_predicate): Update.
30792 (compute_bb_predicates): Update.
30793 (will_be_nonconstant_expr_predicate): Update.
30794 (will_be_nonconstant_predicate): Update.
30795 (phi_result_unknown_predicate): Update.
30796 (predicate_for_phi_result): Update.
30797 (array_index_predicate): Update.
30798 (estimate_function_body_sizes): Update.
30799 (estimate_node_size_and_time): Update.
30800 (estimate_ipcp_clone_size_and_time): Update.
30801 (remap_predicate): Rename to ...
30802 (predicate::remap_after_inlining): ... this one.
30803 (remap_hint_predicate): Update.
30804 (inline_merge_summary): Update.
30805 (inline_update_overall_summary): Update.
30806 (estimate_size_after_inlining): Update.
30807 (read_predicate): Rename to ...
30808 (predicate::stream_in): ... this one.
30809 (read_inline_edge_summary): Update.
30810 (write_predicate): Rename to ...
30811 (predicate::stream_out): ... this one.
30812 (write_inline_edge_summary): Update.
30813 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
30814 (clause_t): Turn to uint32_t
30815 (predicate): Turn to class; implement constructor and operators
30817 (size_time_entry): Update.
30818 (inline_summary): Update.
30819 (inline_edge_summary): Update.
30821 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
30823 * fold-const.c (fold_binary_loc): Move transformation...
30824 * match.pd (C - X CMP X): ... here.
30826 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30828 * config/sparc/sparc.c (sparc_option_override): Set function
30829 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30830 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30832 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30834 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30836 2017-05-18 Marek Polacek <polacek@redhat.com>
30839 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30840 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30841 gimple_assign_load_p.
30843 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30845 PR middle-end/80692
30846 * real.c (do_compare): Give decimal_do_compare preference over
30847 comparing just the signs.
30849 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30851 * doc/md.texi (Canonicalization of Instructions): Describe the
30852 canonical form of instructions that inherently set a condition
30855 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
30857 PR middle-end/80775
30858 * tree-cfg.c: Move deletion of unreachable case statements to after
30859 the merging of consecutive case labels.
30861 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30863 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30864 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
30865 restoring of callee-saved registers.
30867 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
30869 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
30870 * config/visium/visium.c (single_set_and_flags): Likewise.
30871 * config/visium/visium.md (Substitutions): Likewise.
30873 2017-05-17 Martin Liska <mliska@suse.cz>
30875 * cfg.c: Introduce dump_flags_t type and
30876 use it instead of int type.
30878 * cfghooks.c: Likewise.
30879 * cfghooks.h (struct cfg_hooks): Likewise.
30880 * cfgrtl.c: Likewise.
30881 * cfgrtl.h: Likewise.
30882 * cgraph.c (cgraph_node::get_body): Likewise.
30883 * coretypes.h: Likewise.
30884 * domwalk.c: Likewise.
30885 * domwalk.h: Likewise.
30886 * dumpfile.c (struct dump_option_value_info): Likewise.
30887 (dump_enable_all): Likewise.
30888 (dump_switch_p_1): Likewise.
30889 (opt_info_switch_p): Likewise.
30890 * dumpfile.h (enum tree_dump_index): Likewise.
30891 (struct dump_file_info): Likewise.
30892 * genemit.c: Likewise.
30893 * generic-match-head.c: Likewise.
30894 * gengtype.c (open_base_files): Likewise.
30895 * gimple-pretty-print.c: Likewise.
30896 * gimple-pretty-print.h: Likewise.
30897 * graph.c (print_graph_cfg): Likewise.
30898 * graphite-scop-detection.c (dot_all_sese): Likewise.
30899 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30900 * loop-unroll.c (report_unroll): Likewise.
30901 * passes.c (pass_manager::register_one_dump_file): Likewise.
30902 * print-tree.c: Likewise.
30903 * statistics.c: Likewise.
30904 * tree-cfg.c: Likewise.
30905 * tree-cfg.h: Likewise.
30906 * tree-dfa.c: Likewise.
30907 * tree-dfa.h: Likewise.
30908 * tree-dump.c (dump_function): Likewise.
30909 * tree-dump.h (struct dump_info): Likewise.
30910 * tree-pretty-print.c: Likewise.
30911 * tree-pretty-print.h: Likewise.
30912 * tree-ssa-live.c: Likewise.
30913 * tree-ssa-live.h: Likewise.
30914 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30915 * tree-vect-loop.c: Likewise.
30916 * tree-vect-slp.c: Likewise.
30918 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30919 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30921 PR tree-optimization/80457
30922 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30923 of all arguments to a statement as scalar_to_vec operations.
30924 (vectorizable_call): Adjust call to vect_model_simple_cost for
30926 (vectorizable_conversion): Likewise.
30927 (vectorizable_assignment): Likewise.
30928 (vectorizable_shift): Likewise.
30929 (vectorizable_operation): Likewise.
30930 (vectorizable_comparison): Likewise.
30931 (vect_is_simple_cond): Record the def types for operands.
30932 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30933 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30934 for statement argument count.
30936 2017-05-16 Carl Love <cel@us.ibm.com>
30938 * config/rs6000/rs6000-c: Add support for built-in functions
30939 vector unsigned long long vec_bperm (vector unsigned long long,
30940 vector unsigned char)
30941 vector signed long long vec_mule (vector signed int,
30943 vector unsigned long long vec_mule (vector unsigned int,
30944 vector unsigned int)
30945 vector signed long long vec_mulo (vector signed int,
30947 vector unsigned long long vec_mulo (vector unsigned int,
30948 vector unsigned int)
30949 vector signed char vec_sldw (vector signed char,
30950 vector signed char,
30952 vector unsigned char vec_sldw (vector unsigned char,
30953 vector unsigned char,
30955 vector signed short vec_sldw (vector signed short,
30956 vector signed short,
30958 vector unsigned short vec_sldw (vector unsigned short,
30959 vector unsigned short,
30961 vector signed int vec_sldw (vector signed int,
30964 vector unsigned int vec_sldw (vector unsigned int,
30965 vector unsigned int,
30967 vector signed long long vec_sldw (vector signed long long,
30968 vector signed long long,
30970 vector unsigned long long vec_sldw (vector unsigned long long,
30971 vector unsigned long long,
30973 * config/rs6000/rs6000-c: Add support for built-in functions
30974 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30975 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30976 * doc/extend.texi: Update the built-in documentation for the
30977 new built-in functions.
30979 2017-05-16 Marek Polacek <polacek@redhat.com>
30983 * tree.c (save_expr): Don't fold the expression.
30985 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30987 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30988 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30989 and (?*y,m). Update insn attributes.
30991 2017-05-16 Martin Liska <mliska@suse.cz>
30993 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30994 flags argument of print_gimple_stmt, print_gimple_expr,
30995 print_generic_stmt and print_generic_expr.
30996 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30997 * coretypes.h: Likewise.
30998 * except.c (dump_eh_tree): Likewise.
30999 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
31000 * gimple-pretty-print.h: Likewise.
31001 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
31002 (backprop::push_to_worklist): Likewise.
31003 (backprop::pop_from_worklist): Likewise.
31004 (backprop::process_use): Likewise.
31005 (backprop::intersect_uses): Likewise.
31006 (note_replacement): Likewise.
31007 * gimple-ssa-store-merging.c
31008 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31009 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31010 (pass_store_merging::execute): Likewise.
31011 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31012 (ssa_base_cand_dump_callback): Likewise.
31013 (dump_incr_vec): Likewise.
31014 (replace_refs): Likewise.
31015 (replace_mult_candidate): Likewise.
31016 (create_add_on_incoming_edge): Likewise.
31017 (create_phi_basis): Likewise.
31018 (insert_initializers): Likewise.
31019 (all_phi_incrs_profitable): Likewise.
31020 (introduce_cast_before_cand): Likewise.
31021 (replace_one_candidate): Likewise.
31022 * gimplify.c (gimplify_expr): Likewise.
31023 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31024 (set_rename): Likewise.
31025 (rename_uses): Likewise.
31026 (copy_loop_phi_nodes): Likewise.
31027 (add_close_phis_to_merge_points): Likewise.
31028 (copy_loop_close_phi_args): Likewise.
31029 (copy_cond_phi_args): Likewise.
31030 (graphite_copy_stmts_from_block): Likewise.
31031 (translate_pending_phi_nodes): Likewise.
31032 * graphite-poly.c (print_pdr): Likewise.
31033 (dump_gbb_cases): Likewise.
31034 (dump_gbb_conditions): Likewise.
31035 (print_scop_params): Likewise.
31036 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31037 (build_cross_bb_scalars_use): Likewise.
31038 (gather_bbs::before_dom_children): Likewise.
31039 * hsa-dump.c (dump_hsa_immed): Likewise.
31040 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31041 (get_replacement_map): Likewise.
31042 * ipa-inline-analysis.c (dump_condition): Likewise.
31043 (estimate_function_body_sizes): Likewise.
31044 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31045 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31046 * ipa-prop.c (ipa_dump_param): Likewise.
31047 (ipa_print_node_jump_functions_for_edge): Likewise.
31048 (ipa_modify_call_arguments): Likewise.
31049 (ipa_modify_expr): Likewise.
31050 (ipa_dump_param_adjustments): Likewise.
31051 (ipa_dump_agg_replacement_values): Likewise.
31052 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31053 * ipa-pure-const.c (check_stmt): Likewise.
31054 (pass_nothrow::execute): Likewise.
31055 * ipa-split.c (execute_split_functions): Likewise.
31056 * omp-offload.c (dump_oacc_loop_part): Likewise.
31057 (dump_oacc_loop): Likewise.
31058 * trans-mem.c (tm_log_emit): Likewise.
31059 (tm_memopt_accumulate_memops): Likewise.
31060 (dump_tm_memopt_set): Likewise.
31061 (dump_tm_memopt_transform): Likewise.
31062 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31063 (print_loop): Likewise.
31064 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31065 (chkp_gather_checks_info): Likewise.
31066 (chkp_get_check_result): Likewise.
31067 (chkp_remove_check_if_pass): Likewise.
31068 (chkp_use_outer_bounds_if_possible): Likewise.
31069 (chkp_reduce_bounds_lifetime): Likewise.
31070 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31071 (chkp_mark_completed_bounds): Likewise.
31072 (chkp_register_incomplete_bounds): Likewise.
31073 (chkp_mark_invalid_bounds): Likewise.
31074 (chkp_maybe_copy_and_register_bounds): Likewise.
31075 (chkp_build_returned_bound): Likewise.
31076 (chkp_get_bound_for_parm): Likewise.
31077 (chkp_build_bndldx): Likewise.
31078 (chkp_get_bounds_by_definition): Likewise.
31079 (chkp_generate_extern_var_bounds): Likewise.
31080 (chkp_get_bounds_for_decl_addr): Likewise.
31081 * tree-chrec.c (chrec_apply): Likewise.
31082 * tree-data-ref.c (dump_data_reference): Likewise.
31083 (dump_subscript): Likewise.
31084 (dump_data_dependence_relation): Likewise.
31085 (analyze_overlapping_iterations): Likewise.
31086 * tree-inline.c (expand_call_inline): Likewise.
31087 (tree_function_versioning): Likewise.
31088 * tree-into-ssa.c (dump_defs_stack): Likewise.
31089 (dump_currdefs): Likewise.
31090 (dump_names_replaced_by): Likewise.
31091 (dump_update_ssa): Likewise.
31092 (update_ssa): Likewise.
31093 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31094 * tree-parloops.c (build_new_reduction): Likewise.
31095 (try_create_reduction_list): Likewise.
31096 (ref_conflicts_with_region): Likewise.
31097 (oacc_entry_exit_ok_1): Likewise.
31098 (oacc_entry_exit_single_gang): Likewise.
31099 * tree-pretty-print.h: Likewise.
31100 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31101 (get_scalar_evolution): Likewise.
31102 (add_to_evolution): Likewise.
31103 (get_loop_exit_condition): Likewise.
31104 (analyze_evolution_in_loop): Likewise.
31105 (analyze_initial_condition): Likewise.
31106 (analyze_scalar_evolution): Likewise.
31107 (instantiate_scev): Likewise.
31108 (number_of_latch_executions): Likewise.
31109 (gather_chrec_stats): Likewise.
31110 (final_value_replacement_loop): Likewise.
31111 (scev_const_prop): Likewise.
31112 * tree-sra.c (dump_access): Likewise.
31113 (disqualify_candidate): Likewise.
31114 (create_access): Likewise.
31115 (reject): Likewise.
31116 (maybe_add_sra_candidate): Likewise.
31117 (create_access_replacement): Likewise.
31118 (analyze_access_subtree): Likewise.
31119 (analyze_all_variable_accesses): Likewise.
31120 (sra_modify_assign): Likewise.
31121 (initialize_constant_pool_replacements): Likewise.
31122 (find_param_candidates): Likewise.
31123 (decide_one_param_reduction): Likewise.
31124 (replace_removed_params_ssa_names): Likewise.
31125 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31126 * tree-ssa-copy.c (dump_copy_of): Likewise.
31127 (copy_prop_visit_cond_stmt): Likewise.
31128 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31129 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31130 (record_equivalences_from_stmt): Likewise.
31131 * tree-ssa-dse.c (compute_trims): Likewise.
31132 (delete_dead_call): Likewise.
31133 (delete_dead_assignment): Likewise.
31134 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31135 (forward_propagate_into_cond): Likewise.
31136 (pass_forwprop::execute): Likewise.
31137 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31138 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31140 (move_computations_worker): Likewise.
31141 (execute_sm): Likewise.
31142 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31143 (remove_exits_and_undefined_stmts): Likewise.
31144 (remove_redundant_iv_tests): Likewise.
31145 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31146 (adjust_iv_update_pos): Likewise.
31147 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31148 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31149 (value_replacement): Likewise.
31150 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31151 * tree-ssa-pre.c (print_pre_expr): Likewise.
31152 (get_representative_for): Likewise.
31153 (create_expression_by_pieces): Likewise.
31154 (insert_into_preds_of_block): Likewise.
31155 (eliminate_insert): Likewise.
31156 (eliminate_dom_walker::before_dom_children): Likewise.
31157 (eliminate): Likewise.
31158 (remove_dead_inserted_code): Likewise.
31159 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31160 * tree-ssa-reassoc.c (get_rank): Likewise.
31161 (eliminate_duplicate_pair): Likewise.
31162 (eliminate_plus_minus_pair): Likewise.
31163 (eliminate_not_pairs): Likewise.
31164 (undistribute_ops_list): Likewise.
31165 (eliminate_redundant_comparison): Likewise.
31166 (update_range_test): Likewise.
31167 (optimize_range_tests_var_bound): Likewise.
31168 (optimize_vec_cond_expr): Likewise.
31169 (rewrite_expr_tree): Likewise.
31170 (rewrite_expr_tree_parallel): Likewise.
31171 (linearize_expr): Likewise.
31172 (break_up_subtract): Likewise.
31173 (linearize_expr_tree): Likewise.
31174 (attempt_builtin_powi): Likewise.
31175 (attempt_builtin_copysign): Likewise.
31176 (transform_stmt_to_copy): Likewise.
31177 (transform_stmt_to_multiply): Likewise.
31178 (dump_ops_vector): Likewise.
31179 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31180 (print_scc): Likewise.
31181 (set_ssa_val_to): Likewise.
31182 (visit_reference_op_store): Likewise.
31183 (visit_use): Likewise.
31184 (sccvn_dom_walker::before_dom_children): Likewise.
31185 (run_scc_vn): Likewise.
31186 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31188 (expr_hash_elt::print): Likewise.
31189 (const_and_copies::pop_to_marker): Likewise.
31190 (const_and_copies::record_const_or_copy_raw): Likewise.
31191 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31192 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31193 (dump_predicates): Likewise.
31194 (find_uninit_use): Likewise.
31195 (warn_uninitialized_phi): Likewise.
31196 (pass_late_warn_uninitialized::execute): Likewise.
31197 * tree-ssa.c (verify_vssa): Likewise.
31198 (verify_ssa): Likewise.
31199 (maybe_optimize_var): Likewise.
31200 * tree-vrp.c (dump_value_range): Likewise.
31201 (dump_all_value_ranges): Likewise.
31202 (dump_asserts_for): Likewise.
31203 (register_edge_assert_for_2): Likewise.
31204 (vrp_visit_cond_stmt): Likewise.
31205 (vrp_visit_switch_stmt): Likewise.
31206 (vrp_visit_stmt): Likewise.
31207 (vrp_visit_phi_node): Likewise.
31208 (simplify_cond_using_ranges_1): Likewise.
31209 (fold_predicate_in): Likewise.
31210 (evrp_dom_walker::before_dom_children): Likewise.
31211 (evrp_dom_walker::push_value_range): Likewise.
31212 (evrp_dom_walker::pop_value_range): Likewise.
31213 (execute_early_vrp): Likewise.
31215 2017-05-16 Richard Biener <rguenther@suse.de>
31217 * dwarf2out.c (loc_list_from_tree_1): Do not create
31218 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31220 2017-05-16 Richard Biener <rguenther@suse.de>
31222 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31224 (note_variable_value_in_expr): If we resolved the decl ref
31225 do not push to the stack.
31227 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31229 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31230 operations in fast-math mode.
31231 (vaddq_f16): Likewise.
31232 (vmul_f16): Likewise.
31233 (vmulq_f16): Likewise.
31234 (vsub_f16): Likewise.
31235 (vsubq_f16): Likewise.
31236 * config/arm/neon.md (add<mode>3): New.
31238 (fma:<VH:mode>3): New. Also remove outdated comment.
31241 2017-05-16 Martin Liska <mliska@suse.cz>
31245 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31246 (odr_types_equivalent_p): Likewise.
31248 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31250 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31252 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31255 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31256 non-interunit SSE move alternatives with '?'.
31257 (zero-extendsidi peephole2): New peephole to skip intermediate
31258 general register in SSE zero-extend sequence.
31260 2017-05-15 Jeff Law <law@redhat.com>
31262 * reorg.c (relax_delay_slots): Create a new variable to hold
31263 the temporary target rather than clobbering TARGET_LABEL.
31265 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31266 missing argument to extract_bit_field call.
31267 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31269 2017-05-15 Martin Liska <mliska@suse.cz>
31272 * gcc.c (process_command): Do not allow empty argument of -o option.
31274 2017-05-15 Renlin Li <renlin.li@arm.com>
31276 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31277 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31278 * config/aarch64/constraints.md (Usf): Add long call check.
31279 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31280 (call_value): Likewise.
31281 (sibcall): Likewise.
31282 (sibcall_value): Likewise.
31284 (call_value_insn): New.
31285 (sibcall_insn): Update rtx pattern.
31286 (sibcall_value_insn): Likewise.
31287 (call_internal): Remove.
31288 (call_value_internal): Likewise.
31289 (sibcall_internal): Likewise.
31290 (sibcall_value_internal): Likewise.
31291 (call_reg): Likewise.
31292 (call_symbol): Likewise.
31293 (call_value_reg): Likewise.
31294 (call_value_symbol): Likewise.
31296 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31299 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31301 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31303 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31304 compatible with CCGOCmode and with CCZmode.
31306 2017-05-14 Martin Sebor <msebor@redhat.com>
31308 PR middle-end/77671
31309 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31310 (gimple_fold_builtin_snprintf): Same.
31311 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31312 (gimple_fold_builtin_snprintf): Same.
31313 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31314 of character types.
31315 (is_call_safe): New function.
31316 (try_substitute_return_value): Call it.
31317 (try_simplify_call): New function.
31318 (pass_sprintf_length::handle_gimple_call): Call it.
31320 2017-05-14 Martin Sebor <msebor@redhat.com>
31322 PR middle-end/80669
31323 * builtins.c (expand_builtin_stpncpy): Simplify.
31325 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31327 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31328 * config/i386/i386.h
31329 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31330 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31331 (struct machine_function): Add new members call_ms2sysv,
31332 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31333 (struct machine_frame_state): New fields sp_realigned and
31334 sp_realigned_offset.
31335 * config/i386/i386.c
31336 (enum xlogue_stub): New enum.
31337 (enum xlogue_stub_sets): New enum.
31338 (class xlogue_layout): New class.
31339 (struct ix86_frame): New fields stack_realign_allocate_offset,
31340 stack_realign_offset and outlined_save_offset. Modify comments to
31341 detail stack layout when using out-of-line stubs.
31342 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31343 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31344 -mcall-ms2sysv-xlogues.
31345 (stub_managed_regs): New static variable.
31346 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31347 registers managed by out-of-line stub.
31348 (disable_call_ms2sysv_xlogues): New function.
31349 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31350 m->call_ms2sysv when appropriate and compute frame layout for
31352 (sp_valid_at, fp_valid_at): New inline functions.
31353 (choose_basereg): New function.
31354 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31356 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31357 Use align parameter of choose_baseaddr to generated aligned SSE movs
31359 (pro_epilogue_adjust_stack): Modify to track
31360 machine_frame_state::sp_realigned.
31361 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31362 (ix86_nsaved_sseregs): Likewise.
31363 (ix86_emit_save_regs): Likewise.
31364 (ix86_emit_save_regs_using_mov): Likewise.
31365 (ix86_emit_save_sse_regs_using_mov): Likewise.
31366 (get_scratch_register_on_entry): Likewise.
31367 (gen_frame_set): New function.
31368 (gen_frame_load): Likewise.
31369 (gen_frame_store): Likewise.
31370 (emit_outlined_ms2sysv_save): Likewise.
31371 (emit_outlined_ms2sysv_restore): Likewise.
31372 (ix86_expand_prologue): Modify stack re-alignment code and call
31373 emit_outlined_ms2sysv_save when appropriate.
31374 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31375 parameter rtx_insn *insn, which allows the function to be used to only
31376 generate the notes.
31377 (ix86_expand_epilogue): Modify validity checks of frame and stack
31378 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31379 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31380 * config/i386/predicates.md
31381 (save_multiple): New predicate.
31382 (restore_multiple): Likewise.
31383 * config/i386/sse.md
31384 (save_multiple<mode>): New pattern.
31385 (save_multiple_realign<mode>): Likewise.
31386 (restore_multiple<mode>): Likewise.
31387 (restore_multiple_and_return<mode>): Likewise.
31388 (restore_multiple_leave_return<mode>): Likewise.
31389 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31391 2017-05-14 Julia Koval <julia.koval@intel.com>
31393 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31394 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31395 (__builtin_ia32_xsetbv): New builtins.
31396 * config/i386/i386.c (ix86_expand_special_args_builtin):
31398 (ix86_expand_builtin): Special expand for new intrinsics.
31399 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31400 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31401 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31403 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31405 * cfganal.c (inverted_post_order_compute): Change argument type
31407 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31408 * df-core.c (rest_of_handle_df_initialize): Adjust.
31409 (rest_of_handle_df_finish): Likewise.
31410 (df_analyze_1): Likewise.
31411 (df_analyze): Likewise.
31412 (loop_inverted_post_order_compute): Change argument to be a vec *.
31413 (df_analyze_loop): Adjust.
31414 (df_get_n_blocks): Likewise.
31415 (df_get_postorder): Likewise.
31416 * df.h (struct df_d): Change field to be a vec.
31417 * lcm.c (compute_laterin): Adjust.
31418 (compute_available): Likewise.
31419 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31420 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31421 * tree-ssa-pre.c (compute_antic): Likewise.
31423 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31425 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31426 (depth_first_search::depth_first_search): Change structure init
31427 function to this constructor.
31428 (depth_first_search::add_bb): Rename function to this member.
31429 (depth_first_search::execute): Likewise.
31430 (flow_dfs_compute_reverse_finish): Adjust.
31432 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31434 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31435 (longest_simple_path): Likewise.
31436 * shrink-wrap.c (spread_components): Likewise.
31437 (disqualify_problematic_components): Likewise.
31438 (emit_common_heads_for_components): Likewise.
31439 (emit_common_tails_for_components): Likewise.
31440 (insert_prologue_epilogue_for_components): Likewise.
31442 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31444 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31447 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31449 * df-core.c (df_set_blocks): Start using auto_bitmap.
31450 (df_compact_blocks): Likewise.
31451 * df-problems.c (df_rd_confluence_n): Likewise.
31452 * df-scan.c (df_insn_rescan_all): Likewise.
31453 (df_process_deferred_rescans): Likewise.
31454 (df_update_entry_block_defs): Likewise.
31455 (df_update_exit_block_uses): Likewise.
31456 (df_entry_block_bitmap_verify): Likewise.
31457 (df_exit_block_bitmap_verify): Likewise.
31458 (df_scan_verify): Likewise.
31459 * lra-constraints.c (lra_constraints): Likewise.
31460 (undo_optional_reloads): Likewise.
31461 (lra_undo_inheritance): Likewise.
31462 * lra-remat.c (calculate_gen_cands): Likewise.
31463 (do_remat): Likewise.
31464 * lra-spills.c (assign_spill_hard_regs): Likewise.
31465 (spill_pseudos): Likewise.
31466 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31467 (bitmap_set_subtract_values): Likewise.
31469 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31471 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31472 management with auto_bitmap.
31473 (fix_inter_tick): Likewise.
31474 (fix_recovery_deps): Likewise.
31475 * ira.c (add_store_equivs): Likewise.
31476 (find_moveable_pseudos): Likewise.
31477 (split_live_ranges_for_shrink_wrap): Likewise.
31478 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31479 (rtx_reuse_manager::seen_def_p): Likewise.
31480 (rtx_reuse_manager::set_seen_def): Likewise.
31481 * print-rtl.h (class rtx_reuse_manager): Likewise.
31483 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31485 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31487 (migrate_btr_def): Likewise.
31488 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31489 * df-core.c (loop_post_order_compute): Likewise.
31490 (loop_inverted_post_order_compute): Likewise.
31491 * hsa-common.h: Likewise.
31492 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31493 * init-regs.c (initialize_uninitialized_regs): Likewise.
31494 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31495 (inline_small_functions): Likewise.
31496 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31497 * ira.c (combine_and_move_insns): Likewise.
31498 (build_insn_chain): Likewise.
31499 * loop-invariant.c (find_invariants): Likewise.
31500 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31501 * predict.c (tree_predict_by_opcode): Likewise.
31502 (predict_paths_leading_to): Likewise.
31503 (predict_paths_leading_to_edge): Likewise.
31504 (estimate_loops_at_level): Likewise.
31505 (estimate_loops): Likewise.
31506 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31507 (spread_components): Likewise.
31508 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31509 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31510 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31511 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31512 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31513 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31514 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31515 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31516 (mark_threaded_blocks): Likewise.
31517 (thread_through_all_blocks): Likewise.
31518 * tree-ssa.c (verify_ssa): Likewise.
31519 (execute_update_addresses_taken): Likewise.
31520 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31522 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31524 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31526 (post_order_compute): Likewise.
31527 (inverted_post_order_compute): Likewise.
31528 (pre_and_rev_post_order_compute_fn): Likewise.
31530 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31532 * genrecog.c (int_set::int_set): Explicitly construct our
31533 auto_vec base class.
31534 * vec.h (auto_vec::auto_vec): New constructor.
31536 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31538 * bitmap.h (class auto_bitmap): New constructor taking
31539 bitmap_obstack * argument.
31541 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31543 * bitmap.h (class auto_bitmap): Change type of m_bits to
31544 bitmap_head, and adjust ctor / dtor and member operators.
31546 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31548 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31549 when returned register mode doesn't match original mode.
31551 2017-05-12 Jeff Law <law@redhat.com>
31552 Jakub Jelinek <jakub@redhat.com>
31554 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31555 we look for cc setter after the compare-elim changes.
31556 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31557 within the vector to match what compare-elim now expects.
31558 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31559 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31561 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31562 after the compare-elim changes.
31563 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31564 the vector to match what compare-elim now expects.
31565 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31566 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31567 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31568 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31569 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31571 * config/visium/visium.c (single_set_and_flags): Fix where
31572 we look for cc setter after the compare-elim changes.
31573 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31574 with the vector to match what compare-elim now expects.
31575 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31576 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31577 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31578 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31579 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31580 (neg<mode>2_insn_set_overflow): Likewise.
31582 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31584 PR middle-end/79794
31585 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31586 maybe_expand_insn call, set ops[0].target. If still set after call,
31587 set alt_rtl. Add extra arg to recursive calls.
31588 (extract_bit_field): Add alt_rtl argument. Pass to
31590 * expmed.h (extract_bit_field): Fix prototype.
31591 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31592 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31593 to extract_bit_field_calls.
31594 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31595 Pass alt_rtl to extract_bit_field calls.
31596 * calls.c (store_unaligned_arguments_into_psuedos)
31597 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31598 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31600 * optabs.h (struct expand_operand): Add target bitfield.
31602 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31604 * compare-elim.c (try_eliminate_compare): Canonicalize
31605 operation with embedded compare to
31606 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31607 (set (reg) (operation)].
31609 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31611 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31614 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31615 cost of adding a carry flag for ADC instruction.
31616 [case MINUS]: Ignore the cost of subtracting a carry flag
31617 for SBB instruction.
31619 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31621 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31623 * config/rs6000/bmiintrin.h: New file.
31624 * config/rs6000/bmi2intrin.h: New file.
31625 * config/rs6000/x86intrin.h: New file.
31627 2017-05-12 Jeff Law <law@redhat.com>
31629 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31632 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31634 PR middle-end/80707
31635 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31637 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31639 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31640 early expansion of vector divide builtins.
31641 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31642 builtins identified as having unsigned arguments.
31644 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31646 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31647 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31648 expansion of vector logical operations (and, andc, or, xor,
31651 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31653 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31654 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31656 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31658 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31659 early GIMPLE expansion of vector multiplies.
31661 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31663 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31664 TARGET_HAVE_MOVT conditional.
31665 (movt splitter): Likewise.
31667 2017-05-12 Richard Biener <rguenther@suse.de>
31669 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31670 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31671 Fold all stmts not inplace.
31673 2017-05-12 Richard Biener <rguenther@suse.de>
31675 PR tree-optimization/80713
31676 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31677 inserted_exprs bit for not removed stmts.
31679 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31681 PR middle-end/69921
31682 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31683 parallelized" attribute for parallelized OpenACC kernels.
31684 * omp-offload.c (execute_oacc_device_lower): Use it.
31686 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31687 Set "oacc kernels" attribute.
31688 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31689 parameter. Adjust all users.
31690 (oacc_fn_attrib_kernels_p): Remove function.
31691 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31692 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31693 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31694 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31695 assert "oacc kernels" attribute is set.
31697 2017-05-11 Carl Love <cel@us.ibm.com>
31699 * config/rs6000/rs6000-c: Add support for built-in functions
31700 vector unsigned char vec_popcnt (vector signed char)
31701 vector unsigned char vec_popcnt (vector unsigned char)
31702 vector unsigned short vec_popcnt (vector signed short)
31703 vector unsigned short vec_popcnt (vector unsigned short)
31704 vector unsigned int vec_popcnt (vector signed int)
31705 vector unsigned int vec_popcnt (vector unsigned int)
31706 vector unsigned long long vec_popcnt (vector signed long long)
31707 vector unsigned long long vec_popcnt (vector unsigned long long)
31708 vector signed long long vec_slo (vector signed long long,
31709 vector signed char)
31710 vector signed long long vec_slo (vector signed long long,
31711 vector unsigned char)
31712 vector unsigned long long vec_slo (vector unsigned long long,
31713 vector signed char)
31714 vector unsigned long long vec_slo (vector unsigned long long,
31715 vector unsigned char)
31716 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31717 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31718 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31719 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31720 * doc/extend.texi: Update the built-in documentation file for the
31721 new built-in functions.
31723 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31725 * attribs.h (sorted_attr_string): Move machine independent
31726 functions for target clone support from the i386 port to common
31727 code. Rename ix86_function_versions to common_function_versions.
31728 Rename make_name to make_unique_name.
31729 (common_function_versions): Likewise.
31730 (make_unique_name): Likewise.
31731 (make_dispatcher_decl): Likewise.
31732 (is_function_default_version): Likewise.
31733 * attribs.c (attr_strcmp): Likewise.
31734 (sorted_attr_string): Likewise.
31735 (common_function_versions): Likewise.
31736 (make_unique_name): Likewise.
31737 (make_dispatcher_decl): Likewise.
31738 (is_function_default_version): Likewise.
31739 * config/i386/i386.c (attr_strcmp): Likewise.
31740 (sorted_attr_string): Likewise.
31741 (ix86_function_versions): Likewise.
31742 (make_name): Likewise.
31743 (make_dispatcher_decl): Likewise.
31744 (is_function_default_version): Likewise.
31745 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31747 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31750 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31751 Account for direct move costs for vec_construct of integer
31754 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31757 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31758 (UNSPEC_STX_ATOMIC): Ditto.
31759 (loaddi_via_sse): New insn.
31760 (storedi_via_sse): Ditto.
31761 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31762 Update corresponding peephole2 patterns.
31763 (atomic_storedi_fpu): Ditto.
31765 2017-05-11 Julia Koval <julia.koval@intel.com>
31767 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31768 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31770 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31771 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31772 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31774 2017-05-11 Nathan Sidwell <nathan@acm.org>
31776 * graphite-poly.c: Include dumpfile.h.
31778 * dumpfle.h (dump_function): Declare here ...
31779 * tree-dump.h (dump_function): ... not here.
31780 * dumpfile.c: #include tree-cfg.h.
31781 (dump_function): Move here from ...
31782 * tree-dump.c (dump_function): ... here.
31783 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31784 * graphite-poly.c: Don't include tree-dump.h.
31785 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31786 * print-tree.c: Likewise.
31787 * stor-layout.c: Likewise.
31788 * tree-nested.c: Likewise.
31790 * dumpfile.c (dump_start): Use TDF_FLAGS.
31791 (dump_enable_all): Fix TDF_KIND check thinko.
31793 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31795 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31796 array entries to represent two legal parameterizations of the
31797 overloaded __builtin_cmpb function, as represented by the
31798 P6_OV_BUILTIN_CMPB constant.
31799 (altivec_resolve_overloaded_builtin): Add special case handling
31800 for the __builtin_cmpb function, as represented by the
31801 P6_OV_BUILTIN_CMPB constant.
31802 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
31803 (BU_P6_64BIT_2): New macro.
31804 (BU_P6_OVERLOAD_2): New macro
31805 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
31806 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
31807 (CMPB): Add overload support to represent both 32-bit and 64-bit
31808 compare-bytes function.
31809 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
31810 support for TARGET_CMPB.
31811 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
31812 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31813 documentation of the __builtin_cmpb overloaded built-in function.
31815 2017-05-11 Richard Biener <rguenther@suse.de>
31817 PR tree-optimization/80705
31818 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
31819 bases are not vectorizable.
31821 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31823 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
31824 when counting register pressure.
31826 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31828 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31829 (iv_ca_more_deps): Renamed to ...
31830 (iv_ca_compare_deps): ... this.
31831 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31833 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31835 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31837 (determine_group_iv_costs): ... here.
31838 (find_inv_vars_cb): Record inv var if it's not recorded before.
31840 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31842 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31843 (get_shiftadd_cost): Ditto.
31845 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31847 * tree-ssa-address.c: Include header file.
31848 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31850 (add_to_parts): Refactor.
31851 (addr_to_parts): New parameter. Update use of move_hint_to_base.
31852 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
31855 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31857 PR tree-optimization/53090
31858 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31860 (extract_cond_operands): Detect condition with IV on both sides
31861 and return COMP_IV_EXPR_2.
31862 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31863 (rewrite_use_compare): Simplify by removing call to function
31864 extract_cond_operands.
31866 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31868 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
31869 (extract_cond_operands): Detect condition comparing against non-
31870 invariant bound and return appropriate enum value.
31871 (find_interesting_uses_cond): Update use of extract_cond_operands.
31872 Handle its return value accordingly.
31873 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
31875 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31877 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
31878 nonlinear iv_use computation in loop invariant sensitive way.
31880 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31882 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
31883 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31885 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31887 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31888 (dump_cand): Support iv_cand.inv_exprs.
31889 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31891 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31894 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31896 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31898 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31899 as local function. Include necessary header files.
31900 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31902 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31904 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31906 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31908 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31909 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31910 RSHIFT_EXPR and BIT_NOT_EXPR.
31912 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31914 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31915 (adjust_setup_cost): New parameter supporting round up adjustment.
31916 (struct address_cost_data): Delete.
31917 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31918 (split_address_cost, ptr_difference_cost): Delete.
31919 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31920 (struct ainc_cost_data): New struct.
31921 (get_address_cost_ainc): New function.
31922 (get_address_cost, get_computation_cost): Reimplement.
31923 (determine_group_iv_cost_address): Record inv_expr for all uses of
31925 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31926 (iv_ca_has_deps): Reimplemented to ...
31927 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31929 (iv_ca_extend): Call iv_ca_more_deps.
31931 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31933 * tree-ssa-address.c (struct mem_address): Move to header file.
31934 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31935 * tree-ssa-address.h (struct mem_address): Move from C file.
31936 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31938 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31940 * tree-affine.h (aff_combination_type): New interface.
31941 (aff_combination_zero_p): Remove static.
31942 (aff_combination_const_p): New interface.
31943 (aff_combination_singleton_var_p): New interfaces.
31945 2017-05-11 Richard Biener <rguenther@suse.de>
31947 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31948 Skip unreachable blocks and destinations.
31949 (eliminate): Move stmt removal and fixup ...
31950 (fini_eliminate): ... here. Skip inserted exprs.
31951 (pass_pre::execute): Move fini_pre after fini_eliminate.
31952 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31953 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31954 PRE to get rid of dead code that has invalid SSA form and
31955 split critical edges again.
31957 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31959 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31961 2017-05-11 Richard Biener <rguenther@suse.de>
31963 * passes.c (execute_function_todo): Verify loops if they are
31964 said to be up-to-date.
31965 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31966 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31968 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31971 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31972 handle calling assemble_external ourself.
31975 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31976 modes with zero size. Enhance comment.
31978 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31980 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31981 built-ins for vec_xl and vec_xst with short and char pointer
31984 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31986 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31987 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31988 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31989 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31990 (_mm_maskz_min_round_ss): New intrinsics.
31991 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31992 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31993 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31994 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31995 (__builtin_ia32_minss_mask_round): New builtins.
31996 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31997 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31998 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
32000 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
32001 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
32003 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
32006 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32008 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32009 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32010 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32011 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32012 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32013 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32014 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32015 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32016 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32017 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32018 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32019 (__builtin_ia32_mulss_mask_round): New builtins.
32020 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32021 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32022 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32024 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32025 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32027 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32030 2017-05-10 Julia Koval <julia.koval@intel.com>
32032 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32033 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32034 (_mm256_setr_m128i): New intrinsics.
32036 2017-05-10 Julia Koval <julia.koval@intel.com>
32038 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32039 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32040 (_mm_maskz_rcp14_ss): New intrinsics.
32041 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32042 (__builtin_ia32_rcp14ss_mask): New builtins.
32043 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32045 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32047 PR tree-optimization/51513
32048 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32049 (assert_unreachable_fallthru_edge_p): Use it.
32050 (group_case_labels_stmt): Likewise.
32051 * tree-cfg.h: Prototype it.
32052 * stmt.c: Include cfghooks.h and tree-cfg.h.
32053 (emit_case_dispatch_table) <gap_label>: New local variable.
32054 Use it to fill dispatch table gaps.
32055 Test for default_label before updating probabilities.
32056 (expand_case) <default_label>: Remove unneeded initialization.
32057 Test for unreachable default case statement and remove its edge.
32058 Set default_label accordingly.
32059 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32061 2017-05-10 Carl Love <cel@us.ibm.com>
32063 * config/rs6000/rs6000-c: Add support for built-in functions
32064 vector signed char vec_neg (vector signed char)
32065 vector signed short int vec_neg (vector short int)
32066 vector signed int vec_neg (vector signed int)
32067 vector signed long long vec_neg (vector signed long long)
32068 vector float vec_neg (vector float)
32069 vector double vec_neg (vector double)
32070 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32072 * config/rs6000/altivec.h: Add define for vec_neg
32073 * doc/extend.texi: Update the built-in documentation for the
32074 new built-in functions.
32076 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32078 PR tree-optimization/77644
32079 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32081 2017-05-10 Nathan Sidwell <nathan@acm.org>
32083 * dumpfile.h (TDI_lang_all): New.
32084 (TDF_KIND): New. Renumber others
32085 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32087 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32089 (get_dump_file_name): Adjust suffix generation.
32090 (dump_enable_all): Use TDF_KIND.
32091 * doc/invoke.texi (-fdump-lang-all): Document.
32093 * dumpfile.h: Tabify.
32095 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32098 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32099 Move member access before delete.
32101 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32103 * tree-inline.c (expand_call_inline): Split block at stmt
32106 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32109 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32110 are now unused after splitting mov{sf,sd}_hardfloat.
32111 (f32_lr2): Likewise.
32112 (f32_lm): Likewise.
32113 (f32_lm2): Likewise.
32114 (f32_li): Likewise.
32115 (f32_li2): Likewise.
32116 (f32_lv): Likewise.
32117 (f32_sr): Likewise.
32118 (f32_sr2): Likewise.
32119 (f32_sm): Likewise.
32120 (f32_sm2): Likewise.
32121 (f32_si): Likewise.
32122 (f32_si2): Likewise.
32123 (f32_sv): Likewise.
32124 (f32_dm): Likewise.
32125 (f32_vsx): Likewise.
32126 (f32_av): Likewise.
32127 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32128 For movsf, order stores so the VSX stores occur before the GPR
32129 store which encourages the register allocator to use a traditional
32130 FPR instead of a GPR. For movsd, order the stores so that the GPR
32131 store comes before the VSX stores to allow the power6 to work.
32132 This is due to the power6 not having a 32-bit integer store
32133 instruction from a FPR.
32134 (movsf_hardfloat): Likewise.
32135 (movsd_hardfloat): Likewise.
32137 2017-05-09 Martin Sebor <msebor@redhat.com>
32139 PR translation/80280
32140 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32143 PR translation/80280
32144 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32145 data member added in r247778.
32146 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32148 2017-05-09 Nathan Sidwell <nathan@acm.org>
32150 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32152 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32155 2017-05-09 Marek Polacek <polacek@redhat.com>
32157 * doc/invoke.texi: Fix typo.
32159 2017-05-09 Richard Biener <rguenther@suse.de>
32161 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32162 (vrp_val_is_min): Likewise.
32163 (set_value_range_to_value): Likewise.
32164 (set_value_range_to_nonnegative): Likewise.
32165 (gimple_assign_nonzero_p): Likewise.
32166 (gimple_stmt_nonzero_p): Likewise.
32167 (vrp_int_const_binop): Likewise. Remove unreachable case.
32168 (adjust_range_with_scev): Adjust comments.
32169 (compare_range_with_value): Likewise.
32170 (extract_range_from_phi_node): Likewise.
32171 (test_for_singularity): Likewise.
32173 2017-05-09 Richard Biener <rguenther@suse.de>
32175 * tree-vrp.c (get_single_symbol): Add assert that we don't
32176 get overflowed constants as invariant part.
32177 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32178 checks. Use wi::cmp instead of recursing for integer constants.
32179 (compare_values): Just ignore whether we assumed undefined
32180 overflow instead of failing the compare.
32181 (extract_range_for_var_from_comparison_expr): Add comment before the
32182 TREE_NO_WARNING sets.
32183 (test_for_singularity): Likewise.
32184 (extract_range_from_comparison): Do not disable optimization
32185 when we assumed undefined overflow.
32186 (extract_range_basic): Remove init of unused var.
32188 2017-05-09 Richard Biener <rguenther@suse.de>
32190 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32191 (extract_range_from_multiplicative_op_1): Adjust.
32192 (extract_range_from_binary_expr_1): Use int_const_binop.
32194 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32197 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32198 rs6000_store_data_bypass_p in seven define_bypass directives and
32199 in several comments.
32200 * config/rs6000/rs6000-protos.h: Add prototype for
32201 rs6000_store_data_bypass_p function.
32202 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32203 function implements slightly different (rs6000-specific) semantics
32204 than store_data_bypass_p, returning false rather than aborting
32205 with assertion error when arguments do not satisfy the
32206 requirements of store data bypass.
32207 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32208 rs6000_store_data_bypass_p.
32210 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32212 * config/xtensa/xtensa-protos.h
32213 (xtensa_initial_elimination_offset): New declaration.
32214 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32215 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32216 macro definition, add case for FRAME_POINTER_REGNUM when
32217 FRAME_GROWS_DOWNWARD.
32218 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32219 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32220 xtensa_initial_elimination_offset.
32222 2017-05-08 Nathan Sidwell <nathan@acm.org>
32224 * doc/invoke.texi: Alphabetize -fdump options.
32226 2017-05-08 Martin Sebor <msebor@redhat.com>
32228 PR translation/80280
32229 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32231 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32233 * target.def (compute_frame_layout): New optional target hook.
32234 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32235 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32236 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32238 * reload1.c (verify_initial_elim_offsets): Likewise.
32239 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32240 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32241 (arm_get_frame_offsets): Split up into this ...
32242 (arm_compute_frame_layout): ... and this function.
32244 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32246 * config/aarch64/constraints.md (Usa): New constraint.
32247 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32249 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32251 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32252 with_multilib_list after it has been checked.
32254 2017-05-08 Richard Biener <rguenther@suse.de>
32256 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32257 (bitmap_set_subtract_values): Likewise.
32259 2017-05-08 Richard Biener <rguenther@suse.de>
32261 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32262 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32264 (gimple_stmt_nonzero_warnv_p): Rename to ...
32265 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32267 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32268 (extract_range_basic): Adjust, do not disable propagation on
32269 strict overflow sensitive simplification.
32270 (vrp_visit_cond_stmt): Likewise.
32272 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32274 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32275 body size unconditionally.
32277 2017-05-07 Jeff Law <law@redhat.com>
32280 2017-05-06 Jeff Law <law@redhat.com>
32281 PR tree-optimization/78496
32282 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32285 PR tree-optimization/78496
32286 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32287 (simplify_stmt_using_ranges): Call it.
32288 (vrp_dom_walker::before_dom_children): Extract equivalences
32289 from an ASSERT_EXPR with an equality comparison against a
32292 2017-05-06 Jeff Law <law@redhat.com>
32294 PR tree-optimization/78496
32295 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32298 PR tree-optimization/78496
32299 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32300 (simplify_stmt_using_ranges): Call it.
32301 (vrp_dom_walker::before_dom_children): Extract equivalences
32302 from an ASSERT_EXPR with an equality comparison against a
32305 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32307 * lra-constraints.c (lra_copy_reg_equiv): New function.
32308 (split_reg): Use it to copy equivalence information from the
32309 original register to the spill register.
32311 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32313 PR rtl-optimization/75964
32314 * simplify-rtx.c (simplify_const_relational_operation): Remove
32315 invalid handling of comparisons of integer ABS.
32317 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32319 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32320 initialize to zero.
32321 (init_regs): Remove declaration.
32322 (function_arg_advance_32): Initialize error_p as boolean variable.
32324 2017-05-05 Nathan Sidwell <nathan@acm.org>
32326 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32327 lines. Use for (;;).
32329 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32331 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32332 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32333 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32334 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32335 VF=2 that require versioning.
32337 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32339 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32342 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32344 * diagnostic.h (diagnostic_override_option_index): Convert from
32345 macro to inline function.
32347 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32349 * diagnostic.c (last_module_changed_p): New function.
32350 (set_last_module): New function.
32351 (diagnostic_report_current_module): Convert macro usage to
32352 the above functions.
32353 * diagnostic.h (diagnostic_context::last_module): Strengthen
32354 from const line_map * to const line_map_ordinary *.
32355 (diagnostic_last_module_changed): Delete macro.
32356 (diagnostic_set_last_module): Delete macro.
32358 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32360 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32361 with diagnostic_report_diagnostic.
32362 (diagnostic_n_impl_richloc): Likewise.
32363 * diagnostic.h (report_diagnostic): Delete macro.
32364 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32365 with diagnostic_report_diagnostic.
32366 * substring-locations.c (format_warning_va): Likewise.
32368 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32370 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32371 save/restor of format_spec. Move option-printing code to...
32372 (print_option_information): ...this new function, and
32373 reimplement by simply printing to the pretty_printer,
32374 rather than appending to the format string.
32376 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32378 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32379 handling logic into...
32380 (update_effective_level_from_pragmas): ...this new function.
32382 2017-05-04 Andrew Waterman <andrew@sifive.com>
32384 * config/riscv/riscv.opt (mstrict-align): New option.
32385 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32386 (SLOW_UNALIGNED_ACCESS): Define.
32387 (riscv_slow_unaligned_access): Declare.
32388 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32390 (riscv_slow_unaligned_access): New variable.
32391 (rocket_tune_info): Set slow_unaligned_access to true.
32392 (optimize_size_tune_info): Set slow_unaligned_access to false.
32393 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32394 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32395 (riscv_option_override): Set riscv_slow_unaligned_access.
32396 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32398 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32400 * config/riscv/riscv.md: Unify indentation.
32402 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32407 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32409 (extendsi<mode>2): Add support for doing sign extension via
32410 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32411 don't have ISA 3.0 instructions.
32412 (extendsi<mode>2 splitter): Likewise.
32413 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32414 generate the normal insns since SImode can now go in vector
32415 registers. Disallow the special UNSPECs needed for previous
32416 machines to hide SImode being used. Add new insns
32417 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32418 (fix_trunc<mode>si2_stfiwx): Likewise.
32419 (fix_trunc<mode>si2_internal): Likewise.
32420 (fixuns_trunc<mode>si2): Likewise.
32421 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32422 (fctiw<u>z_<mode>_smallint): Likewise.
32423 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32424 of floating point to 32-bit integer from doing a direct move to
32425 the GPR registers to do a store.
32426 (fctiwz_<mode>): Break long line.
32428 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32430 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32431 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32432 (addr_list, addr_offset_valid_p): New.
32433 (split_address_groups): Check offset validity with above function.
32434 (gt-tree-ssa-loop-ivopts.h): Include header file.
32436 2017-05-05 Nathan Sidwell <nathan@acm.org>
32438 * config.gcc (arm*-*-*): Add missing 'fi'.
32440 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32442 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32443 included in -fopt-info does not matter.
32444 * doc/optinfo.texi (-fopt-info): Fix description of default
32445 behavour. Explicitly say order of options included in -fopt-info
32448 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32450 * config.gcc: Allow combinations of aprofile and rmprofile values for
32451 --with-multilib-list.
32452 * config/arm/t-multilib: New file.
32453 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32454 variables. Remove setting of ISA and floating-point ABI in
32455 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32456 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32457 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32458 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32460 * config/arm/t-rmprofile: Likewise except for the matches changes.
32461 * doc/install.texi (--with-multilib-list): Document the combination of
32462 aprofile and rmprofile values and warn about pitfalls in doing that.
32464 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32466 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32467 (movdi_aarch64): Likewise.
32469 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32471 PR tree-optimization/80632
32472 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32474 (build_arrays): Initialize it for virtual phis.
32475 (fix_phi_nodes): Use it for virtual phis.
32477 PR tree-optimization/80558
32478 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32479 [x, y] op z into [x op, y op z] for op & or | if conditions
32482 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32483 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32486 * config/arm/arm.md (use_literal_pool): Remove.
32487 (64-bit immediate split): No longer takes cost into consideration
32488 if arm_disable_literal_pool is enabled.
32489 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32490 used when arm_disable_literal_pool is enabled.
32491 (arm_max_const_double_inline_cost): Remove use of
32492 arm_disable_literal_pool.
32493 (push_minipool_fix): Add assert.
32494 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32495 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32496 (no_literal_pool_sf_immediate): New.
32498 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32500 PR tree-optimization/80613
32501 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32502 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32504 2017-05-05 Richard Biener <rguenther@suse.de>
32506 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32508 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32510 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32511 of this flag from insn conditions due to removal from r247495.
32513 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32515 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32517 (arm_early_store_addr_dep_ptr): Likewise.
32518 * config/arm/aarch-common-protos.h
32519 (arm_early_load_addr_dep_ptr): Add prototype.
32520 (arm_early_store_addr_dep_ptr): Likewise.
32521 * config/arm/cortex-a53.md: Add new bypasses.
32523 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32525 * tree.c (next_type_uid): Change type to unsigned.
32526 (type_hash_canon): Decrement back next_type_uid if
32527 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32528 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32531 2017-05-04 Martin Sebor <msebor@redhat.com>
32533 * builtins.c: Fix a trivial typo in a comment.
32535 PR middle-end/79234
32536 * builtins.c (check_sizes): Adjust to handle reading past the end.
32537 Avoid printing excessive upper bound of ranges. Use %E to print
32538 tree nodes instead of converting them to %wu.
32539 (expand_builtin_memchr): New function.
32540 (compute_dest_size): Rename...
32541 (compute_objsize): ...to this.
32542 (expand_builtin_memcpy): Adjust.
32543 (expand_builtin_mempcpy): Adjust.
32544 (expand_builtin_strcat): Adjust.
32545 (expand_builtin_strcpy): Adjust.
32546 (check_strncat_sizes): Adjust.
32547 (expand_builtin_strncat): Adjust.
32548 (expand_builtin_strncpy): Adjust and simplify.
32549 (expand_builtin_memset): Adjust.
32550 (expand_builtin_bzero): Adjust.
32551 (expand_builtin_memcmp): Adjust.
32552 (expand_builtin): Handle memcmp.
32553 (maybe_emit_chk_warning): Check strncat just once.
32555 2017-05-04 Martin Sebor <msebor@redhat.com>
32557 PR preprocessor/79214
32558 PR middle-end/79222
32559 PR middle-end/79223
32560 * builtins.c (check_sizes): Add inlining context and issue
32561 warnings even when -Wno-system-headers is set.
32562 (check_strncat_sizes): Same.
32563 (expand_builtin_strncat): Same.
32564 (expand_builtin_memmove): New function.
32565 (expand_builtin_stpncpy): Same.
32566 (expand_builtin): Handle memmove and stpncpy.
32568 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32570 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32571 which is not used any more.
32573 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32575 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32577 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32579 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32580 (cortexa53_tunings): Likewise.
32581 (cortexa57_tunings): Likewise.
32582 (cortexa72_tunings): Likewise.
32583 (cortexa73_tunings): Likewise.
32585 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32587 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32588 Set loop alignment to 8.
32590 2017-05-04 Martin Sebor <msebor@redhat.com>
32592 PR translation/80280
32593 * builtins.c (expand_builtin_object_size): Add missing quoting to
32594 %D and like directives.
32595 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32596 (hsa_type_for_tree_type): Same.
32597 (verify_function_arguments): Same.
32598 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32599 * varasm.c (get_section): Same.
32602 2017-05-04 Martin Sebor <msebor@redhat.com>
32604 PR translation/80280
32605 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32607 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32609 * config/aarch64/aarch64.c (generic_addrcost_table):
32610 Change HI/TI mode setting.
32612 2017-05-04 Martin Jambor <mjambor@suse.cz>
32614 PR tree-optimization/80622
32615 * tree-sra.c (comes_initialized_p): New function.
32616 (build_accesses_from_assign): Only set write lazily when
32617 comes_initialized_p is false.
32618 (analyze_access_subtree): Use comes_initialized_p.
32619 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32620 instead of testing for PARM_DECL.
32622 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32624 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32625 constraint on operand 0 to allow more general addressing modes.
32626 Adjust output template.
32627 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32629 * config/aarch64/aarch64-protos.h
32630 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32631 * config/aarch64/constraints.md (Dp): New address constraint.
32632 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32635 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32637 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32638 update use of estimate_ipcp_clone_size_and_time.
32639 (estimate_local_effects): Update use of
32640 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32641 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32642 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32643 Return nonspecialized time.
32645 2017-05-04 Richard Biener <rguenther@suse.de>
32647 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32648 for the last VUSE which def dominates the PHI. Directly call
32650 (get_continuation_for_phi_1): Remove.
32652 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32654 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32655 to explain the use of truncating division. Cap the number of
32656 iterations to the maximum given by nb_iterations_upper_bound,
32659 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32661 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32662 * configure: Regenerate.
32663 * config.in: Regenerate.
32664 * config/i386/driver-mingw32.c: new file.
32665 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32666 * config.host: Link driver-mingw32.o on MinGW host.
32667 * doc/install.texi: Document new --enable-mingw-wildcard configure
32670 2017-05-04 Marek Polacek <polacek@redhat.com>
32672 PR tree-optimization/80612
32673 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32675 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32676 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32678 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32679 (movt splitter): Likewise.
32680 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32681 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32682 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32683 block for Thumb-1 with MOVT.
32684 (thumb2_legitimate_address_p): Move code block ...
32685 (can_avoid_literal_pool_for_label_p): ... into this new function.
32686 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32688 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32689 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32690 "M-profile targets with the MOVT instruction".
32692 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32694 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32695 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32696 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32698 2017-05-04 Martin Liska <mliska@suse.cz>
32700 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32701 variable cond_code.
32703 2017-05-04 Richard Biener <rguenther@suse.de>
32705 * tree.c (array_at_struct_end_p): Handle arrays at struct
32706 end with flexarrays more conservatively. Refactor and treat
32707 arrays of arrays or aggregates more strict. Fix
32708 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32709 * tree.c (array_at_struct_end_p): Adjust prototype.
32710 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32711 * gimple-fold.c (get_range_strlen): Likewise.
32712 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32714 2017-05-04 Richard Biener <rguenther@suse.de>
32716 PR tree-optimization/31130
32717 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32719 (supports_overflow_infinity): Likewise.
32720 (is_negative_overflow_infinity): Likewise.
32721 (is_positive_overflow_infinity): Likewise.
32722 (is_overflow_infinity): Likewise.
32723 (stmt_overflow_infinity): Likewise.
32724 (overflow_infinity_range_p): Likewise.
32725 (usable_range_p): Remove as always returning true.
32726 (make_overflow_infinity): Remove.
32727 (negative_overflow_infinity): Likewise.
32728 (positive_overflow_infinity): Likewise.
32729 (avoid_overflow_infinity): Likewise.
32730 (set_value_range): Adjust accordingly.
32731 (set_value_range_to_nonnegative): Likewise, remove now unused
32732 overflow_infinity arg.
32733 (vrp_operand_equal_p): Adjust.
32734 (update_value_range): Likewise.
32735 (range_int_cst_singleton_p): Likewise.
32736 (operand_less_p): Likewise.
32737 (compare_values_warnv): Likewise.
32738 (extract_range_for_var_from_comparison_expr): Likewise.
32739 (vrp_int_const_binop): Likewise.
32740 (zero_nonzero_bits_from_vr): Likewise.
32741 (extract_range_from_multiplicative_op_1): Likewise.
32742 (extract_range_from_binary_expr_1): Likewise.
32743 (extract_range_from_unary_expr): Likewise.
32744 (extract_range_from_comparison): Likewise.
32745 (extract_range_basic): Likewise.
32746 (adjust_range_with_scev): Likewise.
32747 (compare_ranges): Likewise.
32748 (compare_range_with_value): Likewise.
32749 (dump_value_range): Likewise.
32750 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32752 (simplify_cond_using_ranges): Adjust.
32754 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32756 * brig-builtins.def: Added a builtin for class_f64.
32757 * builtin-types.def: Added a builtin type needed by class_f64.
32759 2017-05-03 Jason Merrill <jason@redhat.com>
32761 * timevar.def: Add TV_CONSTEXPR.
32763 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32765 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32767 2017-05-03 Martin Jambor <mjambor@suse.cz>
32769 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32770 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32771 * ipa-cp.c (ipcp_propagate_stage): Do not call
32772 ipa_update_after_lto_read.
32773 * ipa-inline.c (ipa_inline): Likewise.
32775 2017-05-03 Martin Jambor <mjambor@suse.cz>
32777 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32778 tag. Added a default constructor and a destructor.
32779 (ipa_edge_args_sum_t): New class;
32780 (ipa_edge_args_sum): Declare.
32781 (ipa_edge_args_vector): Remove declaration.
32782 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32783 (ipa_free_edge_args_substructures): Remove declaration.
32784 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32785 (ipa_edge_args_info_available_for_edge_p): Likewise.
32786 * ipa-prop.c (ipa_edge_args_vector): Removed.
32787 (edge_removal_hook_holder): Likewise.
32788 (edge_duplication_hook_holder): Likewise.
32789 (ipa_edge_args_sum): New variable.
32790 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32791 ipa_edge_args_vector.
32792 (ipa_free_edge_args_substructures): Likewise.
32793 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32794 ipa_edge_args_vector.
32795 (ipa_edge_removal_hook): Turned into method
32796 ipa_edge_args_sum_t::remove.
32797 (ipa_edge_duplication_hook): Turned into method
32798 ipa_edge_args_sum_t::duplicate.
32799 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
32800 registering edge hooks.
32801 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
32802 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
32803 ipa_edge_args_sum instead of ipa_edge_args_vector.
32804 * ipa-profile.c (ipa_profile): Likewise.
32806 2017-05-03 Martin Jambor <mjambor@suse.cz>
32808 * symbol-summary.h (function_summary): New method exists.
32809 (function_summary::symtab_removal): Deallocate through release.
32810 (call_summary): New class.
32811 (gt_ggc_mx): New overload.
32812 (gt_pch_nx): Likewise.
32813 (gt_pch_nx): Likewise.
32815 2017-05-03 Jeff Law <law@redhat.com>
32817 PR tree-optimization/78496
32818 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
32819 from simplify_cond_using_ranges. Split off code to walk
32820 backwards through casts into ...
32821 (simplify_cond_using_ranges_2): New function.
32822 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
32823 (execute_vrp): After identifying jump threads, call
32824 simplify_cond_using_ranges_2.
32826 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32829 * ipa-inline.h (inline_summary): Add ctor.
32830 (create_ggc): Do not use ggc_cleared_alloc.
32832 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32833 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32835 * gcc.c (handle_braces): Support escaping in switch matching
32837 * doc/invoke.texi (Spec Files): Document it.
32838 Remove superfluous @code markup in items.
32840 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32842 * diagnostic-show-locus.c (struct column_range): New struct.
32843 (get_affected_columns): New function.
32844 (get_printed_columns): New function.
32845 (struct correction): New struct.
32846 (correction::ensure_capacity): New function.
32847 (correction::ensure_terminated): New function.
32848 (struct line_corrections): New struct.
32849 (line_corrections::~line_corrections): New dtor.
32850 (line_corrections::add_hint): New function.
32851 (layout::print_trailing_fixits): Reimplement in terms of the new
32853 (selftest::test_overlapped_fixit_printing): New function.
32854 (selftest::diagnostic_show_locus_c_tests): Call it.
32856 2017-05-03 Nathan Sidwell <nathan@acm.org>
32858 Canonicalize canonical type hashing
32859 * tree.h (type_hash_canon_hash): Declare.
32860 * tree.c (type_hash_list, attribute_hash_list): Move into
32861 type_hash_canon_hash.
32862 (build_type_attribute_qual_variant): Break out hash code calc into
32863 type_hash_canon_hash.
32864 (type_hash_canon_hash): New. Generic type hash computation.
32865 (build_range_type_1, build_array_type_1, build_function_type,
32866 build_method_type_directly, build_offset_type, build_complex_type,
32867 make_vector_type): Call it.
32869 2017-05-03 Richard Biener <rguenther@suse.de>
32871 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
32872 When all DRs have unknown misaligned do not always peel
32873 when there is a store but apply the same costing model as if
32874 there were only loads.
32876 2017-05-03 Richard Biener <rguenther@suse.de>
32879 PR tree-optimization/80492
32880 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32881 compare_base_decls returning dont-know properly.
32883 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32885 * config/arm/iterators.md (CCSI): New mode iterator.
32886 (arch): New mode attribute.
32887 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32888 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32889 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32890 code iterator for success result mode.
32891 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32892 the corresponding new insn generators.
32894 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32897 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32898 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32900 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32902 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32903 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32904 (DDR_A): Wrap DDR argument in brackets.
32905 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32906 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32907 (DDR_REVERSED_P): Likewise.
32909 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32911 PR tree-optimization/79472
32912 * tree-switch-conversion.c (struct switch_conv_info): Add
32913 contiguous_range and default_case_nonstandard fields.
32914 (collect_switch_conv_info): Compute contiguous_range and
32915 default_case_nonstandard fields, don't clear final_bb if
32916 contiguous_range and only the default case doesn't have the required
32918 (check_all_empty_except_final): Set default_case_nonstandard instead
32919 of failing if contiguous_range and the default case doesn't have empty
32921 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32922 and only the default case doesn't have the required constants. Skip
32924 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32925 if default_case_nonstandard.
32926 (build_constructors): Build constant 1 just once. Assert that default
32927 values aren't inserted in between cases if contiguous_range. Skip
32929 (build_arrays): Skip virtual phis.
32930 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32931 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32932 Handle virtual phis.
32933 (gen_inbound_check): Handle default_case_nonstandard case.
32934 (process_switch): Adjust check_final_bb caller. Call
32935 gather_default_values with the first non-default case instead of
32936 default case if default_case_nonstandard.
32938 2017-05-02 Nathan Sidwell <nathan@acm.org>
32940 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32941 check. Fix formatting.
32943 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32945 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32946 errors when comparing specialized and unspecialized times.
32948 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32950 * diagnostic-show-locus.c
32951 (layout::should_print_annotation_line_p): Make private.
32952 (layout::print_annotation_line): Make private.
32953 (layout::annotation_line_showed_range_p): Make private.
32954 (layout::show_ruler): Make private.
32955 (layout::print_source_line): Make private. Pass in line and
32956 line_width, rather than calling location_get_source_line. Drop
32958 (layout::print_leading_fixits): New method.
32959 (layout::print_any_fixits): Rename to...
32960 (layout::print_trailing_fixits): ...this, and make private.
32961 Don't print newline fixits.
32962 (diagnostic_show_locus): Move logic for printing one row into...
32963 (layout::print_line): ...this new function. Move the
32964 location_get_source_line call and error-handling from
32965 print_source_line to here. Call print_leading_fixits, and rename
32966 print_any_fixits to print_trailing_fixits.
32967 (selftest::test_fixit_insert_containing_newline): Update now that
32968 newlines are partially supported.
32969 (selftest::test_fixit_insert_containing_newline_2): New test.
32970 (selftest::test_fixit_replace_containing_newline): Update comments.
32971 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32972 * edit-context.c (class added_line): New class.
32973 (class edited_line): Describe newline handling in comment.
32974 (edited_line::actually_edited_p): New method.
32975 (edited_line::print_content): Delete redundant decl.
32976 (edited_line::m_predecessors): New field.
32977 (edited_file::print_content): Call edited_line::print_content.
32978 (edited_file::print_diff): Update to support newlines.
32979 (edited_file::print_diff_hunk): Likewise.
32980 (edited_file::print_run_of_changed_lines): New function.
32981 (edited_file::print_diff_line): Convert to...
32982 (print_diff_line): ...this.
32983 (edited_file::get_effective_line_count): New function.
32984 (edited_line::edited_line): Initialize new field m_predecessors.
32985 (edited_line::~edited_line): Clean up m_predecessors.
32986 (edited_line::apply_fixit): Handle newlines.
32987 (edited_line::get_effective_line_count): New function.
32988 (edited_line::print_content): New function.
32989 (edited_line::print_diff_lines): New function.
32990 (selftest::test_applying_fixits_insert_containing_newline): New
32992 (selftest::test_applying_fixits_replace_containing_newline): New
32994 (selftest::insert_line): New function.
32995 (selftest::test_applying_fixits_multiple_lines): Add example of
32997 (selftest::edit_context_c_tests): Call the new tests.
32999 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33001 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
33002 parameter cand. Update dump information.
33003 (get_computation_cost): Update uses.
33005 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33007 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
33008 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33009 (get_computation_at, rewrite_use_address): Update use of
33010 get_computation_aff.
33012 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33014 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33015 (get_computation): Delete.
33016 (get_computation_cost): Implement like get_computation_cost_at.
33017 Use get_computation_at.
33018 (get_computation_cost_at): Delete.
33019 (rewrite_use_nonlinear_expr): Use get_computation_at.
33020 (rewrite_use_compare, remove_unused_ivs): Ditto.
33022 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33024 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33026 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33028 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33029 (ivopts_global_cost_for_size): Rename parameter and update uses.
33030 (iv_ca_recount_cost): Update uses.
33031 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33032 candidates seperately in n_invs and n_cands.
33033 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33035 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33037 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33038 (find_inv_vars_cb): New.
33039 (find_depends): Renamed to ...
33040 (find_inv_vars): ... this.
33041 (add_candidate_1, force_var_cost): Call find_inv_vars.
33042 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33044 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33046 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33047 inv_vars. Add inv_exprs.
33048 (struct iv_cand): Rename depends_on to inv_vars.
33049 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33050 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33051 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33052 (dump_cand): Dump inv_vars.
33053 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33054 (record_invariant, find_depends, add_candidate_1): Ditto.
33055 (set_group_iv_cost, force_var_cost): Ditto.
33056 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33057 (get_computation_cost_at, get_computation_cost): Ditto.
33058 (determine_group_iv_cost_generic): Ditto.
33059 (determine_group_iv_cost_address): Ditto.
33060 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33061 (determine_group_iv_costs): Ditto.
33062 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33063 (iv_ca_set_remove_invariants): Renamed to ...
33064 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33065 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33066 (iv_ca_set_add_invariants): Renamed to ...
33067 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33068 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33069 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33070 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33071 (create_new_ivs): Remove useless dump.
33073 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33075 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33077 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33078 (iv_ca_set_no_cp, create_new_iv): Ditto.
33080 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33082 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33084 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33086 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33087 function tree_check2.
33089 2017-05-02 Martin Liska <mliska@suse.cz>
33091 * doc/gcov.texi: Add missing preposition.
33092 * gcov.c (function_info::function_info): Properly fill up
33093 all member variables.
33095 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33097 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33099 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33101 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33103 2017-05-02 Martin Liska <mliska@suse.cz>
33106 * lto-streamer-in.c (lto_read_tree_1): Remove
33107 LTO_STREAMER_DEBUG.
33108 * lto-streamer.c (struct tree_hash_entry): Likewise.
33109 (struct tree_entry_hasher): Likewise.
33110 (tree_entry_hasher::hash): Likewise.
33111 (tree_entry_hasher::equal): Likewise.
33112 (lto_streamer_init): Likewise.
33113 (lto_orig_address_map): Likewise.
33114 (lto_orig_address_get): Likewise.
33115 (lto_orig_address_remove): Likewise.
33116 * lto-streamer.h: Likewise.
33117 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33118 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33120 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33122 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33123 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33124 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33125 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33126 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33127 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33128 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33129 (mm_maskz_sub_ss): New intrinsics.
33130 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33131 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33132 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33133 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33134 (__builtin_ia32_subss_mask_round): New builtins.
33135 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33136 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33137 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33139 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33140 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33142 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33145 2017-05-02 Martin Jambor <mjambor@suse.cz>
33147 PR tree-optimization/78687
33148 * tree-sra.c (access): New field parent.
33149 (process_subtree_disqualification): New function.
33150 (disqualify_candidate): Call it.
33151 (build_accesses_from_assign): Reset write flag if creating an
33153 (build_access_subtree): Fill in parent field and also prpagate
33154 down grp_write flag.
33155 (create_artificial_child_access): New parameter set_grp_write, set
33156 grp_write to its value.
33157 (propagate_subaccesses_across_link): Also propagate grp_write flag
33159 (propagate_all_subaccesses): Push the closest parent back to work
33160 queue if add_access_to_work_queue returned true.
33162 2017-05-02 Richard Biener <rguenther@suse.de>
33164 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33165 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33166 -fstrict-overflow documentation.
33167 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33168 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33169 flag_strict_overflow.
33170 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33171 * lto-opts.c (lto_write_options): Do not stream it.
33172 * lto-wrapper.c (merge_and_complain): Do not handle it.
33173 * opts.c (default_options_table): Do not set -fstrict-overflow.
33174 (finish_options): Likewise do not clear it when sanitizing.
33175 * simplify-rtx.c (simplify_const_relational_operation): Do not
33176 test flag_strict_overflow.
33178 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33180 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33181 using enabled attribute.
33182 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33183 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33184 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33185 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33186 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33187 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33188 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33189 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33190 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33191 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33193 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33195 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33197 2017-05-02 Richard Biener <rguenther@suse.de>
33199 PR tree-optimization/80591
33201 2017-04-10 Richard Biener <rguenther@suse.de>
33203 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33206 2017-05-02 Richard Biener <rguenther@suse.de>
33208 PR tree-optimization/80549
33209 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33210 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33211 headers if they do not have a preheader.
33213 2017-05-02 Martin Liska <mliska@suse.cz>
33216 * common.opt: Fix typo.
33217 * doc/invoke.texi: Likewise.
33219 2017-05-01 Jan Beulich <jbeulich@suse.com>
33221 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33222 swapping, add (x,x,m,x,n) alternative.
33224 2017-05-01 Nathan Sidwell <nathan@acm.org>
33226 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33227 unnecessary unadjusted_alignment check.
33229 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33232 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33234 * gimplify.c (gimplify_cilk_detach): New function.
33235 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33236 * tree-core.h: Document EXPR_CILK_SPAWN.
33237 * tree.h (EXPR_CILK_SPAWN): Define.
33239 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33241 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33242 to use new fixit_hint representation, using the "replace" logic.
33243 (get_line_span_for_fixit_hint): Likewise.
33244 (layout::print_any_fixits): Likewise.
33245 (selftest::test_one_liner_many_fixits): Rename to...
33246 (selftest::test_one_liner_many_fixits_1): ...this, and update
33247 comment and expected output to reflect that the multiple fix-it
33248 hints are now consolidated into one insertion.
33249 (selftest::test_one_liner_many_fixits_2): New test.
33250 (selftest::test_diagnostic_show_locus_one_liner): Update for
33252 (selftest::test_fixit_consolidation): Update for fix-it API
33254 * diagnostic.c (print_parseable_fixits): Likewise.
33255 * edit-context.c (edited_line::m_line_events): Convert from
33256 auto_vec <line_event *> to auto_vec <line_event>.
33257 (class line_event): Convert from abstract base class to a concrete
33258 class, taking over the role of replace_event.
33259 (class insert_event): Delete.
33260 (class replace_event): Rename to class line_event. Convert to
33262 (edit_context::add_fixits): Reimplement.
33263 (edit_context::apply_insert): Delete.
33264 (edit_context::apply_replace): Rename to...
33265 (edit_context::apply_fixit): ...this. Convert to half-open range.
33266 (edited_file::apply_insert): Delete.
33267 (edited_file::apply_replace): Rename to...
33268 (edited_file::apply_fixit): ...this.
33269 (edited_line::~edited_line): Drop deletion of events.
33270 (edited_line::apply_insert): Delete.
33271 (edited_line::apply_replace): Rename to...
33272 (edited_line::apply_fixit): ...this. Convert to half-open range.
33273 Update for change to type of m_line_events.
33274 * edit-context.h (edit_context::apply_insert): Delete.
33275 (edit_context::apply_replace): Rename to...
33276 (edit_context::apply_fixit): ...this.
33278 2017-05-01 Martin Sebor <msebor@redhat.com>
33280 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33283 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33286 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33287 __get_cpuid_max returns 0.
33288 (__get_cpuid_count): Ditto.
33290 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33292 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33293 replacement expression is another instance of one of its arguments.
33295 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33298 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33299 check for stack push/pop autoinc.
33300 * config/i386/i386.c (ix86_agi_dependent): Return false
33301 if the only reason why modified_in_p returned true is that
33302 addr is SP based and set_insn is a push or pop.
33304 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33306 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33309 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33312 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33313 (account_size_time): Use two predicates - exec_pred and
33315 (evaluate_conditions_for_known_args): Compute both clause and
33317 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33318 (inline_summary_t::duplicate): Update.
33319 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33321 (compute_inline_parameters): Likewise.
33322 (estimate_edge_size_and_time): Update caluclation of time.
33323 (estimate_node_size_and_time): Compute both time and nonspecialized
33325 (estimate_ipcp_clone_size_and_time): Update.
33326 (inline_merge_summary): Update.
33327 (do_estimate_edge_time): Update.
33328 (do_estimate_edge_size): Update.
33329 (do_estimate_edge_hints): Update.
33330 (inline_read_section, inline_write_summary): Stream both new predicates.
33331 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33333 (compute_inlined_call_time): Cleanup.
33334 (big_speedup_p): Update.
33335 (edge_badness): Update.
33336 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33337 (size_time_entry): Replace predicate by exec_predicate and
33338 nonconst_predicate.
33339 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33340 (estimate_edge_time): Return also nonspec_time.
33341 (reset_edge_growth_cache): Update.
33343 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33345 PR rtl-optimization/80491
33346 * ifcvt.c (noce_process_if_block): When looking for x setter
33347 with missing else_bb, don't check only the insn right before
33348 cond_earliest, but look for the last insn that x is modified in
33349 within the same bb.
33351 PR rtl-optimization/80491
33352 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33354 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33356 PR tree-optimization/80487
33357 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33359 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33361 PR tree-optimization/79697
33362 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33363 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33364 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33366 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33367 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33369 2017-04-28 Martin Sebor <msebor@redhat.com>
33371 PR tree-optimization/80523
33372 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33373 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33375 (maybe_warn, format_directive, parse_directive): Use new functions.
33376 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33378 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33380 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33382 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33384 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33385 target_header_dir): Set correctly.
33386 * configure: Regenerated.
33387 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33388 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33389 instead of SYSTEM_HEADER_DIR.
33391 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33393 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33394 (estimate_local_effects): Likewise.
33395 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33396 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33397 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33398 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33399 do_estimate_edge_time, estimate_edge_time): Likewise.
33400 * ipa-inline-analysis.c (estimate_node_size_and_time,
33401 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33402 (estimate_time_after_inlining): Remove.
33404 2017-04-28 Martin Liska <mliska@suse.cz>
33406 * doc/gcov.texi: Enhance documentation of gcov.
33408 2017-04-28 Martin Liska <mliska@suse.cz>
33410 * doc/gcov.texi: Sort options in alphabetic order.
33411 * doc/gcov-dump.texi: Likewise.
33412 * doc/gcov-tool.texi: Likewise.
33413 * gcov.c (print_usage): Likewise.
33414 * gcov-dump.c (print_usage): Likewise.
33415 * gcov-tool.c (print_merge_usage_message): Likewise.
33416 (print_rewrite_usage_message): Likewise.
33417 (print_overlap_usage_message): Likewise.
33419 2017-04-28 Martin Liska <mliska@suse.cz>
33421 PR gcov-profile/53915
33422 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33424 2017-04-28 Martin Liska <mliska@suse.cz>
33426 PR gcov-profile/79891
33427 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33428 is marked by compiler as living on a line.
33429 (get_cycles_count): Remove usage of the union.
33430 (output_intermediate_file): Likewise.
33431 (find_source): Fix GNU coding style.
33432 (accumulate_line_counts): Remove old non-all block mode.
33433 (output_lines): Remove usage of the union.
33434 * profile.c (output_location): Include all BBs, even if
33435 belonging to a same line (and file) as a previous BB.
33437 2017-04-28 Martin Liska <mliska@suse.cz>
33439 * gcov.c (process_args): Handle new argument 'w'.
33440 (read_graph_file): Assign ID to BBs.
33441 (output_branch_count): Display BB # if verbose flag is set.
33442 (output_lines): Likewise for arcs.
33443 (print_usage): Add '--verbose' option help.
33444 * doc/gcov.texi: Document --verbose (-w) option.
33446 2017-04-28 Martin Liska <mliska@suse.cz>
33448 * gcov.c (struct block_location_info): New struct.
33449 (process_file): Fill up the new structure.
33450 (read_graph_file): Replace usage of encoding by the newly added
33452 (add_line_counts): Likewise.
33453 (accumulate_line_counts): Remove usage of the union.
33454 (function_info::function_info): New function.
33455 (function_info::~function_info): Likewise.
33456 (process_file): Call delete instead of release_function.
33457 (release_function): Release the function.
33458 (release_structures): Call delete instead of release_function.
33459 (solve_flow_graph): Replace usage of num_blocks.
33460 (find_exception_blocks): Likewise.
33461 (output_lines): Fix GNU coding style.
33463 2017-04-28 Martin Liska <mliska@suse.cz>
33466 * coverage.c (coverage_remove_note_file): New function.
33467 * coverage.h: Declare the function.
33468 * toplev.c (finalize): Clean if an error has been seen.
33470 2017-04-28 Martin Liska <mliska@suse.cz>
33472 PR gcov-profile/80031
33473 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33474 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33475 * gcov.c (read_graph_file): Read just number of blocks.
33476 * profile.c (branch_prob): Do not stream 0 flags per a basic
33479 2017-04-28 Martin Liska <mliska@suse.cz>
33481 * gcov-dump.c (tag_*): Add new argument to declarations.
33482 (dump_gcov_file): Likewise.
33483 (tag_blocks): Add and use new argument depth.
33484 (tag_arcs): Likewise.
33485 (tag_lines): Likewise.
33486 (tag_counters): Likewise.
33487 (tag_summary): Likewise.
33488 (dump_working_sets): Use depth to do a proper indentation.
33490 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33493 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33494 * symtab.c (symtab_node::debug_symtab): Move definition here.
33496 2017-04-28 Richard Biener <rguenther@suse.de>
33498 * lto-streamer.h (LTO_major_version): Bump to 7.
33500 2017-04-28 Richard Biener <rguenther@suse.de>
33502 * tree-vrp.c (assert_info): New struct.
33503 (add_assert_info): New helper.
33504 (register_edge_assert_for_2): Refactor to add asserts to a vector
33506 (register_edge_assert_for_1): Likewise.
33507 (register_edge_assert_for): Likewise.
33508 (finish_register_edge_assert_for): New helper actually registering
33509 asserts where live on edge.
33510 (find_conditional_asserts): Adjust.
33511 (find_switch_asserts): Likewise.
33512 (evrp_dom_walker::try_find_new_range): Generalize.
33513 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33515 2017-04-27 Marek Polacek <polacek@redhat.com>
33518 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33519 arg10 and arg11 to itype.
33521 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33523 * doc/extend.texi (Object Size Checking): Improve grammar.
33525 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33528 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33529 that the logic for permitting reciprocal estimates matches that
33532 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33535 * tree.c (type_cache_hasher::equal): Only compare
33536 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33537 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33538 non-aggregate element types.
33539 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33540 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33542 2017-04-27 Richard Biener <rguenther@suse.de>
33544 PR middle-end/80533
33545 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33546 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33547 keeping a reference to a trailing array.
33549 2017-04-27 Richard Biener <rguenther@suse.de>
33551 PR middle-end/80539
33552 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33553 being in loop-closed SSA form conservatively.
33554 (chrec_fold_multiply_poly_poly): Likewise.
33556 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33558 PR middle-end/79665
33559 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33560 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33562 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33565 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33566 (aarch64_function_arg_alignment): Return unsigned int again, but still
33567 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33568 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33569 Don't emit -Wpsabi note.
33570 (aarch64_function_arg_boundary): Likewise.
33571 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33574 2017-04-26 Nathan Sidwell <nathan@acm.org>
33576 * tree.h (crc32_unsigned_n): Declare.
33577 (crc32_unsigned, crc32_unsigned): Make inline.
33578 * tree.c (crc32_unsigned_bits): Replace with ...
33579 (crc32_unsigned_n): ... this.
33580 (crc32_unsigned, crc32_byte): Remove.
33581 (crc32_string): Remove unnecessary braces.
33583 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33585 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33586 * ipa-inline-analysis.c (MAX_TIME): Remove.
33587 (account_size_time): Use sreal for time.
33588 (dump_inline_summary): Update.
33589 (estimate_function_body_sizes): Update.
33590 (estimate_edge_size_and_time): Update.
33591 (estimate_calls_size_and_time): Update.
33592 (estimate_node_size_and_time): Update.
33593 (inline_merge_summary): Update.
33594 (inline_update_overall_summary): Update.
33595 (estimate_time_after_inlining): Update.
33596 (inline_read_section): Update.
33597 (inline_write_summary): Update.
33598 * ipa-inline.c (compute_uninlined_call_time): Update.
33599 (compute_inlined_call_time): Update.
33600 (recursive_inlining): Update.
33601 (inline_small_functions): Update.
33602 (dump_overall_stats): Update.
33603 * ipa-inline.h: Include sreal.h.
33604 (size_time_entry): Turn time to sreal.
33605 (inline_summary): Turn self_time nad time to sreal.
33607 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33609 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33611 (sreal::stream_out, sreal::stream_in): New.
33612 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33614 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33616 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33619 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33622 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33624 (dimode_scalar_chain::compute_convert_gain): Ditto.
33625 (dimode_scalar_chain::make_vector_copies): Ditto.
33626 (dimode_scalar_chain::convert_reg): Ditto.
33627 (dimode_scalar_chain::convert_insn): Ditto.
33628 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33629 (VI248_AVX512BW_1): New mode iterator.
33630 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33631 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33634 2017-04-25 Martin Sebor <msebor@redhat.com>
33636 PR tree-optimization/80497
33637 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33638 constants are representable in HOST_WIDE_INT.
33639 (parse_directive): Ditto.
33641 2017-04-25 Martin Sebor <msebor@redhat.com>
33644 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33645 (new_zero_array): Adjust signature.
33646 (dom_info::dom_init): Used unsigned rather that size_t.
33647 (dom_info::dom_info): Same.
33649 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33650 Jakub Jelinek <jakub@redhat.com>
33653 * config/arm/arm.c: Include gimple.h.
33654 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33655 returns negative, increment ncrn only if it returned positive.
33656 (arm_needs_doubleword_align): Return int instead of bool,
33657 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33658 members, but if there is any such non-FIELD_DECL
33659 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33660 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33661 returns negative, increment nregs only if it returned positive.
33662 (arm_setup_incoming_varargs): Likewise.
33663 (arm_function_arg_boundary): Emit -Wpsabi note if
33664 arm_needs_doubleword_align returns negative, return
33665 DOUBLEWORD_ALIGNMENT only if it returned positive.
33667 2017-04-25 Marek Polacek <polacek@redhat.com>
33670 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33671 first argument to type.
33673 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33676 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33677 type checks to test for compatibility instead of equality.
33679 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33680 Jakub Jelinek <jakub@redhat.com>
33683 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33685 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33686 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33687 the alignment computation, but return their maximum in warn_alignment.
33688 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33689 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33691 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33692 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33695 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33697 * config/arc/simdext.md (dmpyh): Fix typo.
33699 2017-04-25 Richard Biener <rguenther@suse.de>
33701 PR tree-optimization/80492
33702 * alias.c (compare_base_decls): Handle registers with asm
33703 specification conservatively.
33704 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33705 compare_base_decls returning dont-know properly.
33707 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33709 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33710 (legitimate_offset_address_p): New function.
33711 (arc_legitimate_address_p): Use above function.
33713 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33715 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33717 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33719 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33720 ACCH registers whenever they are available.
33722 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33724 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33725 double regs fix when not used.
33727 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33729 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33731 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33732 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33734 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33736 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33737 when emitting short ADD instructions.
33739 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33741 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33743 (cmpsi_cc_c_insn): Likewise.
33744 (cbranchsi4_scratch): Compute proper instruction length using
33745 compact_hreg_operand.
33746 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33748 2017-04-25 Richard Biener <rguenther@suse.de>
33750 PR middle-end/80509
33751 * passes.c (pass_manager::pass_manager): Initialize
33752 m_name_to_pass_map.
33754 2017-04-25 Richard Biener <rguenther@suse.de>
33756 PR tree-optimization/79201
33757 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33759 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33762 * config/s390/vector.md: Split MEM->GPR vector moves for
33763 non-s_operand addresses.
33765 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33768 * config/s390/predicates.md (reload_const_wide_int_operand): New
33770 * config/s390/s390.md ("movti"): Remove d/P alternative.
33771 ("movti_bigconst"): New pattern definition.
33773 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33776 * s390-protos.h (s390_expand_cs_hqi): Removed.
33777 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33778 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33779 modes as well as CCZ1mode and CCZmode.
33780 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33781 signature of s390_emit_compare_and_swap.
33782 (s390_expand_cs_hqi): Likewise, make static.
33783 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33784 compare-and-swap, in some cases.
33785 (s390_expand_cs): Wrapper function.
33786 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33788 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33789 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33790 patterns for small and large integers. Forbid symref memory operands.
33791 Move expander to s390.c. Require cc register.
33792 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33793 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33794 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33795 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33796 symref memory operands. Remove CC mode and call s390_match_ccmode
33798 ("atomic_exchange<mode>"): Allow and implement all integer modes.
33800 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33802 * config/s390/s390.md (define_peephole2): New peephole to help
33803 combining the load-and-test pattern with volatile memory.
33805 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33807 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
33808 with CCZmode for TARGET_Z196.
33810 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33812 PR rtl-optimization/80501
33813 * combine.c (make_compound_operation_int): Set subreg_code to SET
33814 even for AND with mask of the sign bit of mode.
33816 PR rtl-optimization/80500
33817 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
33818 sum's initial value.
33820 2017-04-25 Julian Brown <julian@codesourcery.com>
33821 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33823 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
33825 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33827 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33829 2017-04-25 Julian Brown <julian@codesourcery.com>
33830 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33832 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33833 (thunderx2t99_sha): New Reservation.
33835 2017-04-25 Julian Brown <julian@codesourcery.com>
33836 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33838 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33839 type for 1-element load.
33841 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33843 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33845 2017-04-24 Martin Jambor <mjambor@suse.cz>
33847 PR tree-optimization/80293
33848 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33849 char arrays not totally scalarizable if it is false.
33850 (analyze_all_variable_accesses): Pass correct value in the new
33851 parameter. Add a statistics counter.
33853 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
33855 PR middle-end/79931
33856 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
33858 2017-04-24 Richard Biener <rguenther@suse.de>
33860 PR tree-optimization/80494
33861 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33862 out for complex types.
33864 2017-04-24 Richard Biener <rguenther@suse.de>
33866 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
33867 * tree-ssa-sccvn.c (print_scc): Print SCC size.
33868 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
33869 (DFS): Adjust and never fail.
33870 (sccvn_dom_walker::fail): Remove.
33871 (sccvn_dom_walker::before_dom_children): Adjust.
33872 (run_scc_vn): Likewise and never fail.
33873 * tree-ssa-pre.c (pass_pre::execute): Adjust.
33874 (pass_fre::execute): Likewise.
33876 2017-04-24 Richard Biener <rguenther@suse.de>
33878 PR tree-optimization/79725
33879 * tree-ssa-sink.c (statement_sink_location): Return whether
33880 failure reason was zero uses. Move that check later.
33881 (sink_code_in_bb): Deal with zero uses by removing the stmt
33884 2017-04-24 Richard Biener <rguenther@suse.de>
33887 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33888 pointer-based references.
33890 2017-04-24 Richard Biener <rguenther@suse.de>
33893 * pass_manager.h (pass_manager::operator new): Remove.
33894 (pass_manager::operator delete): Likewise.
33895 * passes.c (pass_manager::operator new): Remove.
33896 (pass_manager::operator delete): Likewise.
33897 (pass_manager::pass_manager): Zero individual pass members.
33899 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33902 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33903 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33904 Check "XEXP (src, 1)" operand here.
33905 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33906 Check "XEXP (src, 1)" operand here.
33907 (dimode_scalar_chain::make_vector_copies): Detect count register
33908 of a shift instruction. Zero extend count register from QImode
33909 to DImode to satisfy vector shift pattern count operand predicate.
33910 Substitute vector shift count operand with a DImode copy.
33911 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33914 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33916 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33917 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33918 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33919 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33920 (UNSPEC_NOREX_MEM): Remove definition.
33922 2017-04-21 Richard Biener <rguenther@suse.de>
33924 PR tree-optimization/79547
33925 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33926 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33927 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33928 without any constraints.
33930 2017-04-21 Richard Biener <rguenther@suse.de>
33932 PR tree-optimization/78847
33933 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33935 2017-04-21 Richard Biener <rguenther@suse.de>
33937 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33938 (build_distinct_type_copy): Likewise.
33939 (build_variant_type_copy): Likewise.
33940 * tree.c (build_qualified_type): Pass down mem-stat info.
33941 (build_distinct_type_copy): Likewise.
33942 (build_variant_type_copy): Likewise.
33944 2017-04-21 Richard Biener <rguenther@suse.de>
33946 PR tree-optimization/80237
33947 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33949 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33950 for a simplified result.
33952 2016-04-21 Richard Biener <rguenther@suse.de>
33954 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33955 sth as strict as a simple_iv but a chrec without symbols and an
33956 operand defined in the loop we are peeling (and not some subloop).
33957 (propagate_constants_for_unrolling): Propagate all constants.
33959 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33962 * config/i386/i386.c (print_reg): Remove assert for disalowed
33963 regno values, call output_operand_lossage instead.
33965 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33968 * config/i386/constraints.md (Yc): New register constraint.
33969 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33970 Use Yc constraint for alternative 2 of operand 0. Remove
33971 preferred_for_speed attribute.
33973 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33975 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33976 lastprivate clauses in SIMT case.
33978 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33980 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33982 2017-04-20 Richard Biener <rguenther@suse.de>
33984 PR tree-optimization/57796
33985 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33986 as N scalar stores.
33987 (vect_model_load_cost): Cost gathers as N scalar loads.
33989 2017-04-20 Richard Biener <rguenther@suse.de>
33991 * ggc-page.c (ggc_allocated_p): Rename to ...
33992 (safe_lookup_page_table_entry): ... this and return the lookup
33994 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33996 2017-04-20 Richard Biener <rguenther@suse.de>
33998 PR tree-optimization/80453
33999 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
34000 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
34001 from the conditions.
34002 (vn_phi_eq): Pass them down.
34003 (vn_phi_lookup): Record them.
34004 (vn_phi_insert): Likewise.
34006 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
34008 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34009 uninitialized variable warning to avoid buffer overrun.
34011 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34014 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34015 is suppressed for '{ 0 }' in C.
34017 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34019 * BASE-VER: Set to 8.0.0.
34021 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34023 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34024 priority .init_array and .fini_array section with SECTION_NOTYPE
34027 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34029 PR middle-end/80423
34030 * tree.h (build_array_type): Add typeless_storage default argument.
34031 * tree.c (type_cache_hasher::equal): Also compare
34032 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34033 (build_array_type): Add typeless_storage argument, set
34034 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34036 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34037 (build_array_type): Likewise. Add typeless_storage argument.
34039 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34040 Jakub Jelinek <jakub@redhat.com>
34042 PR tree-optimization/80426
34043 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34044 operation on symbolic operands, also compute the overflow for the
34045 invariant part when the operation degenerates into a negation.
34047 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34050 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34051 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34054 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34056 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34059 * config/avr/avr.c (tree.h): Include it.
34060 (cgraph.h): Include it.
34061 (avr_encode_section_info): Don't warn for uninitialized progmem
34062 variable if it's just an alias.
34064 2017-04-19 Richard Biener <rguenther@suse.de>
34067 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34068 when needed by AutoPGO.
34070 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34073 * doc/lto.texi: Remove an extra 'that'.
34075 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34077 PR rtl-optimization/80429
34078 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34079 are only used in debug insns.
34081 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34082 Vladimir Makarov <vmakarov@redhat.com>
34084 * config/sparc/predicates.md (input_operand): Add comment. Return
34085 true for any memory operand when LRA is in progress.
34086 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34088 2017-04-18 Jeff Law <law@redhat.com>
34091 * mips.md ({return,simple_return}_internal): Do not overwrite
34094 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34096 PR tree-optimization/80443
34097 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34098 instead of adding 1, subtract -1 and similarly instead of subtracting
34101 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34103 PR rtl-optimization/80357
34104 * haifa-sched.c (tmp_bitmap): New variable.
34105 (model_recompute): Handle duplicate use records.
34106 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34107 (free_global_sched_pressure_data): Free it.
34109 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34112 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34113 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34114 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34115 instead of SYSTEM_HEADER_DIR.
34117 2017-04-18 Jeff Law <law@redhat.com>
34119 PR middle-end/80422
34120 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34121 predecessors after walking up the insn chain.
34123 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34126 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34127 sizetype type into debug info.
34129 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34132 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34133 unneeded test for TARGET_UPPER_REGS_SF.
34134 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34136 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34139 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34140 instead of gsi_after_labels.
34142 2017-04-18 Jeff Law <law@redhat.com>
34144 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34148 2017-04-13 Jeff Law <law@redhat.com>
34149 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34150 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34152 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34155 * config/avr/avr.c (intl.h): Include it.
34156 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34158 2017-04-18 Martin Liska <mliska@suse.cz>
34160 PR gcov-profile/78783
34161 * gcov-tool.c (gcov_output_files): Validate that destination
34162 file is either removed by the tool or by a user.
34164 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34165 Guy Benyei <guybe@mellanox.com>
34167 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34168 block, and do not negate it, the stored id is already negative.
34170 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34172 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34174 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34177 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34178 masks of options that should be turned off if the VSX vector
34179 options are turned off.
34180 (OTHER_P8_VECTOR_MASKS): Likewise.
34181 (OTHER_VSX_VECTOR_MASKS): Likewise.
34182 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34183 rs6000_disable_incompatible_switches to validate no type switches
34185 (rs6000_incompatible_switch): New function to disallow turning on
34186 other vector options if -mno-vsx, -mno-power8-vector, or
34187 -mno-power9-vector are specified.
34189 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34191 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34193 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34195 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34196 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34197 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34198 (ARG_POINTER_CFA_OFFSET): Likewise.
34200 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34202 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34203 conditions to take advantage of various optimizations.
34205 2017-04-13 Jeff Law <law@redhat.com>
34207 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34208 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34209 (zero_extendsidi2_dext): Likewise.
34211 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34214 * fold-const.c (fold_ternary_loc): Revert
34215 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34218 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34220 PR rtl-optimization/80343
34221 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34222 new scratch pseudo.
34224 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34227 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34228 to ubsan_encode_value.
34230 2017-04-13 Jeff Law <law@redhat.com>
34232 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34233 appearing in DEBUG_INSNs.
34235 2017-04-13 Martin Liska <mliska@suse.cz>
34237 PR gcov-profile/80413
34238 * gcov-io.c (gcov_write_string): Copy to buffer just when
34239 allocated size is greater than zero.
34241 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34244 * dwarf2out.c (decls_for_scope): Ignore declarations of
34245 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34247 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34250 * ipa-visibility.c (non_local_p): Fix typos.
34251 (localize_node): When localizing symbol in same comdat group,
34252 dissolve the group only when we know external symbols are going
34254 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34256 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34258 PR tree-optimization/79390
34259 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34260 order does not result in usable sequence, retry with reversed operand
34266 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34267 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34268 op0 instead of fold_convert_loc (loc, type, arg0).
34270 2017-04-12 Jeff Law <law@redhat.com>
34272 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34273 has a delay slot in the generated code.
34275 * config/cris/cris.md (cris_preferred_reload_class): Return
34276 GENNONACR_REGS rather than GENERAL_REGS.
34278 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34281 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34282 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34283 signedness of the result type.
34285 2017-04-12 Richard Biener <rguenther@suse.de>
34286 Jeff Law <law@redhat.com>
34288 PR tree-optimization/80359
34289 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34290 trim stores to TARGET_MEM_REFs.
34292 2017-04-12 Richard Biener <rguenther@suse.de>
34294 PR tree-optimization/79390
34295 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34296 threading case even more.
34298 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34301 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34302 for quad_address_p for TImode, instead of just not indexed_address.
34304 2017-04-12 Richard Biener <rguenther@suse.de>
34305 Bernd Edlinger <bernd.edlinger@hotmail.de>
34307 PR middle-end/79671
34308 * alias.c (component_uses_parent_alias_set_from): Handle
34309 TYPE_TYPELESS_STORAGE.
34310 (get_alias_set): Likewise.
34311 * tree-core.h (tree_type_common): Add typeless_storage flag.
34312 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34313 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34314 for types containing members with TYPE_TYPELESS_STORAGE.
34315 (place_field): Likewise.
34316 (layout_type): Likewise for ARRAY_TYPE.
34317 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34318 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34319 TYPE_TYPELESS_STORAGE.
34320 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34322 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34325 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34326 first argument to type.
34328 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34332 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34333 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34334 (rs6000_expand_binop_builtin): Likewise.
34335 (rs6000_expand_ternop_builtin): Likewise; also add missing
34336 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34337 vshasigma built-ins.
34338 * doc/extend.texi: Document that vec_xxpermdi's third argument
34339 must be a constant.
34341 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34343 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34344 Use shift_const cost parameter when calculating gain of STV shifts.
34346 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34348 PR rtl-optimization/70478
34349 * lra-constraints.c (process_alt_operands): Check memory for
34350 disfavoring memory insn operand.
34352 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34354 PR middle-end/80100
34355 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34356 left shift in unsigned HOST_WIDE_INT type.
34358 PR rtl-optimization/80385
34359 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34360 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34363 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34364 if they have any depend clauses.
34366 2017-04-11 Martin Liska <mliska@suse.cz>
34369 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34370 * ipa-split.c (split_function): Create a local comdat symbol
34371 if caller is in a comdat group.
34373 2017-04-11 Martin Liska <mliska@suse.cz>
34376 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34379 2017-04-11 Martin Sebor <msebor@redhat.com>
34381 PR middle-end/80364
34382 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34383 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34385 (directive::set_width, directive::set_precision, format_character):
34387 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34390 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34393 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34394 conflict, set target->arch_name instead of target->cpu_name.
34396 2017-04-11 Richard Biener <rguenther@suse.de>
34398 PR tree-optimization/80374
34399 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34400 build_zero_cst, remove fold_convertible_p check again.
34402 2017-04-11 Martin Liska <mliska@suse.cz>
34405 * ubsan.c (instrument_object_size): Do not instrument register
34408 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34411 * config/i386/i386-builtin-types.def
34412 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34413 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34414 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34415 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34416 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34417 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34418 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34419 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34420 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34421 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34422 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34423 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34424 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34425 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34426 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34427 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34428 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34429 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34430 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34431 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34432 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34433 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34434 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34435 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34436 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34437 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34438 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34439 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34440 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34441 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34442 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34443 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34444 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34445 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34446 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34447 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34448 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34449 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34450 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34451 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34452 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34453 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34454 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34455 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34456 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34457 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34459 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34460 flag to second_arg_count, handle 4 argument function type _COUNT
34461 aliases, handle second_arg_count on second argument rather than last.
34463 2017-04-10 Jeff Law <law@redhat.com>
34465 PR tree-optimization/80374
34466 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34467 record anything if we can not convert integer_zero_node to the
34470 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34473 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34474 Enhance special handling given to the TARGET_P9_MINMAX option in
34475 relation to certain other options.
34477 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34479 PR tree-optimization/80153
34480 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34481 remove POINTER_PLUS_EXPR's base part directly, rather than through
34484 2017-04-10 Richard Biener <rguenther@suse.de>
34485 Bin Cheng <bin.cheng@arm.com>
34487 PR tree-optimization/80153
34488 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34489 the first element of pointer type aff_tree. Build result expr in
34491 (add_elt_to_tree): Convert to type unconditionally. Remove other
34492 fold_convert calls.
34493 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34494 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34496 2017-04-10 Richard Biener <rguenther@suse.de>
34498 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34501 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34503 PR rtl-optimization/70478
34504 * lra-constraints.c (curr_small_class_check): New.
34505 (update_and_check_small_class_inputs): New.
34506 (process_alt_operands): Update curr_small_class_check. Disfavor
34507 alternative insn memory operands. Check available regs for small
34510 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34513 * config/mips/mips.opt (-mvirt): Update description.
34514 * doc/invoke.texi (-mvirt): Likewise.
34516 2017-04-10 Richard Biener <rguenther@suse.de>
34518 PR middle-end/80362
34519 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34520 looking for NEGATE_EXPR in -A / -B to A / B folding.
34522 2017-04-10 Martin Liska <mliska@suse.cz>
34524 PR gcov-profile/80224
34525 * gcov.c (print_usage): Fix usage string.
34526 (get_gcov_intermediate_filename): Remove.
34527 (output_gcov_file): Use both for normal and intermediate format.
34528 (generate_results): Do not initialize special file for
34529 intermediate format.
34531 2017-04-10 Richard Biener <rguenther@suse.de>
34533 PR tree-optimization/80304
34534 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34537 2017-04-10 Nathan Sidwell <nathan@acm.org>
34540 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34541 (rs6000_init_builtins): Use it.
34543 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34545 * config/arm/arm.md (<mrc>): Add mode to SET source.
34546 (<mrrc>): Likewise.
34548 2017-04-10 Richard Biener <rguenther@suse.de>
34550 PR middle-end/80344
34551 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34553 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34556 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34557 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34558 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34559 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34560 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34561 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34562 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34563 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34564 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34565 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34566 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34567 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34568 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34569 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34570 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34571 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34572 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34573 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34574 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34575 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34576 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34577 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34578 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34580 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34582 PR rtl-optimization/70478
34583 * lra-constraints.c: Reverse the last patch.
34585 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34587 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34588 Add comment for WCHAR_T.
34590 2017-04-08 Martin Liska <mliska@suse.cz>
34593 2017-04-07 Martin Liska <mliska@suse.cz>
34596 * ipa-split.c (split_function): Add function part to a same comdat
34599 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34602 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34604 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34606 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34607 (vec_pairing): ...to this.
34608 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34609 (rs6000_sched_init): Adjust for name change.
34610 (struct rs6000_sched_context): Likewise.
34611 (rs6000_init_sched_context): Likewise.
34612 (rs6000_set_sched_context): Likewise.
34614 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34620 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34622 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34623 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34624 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34626 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34628 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34630 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34632 PR rtl-optimization/70703
34633 * ira-color.c (update_conflict_hard_regno_costs): Use
34634 int64_t instead of HOST_WIDE_INT.
34636 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34638 PR rtl-optimization/70478
34639 * lra-constraints.c (process_alt_operands): Disfavor alternative
34640 insn memory operands.
34642 2017-04-07 Jeff Law <law@redhat.com>
34644 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34645 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34647 2017-04-07 Martin Liska <mliska@suse.cz>
34650 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34651 Show error message instead of an ICE.
34653 2017-04-07 Martin Liska <mliska@suse.cz>
34656 * ipa-split.c (split_function): Add function part to a same comdat
34659 2017-04-07 Richard Biener <rguenther@suse.de>
34661 PR middle-end/80341
34662 * tree.c (get_unwidened): Also handle ! for_type case for
34664 * convert.c (do_narrow): Split out from ...
34665 (convert_to_integer_1): ... here. Do not pass final truncation
34666 type to get_unwidened for TRUNC_DIV_EXPR.
34668 2017-04-07 Richard Biener <rguenther@suse.de>
34670 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34672 (aff_combination_const): Adjust.
34673 (aff_combination_scale): Likewise.
34674 (aff_combination_add_elt): Likewise.
34675 (aff_combination_add_cst): Likewise.
34676 (aff_combination_convert): Likewise.
34677 (add_elt_to_tree): Likewise. Remove unused argument.
34678 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34680 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34682 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34684 * config/arm/arm.c (arm_default_short_enums): Use
34685 ARM_DEFAULT_SHORT_ENUMS.
34686 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34688 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34691 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34692 members with redundant out-of-class redeclaration.
34694 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34697 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34698 * config/i386/i386.md (*zero_extendsidi2):
34699 Add (?*x,*x) and (?*v,*v) alternatives.
34701 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34704 * config/i386/i386.c (ix86_expand_builtin)
34705 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34706 mode from insn data. Convert operands to insn operand mode.
34707 Copy operands that don't satisfy insn predicate to a register.
34709 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34711 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34714 2017-04-06 Richard Biener <rguenther@suse.de>
34716 PR tree-optimization/80334
34717 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34718 preserve alignment of accesses.
34720 2017-04-06 Richard Biener <rguenther@suse.de>
34722 PR tree-optimization/80262
34723 * tree-sra.c (build_ref_for_offset): Preserve address-space
34725 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34726 Drop useless address-space information on MEM_REF offsets.
34728 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34730 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34732 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34734 PR rtl-optimization/70703
34735 * ira-color.c (update_conflict_hard_regno_costs): Use
34736 HOST_WIDE_INT instead of long.
34738 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34741 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34742 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34744 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34745 for x86_64 target. Handle -m3dnowa option.
34747 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34749 PR rtl-optimization/70703
34750 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34751 (update_conflict_hard_regno_costs): Use long instead of unsigned
34752 arithmetic for cost calculation.
34754 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34755 Bernd Edlinger <bernd.edlinger@hotmail.de>
34758 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34761 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34764 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34765 ptr_mode with Pmode throughout.
34766 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34767 into probe_stack_range and use DImode.
34769 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34772 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34773 call_eh_return is true.
34775 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34777 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34778 Initialize last_match_fntype_index.
34780 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34783 * tree-nvr.c: Include internal-fn.h.
34784 (pass_return_slot::execute): Ignore internal calls without
34787 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34788 Richard Biener <rguenther@suse.de>
34791 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34792 captures used multiple times, except for the last use.
34793 * generic-match-head.c: Include gimplify.h.
34795 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34797 PR tree-optimization/79390
34798 * target.h (struct noce_if_info): Declare.
34799 * targhooks.h (default_noce_conversion_profitable_p): Declare.
34800 * target.def (noce_conversion_profitable_p): New target hook.
34801 * ifcvt.h (struct noce_if_info): New type, moved from ...
34802 * ifcvt.c (struct noce_if_info): ... here.
34803 (noce_conversion_profitable_p): Renamed to ...
34804 (default_noce_conversion_profitable_p): ... this. No longer
34806 (noce_try_store_flag_constants, noce_try_addcc,
34807 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
34808 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
34809 instead of noce_conversion_profitable_p.
34810 * config/i386/i386.c: Include ifcvt.h.
34811 (ix86_option_override_internal): Don't override
34812 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
34813 (ix86_noce_conversion_profitable_p): New function.
34814 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
34815 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
34816 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
34817 * doc/tm.texi: Regenerated.
34819 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34821 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
34824 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34827 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34828 instructions for small multiply cores.
34830 2017-04-04 Jeff Law <law@redhat.com>
34832 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34834 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34835 that are not set by the loop over the elements.
34837 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34840 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34841 int mode, convert_modes it to mode as unsigned, otherwise use
34842 lowpart_subreg to mode rather than SImode.
34843 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34844 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34845 Use DImode instead of SImode for the shift count operand.
34846 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34849 2017-04-04 Richard Biener <rguenther@suse.de>
34851 PR middle-end/80281
34852 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
34853 arithmetic done for the negate or the plus. Simplify.
34854 (A - (-B) -> A + B): Likewise.
34855 * fold-const.c (split_tree): Make sure to not negate pointers.
34857 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
34859 PR rtl-optimization/60818
34860 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34861 a compare of comparisons with the thing compared if this results
34862 in a different machine mode.
34864 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
34866 * alias.c (base_alias_check): Fix typo in comment.
34867 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
34868 * cgraphunit.c (symbol_table::compile): Likewise.
34869 * collect2.c (maybe_run_lto_and_relink): Likewise.
34870 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
34871 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
34872 * config/avr/avr.c (avr_map_op_t): Likewise.
34873 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
34874 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
34875 * config/epiphany/epiphany.md (movcc): Likewise.
34876 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
34877 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
34879 * config/mips/mips.c (mips_save_restore_reg): Likewise.
34880 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
34881 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
34882 * config/sh/sh.c (sh_rtx_costs): Likewise.
34883 * fold-const.c (fold_truth_andor): Likewise.
34884 * genautomata.c (collapse_flag): Likewise.
34885 * gengtype.h (struct type::u::s): Likewise.
34886 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34887 * input.c (FORMAT_AMOUNT): Likewise.
34888 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34889 (known_aggs_to_agg_replacement_list): Likewise.
34890 * ipa-inline-analysis.c: Likewise.
34891 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34892 * ipa-polymorphic-call.c
34893 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34894 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34895 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34897 * modulo-sched.c (apply_reg_moves): Likewise.
34898 * omp-expand.c (build_omp_regions_1): Likewise.
34899 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34900 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34901 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34902 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34903 * value-prof.c: Likewise.
34904 * var-tracking.c (val_reset): Likewise.
34906 2017-04-03 Richard Biener <rguenther@suse.de>
34908 PR tree-optimization/80275
34909 * fold-const.c (split_address_to_core_and_offset): Handle
34912 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34914 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34915 descriptors is at least equal to that of functions.
34917 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34919 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34921 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34924 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34925 (mov<IMOD4:mode>): New expander.
34926 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34928 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34930 PR rtl-optimization/79405
34931 * fwprop.c (propagations_left): New variable.
34932 (forward_propagate_into): Decrement it.
34933 (fwprop_init): Initialize it.
34934 (fw_prop): If the variable has reached zero, stop propagating.
34935 (fwprop_addr): Ditto.
34937 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34940 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34941 a FUNCTION_DECL, pass it as decl instead of origin to
34944 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34946 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34949 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34952 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34953 TARGET_VSX_SMALL_INTEGER.
34955 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34957 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34958 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34960 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34962 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34963 extraction from odd-numbered MSA register.
34965 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34967 PR middle-end/80173
34968 * expmed.c (store_bit_field_1): Don't attempt to create
34969 a word subreg out of hard registers wider than word if they
34970 have HARD_REGNO_NREGS of 1 for their mode.
34972 PR middle-end/80163
34973 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34974 conversions to integer types wider than word and pointer.
34977 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34978 (rtx_equal_for_cselib_p): Pass 0 to it.
34979 * cselib.c (cselib_hasher::equal): Likewise.
34980 (rtx_equal_for_cselib_1): Add depth argument. If depth
34981 is 128, don't look up VALUE locs and punt. Increment
34982 depth in recursive calls when walking VALUE locs.
34984 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34986 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34987 (make_gcov_file_name): Use the canonical path name for generating
34989 (read_line): Fix handling of files with ascii null bytes.
34991 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34993 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34994 to initialise a vector register instead
34995 of using a const_int.
34997 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34999 PR translation/80189
35000 * gimplify.c (omp_default_clause): Use %qs instead of %s in
35001 diagnostic messages.
35003 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
35006 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
35007 (dfp_diex_<mode>): Update mode of operand 1.
35008 * doc/extend.texi (dxex, dxexq): Document change to return type.
35009 (diex, diexq): Document change to argument type.
35011 2017-03-30 Martin Jambor <mjambor@suse.cz>
35014 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35015 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35016 it reflects the signature changes performed at the callee side.
35017 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35018 to cgraph_build_function_type_skip_args.
35019 (build_function_decl_skip_args): Adjust call to the above function.
35021 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35024 * config/i386/sse.md
35025 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35026 register as dest whenever it is a MEM not rtx_equal_p to the
35027 corresponding dup operand, and when forcing into reg move the
35028 reg into the memory afterwards.
35029 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35030 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35031 for the force_reg mode.
35032 (avx512vl_vextractf128<mode>): Use register as dest either
35033 always when a MEM, or when it is a MEM not rtx_equal_p to the
35034 corresponding dup operand, or even not when it is a CONST_VECTOR
35035 depending on the mode and lo vs. hi.
35036 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35038 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35039 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35040 Likewise. Require that operands[2] is even.
35041 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35042 Remove extraneous parens. Require that operands[2] is a multiple
35044 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35045 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35046 disallow memory then.
35048 2017-03-30 Richard Biener <rguenther@suse.de>
35050 PR tree-optimization/77498
35051 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35052 to non-constants over backedges.
35054 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35056 PR rtl-optimization/80233
35057 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35058 as last_combined_insn. Do not test for BARRIER_P separately.
35060 2017-03-29 Andreas Schwab <schwab@suse.de>
35063 * calls.c (prepare_call_address): Convert funexp to Pmode before
35064 copying to temp reg.
35066 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35068 PR tree-optimization/80158
35069 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35070 Handle possible future case of more than one alternate
35072 (replace_rhs_if_not_dup): Likewise.
35073 (replace_one_candidate): Likewise.
35075 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35077 PR rtl-optimization/80193
35078 * ira.c (ira): Do not check allocation for LRA.
35080 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35082 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35083 (nvptx_output_simt_exit): Declare.
35084 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35085 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35086 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35087 (nvptx_declare_function_name): ...here. Emit declaration of local
35088 memory space buffer for omp_simt_enter insn.
35089 (nvptx_output_unisimt_switch): New.
35090 (nvptx_output_softstack_switch): New.
35091 (nvptx_output_simt_enter): New.
35092 (nvptx_output_simt_exit): New.
35093 * config/nvptx/nvptx.h (struct machine_function): New fields
35094 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35095 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35096 (UNSPECV_SIMT_EXIT): Ditto.
35097 (omp_simt_enter_insn): New insn.
35098 (omp_simt_enter): New expansion.
35099 (omp_simt_exit): New insn.
35100 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35102 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35103 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35104 (expand_GOMP_SIMT_EXIT): New.
35105 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35106 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35107 (GOMP_SIMT_EXIT): Ditto.
35108 * target-insns.def (omp_simt_enter): New insn.
35109 (omp_simt_exit): Ditto.
35110 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35112 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35113 (lower_rec_input_clauses): Likewise.
35114 (lower_lastprivate_clauses): Handle SIMT privatization.
35116 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35117 (ompdevlow_adjust_simt_enter): New.
35118 (find_simtpriv_var_op): New.
35119 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35120 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35122 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35123 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35124 (copy_decl_for_dup_finish): Ditto.
35126 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35128 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35131 * config/i386/i386.c (ix86_option_override_internal): Always
35132 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35134 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35136 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35138 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35140 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35141 mark new edge's irreducible flag accordign to it.
35142 (vect_do_peeling): Check loop preheader edge's irreducible flag
35143 and pass it to function slpeel_add_loop_guard.
35145 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35147 PR tree-optimization/80218
35148 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35149 Update block frequencies and counts.
35151 2017-03-28 Richard Biener <rguenther@suse.de>
35153 PR tree-optimization/78644
35154 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35155 of a simplification result we may not use it at all.
35157 2017-03-28 Richard Biener <rguenther@suse.de>
35160 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35161 without arguments, generate default definition of a SSA name.
35163 2017-03-28 Richard Biener <rguenther@suse.de>
35165 PR middle-end/80222
35166 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35167 TYPE_REF_CAN_ALIAS_ALL references.
35168 * fold-const.c (fold_indirect_ref_1): Likewise.
35170 2017-03-28 Martin Liska <mliska@suse.cz>
35173 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35174 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35176 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35177 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35179 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35180 (EXTRA_SPECS): Define.
35181 (SUBTARGET_EXTRA_SPECS): Likewise.
35182 (SUBTARGET_CPP_SPEC): Likewise.
35183 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35184 SUBTARGET_EXTRA_SPECS.
35185 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35187 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35189 * config/arc/simdext.md (vst64_insn): Update pattern.
35190 (vld32wh_insn): Likewise.
35191 (vld32wl_insn): Likewise.
35192 (vld64_insn): Likewise.
35193 (vld32_insn): Likewise.
35195 2017-03-28 Marek Polacek <polacek@redhat.com>
35198 * fold-const.c (fold_comparison): Use protected_set_expr_location
35199 instead of SET_EXPR_LOCATION.
35201 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35203 * tree.c (add_expr): Avoid name lookup warning.
35205 2017-03-27 Jeff Law <law@redhat.com>
35207 PR tree-optimization/80216
35208 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35209 function name. Limit recursion depth.
35210 (record_temporary_equivalences): Corresponding changes.
35212 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35214 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35217 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35220 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35222 * cfgcleanup.c (reg_note_cfa_p): New array.
35223 (insns_have_identical_cfa_notes): New function.
35224 (old_insns_match_p): Don't cross-jump in between /f
35225 and non-/f instructions. If both i1 and i2 are frame related,
35226 verify all CFA notes, their order and content.
35228 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35231 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35232 HImode and SImode with zero extend to DImode to one insn.
35233 (bswap<mode>2_extenddi): Likewise.
35234 (bswapsi2_extenddi): Likewise.
35235 (bswaphi2_extendsi): Likewise.
35236 (bswaphi2): Combine bswap HImode and SImode into one insn.
35237 Separate memory insns from swapping register.
35238 (bswapsi2): Likewise.
35239 (bswap<mode>2): Likewise.
35240 (bswaphi2_internal): Delete, no longer used.
35241 (bswapsi2_internal): Likewise.
35242 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35243 store, and gpr<-gpr swap insns.
35244 (bswap<mode>2_store): Likewise.
35245 (bswaphi2_reg): Register only splitter, combine with the splitter.
35246 (bswaphi2 splitter): Likewise.
35247 (bswapsi2_reg): Likewise.
35248 (bswapsi2 splitter): Likewise.
35249 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35250 the insns into load, store, and register/register insns.
35251 (bswapdi2_ldbrx): Likewise.
35252 (bswapdi2_load): Likewise.
35253 (bswapdi2_store): Likewise.
35254 (bswapdi2_reg): Likewise.
35256 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35258 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35259 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35261 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35264 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35266 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35267 special handling for target option conflicts between dform
35268 options (-mpower9-dform, -mpower9-dform-vector,
35269 -mpower9-dform-scalar) and -mno-direct-move.
35271 2017-03-27 Richard Biener <rguenther@suse.de>
35273 PR tree-optimization/80181
35274 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35276 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35278 * config/arc/predicates.md (move_double_src_operand): Replace the
35279 call to move_double_src_operand with a call to address_operand.
35281 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35283 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35284 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35285 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35287 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35289 * config/arc/predicates.md (long_immediate_loadstore_operand):
35290 Consider scaled addresses cases.
35292 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35294 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35295 restored when in interrupt.
35296 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35297 doesn't have delay slot.
35299 2017-03-27 Richard Biener <rguenther@suse.de>
35302 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35303 inlined thunk clones.
35305 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35308 * asan.c (instrument_derefs): Copy over last operand from
35309 original COMPONENT_REF to the new COMPONENT_REF with
35310 DECL_BIT_FIELD_REPRESENTATIVE.
35311 * ubsan.c (instrument_object_size): Likewise.
35313 2017-03-27 Richard Biener <rguenther@suse.de>
35315 PR tree-optimization/80170
35316 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35317 sure DR/SCEV didnt fold in constants we do not see when looking
35318 at the reference base alignment.
35320 2017-03-27 Richard Biener <rguenther@suse.de>
35322 PR middle-end/80171
35323 * gimple-fold.c (fold_ctor_reference): Properly guard against
35324 NULL return value from canonicalize_constructor_val.
35326 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35329 * config/i386/i386.c (ix86_expand_builtin)
35330 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35331 flags reg setting and flags reg using instructions.
35332 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35333 clobbering instructions to zero extend op2.
35335 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35337 * doc/install.texi (Configuration) <--with-aix-soname>:
35338 Update link to AIX ld.
35340 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35342 PR rtl-optimization/80160
35343 PR rtl-optimization/80159
35344 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35345 reg_alternate_class into account.
35347 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35350 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35351 to consider in curr_insn_transform.
35353 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35355 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35356 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35357 and emit_mode_inner.
35359 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35361 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35362 argument to the overloaded builtin variants. Use the new flag to
35363 deprecate certain builtin variants.
35364 * config/s390/s390-builtin-types.def: Add new builtin types.
35365 * config/s390/s390-builtins.h: Support new flags field for
35366 overloaded builtins.
35367 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35368 (s390_macro_to_expand): Enable vector float data type.
35369 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35370 builtins by incrementing the __VEC__ version number.
35371 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35373 (s390_resolve_overloaded_builtin): Emit error messages depending
35374 on the builtin flags.
35375 * config/s390/s390.c (s390_expand_builtin): Support additional
35376 flags argument. Change error message to match the messages
35377 emitted in s390-c.c.
35378 * config/s390/s390.md: New UNSPEC_* constants.
35379 (op_type): Add new instruction types.
35380 * config/s390/vecintrin.h: Add new builtins and test data class
35382 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35383 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35384 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35385 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35386 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35387 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35389 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35390 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35391 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35392 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35394 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35395 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35396 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35397 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35399 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35400 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35401 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35402 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35403 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35404 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35405 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35407 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35408 ("vec_scatter_element<V_HW_4:mode>_DI")
35409 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35410 ("vec_fpint<mode>", "vflls")
35411 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35412 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35413 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35414 ("*vec_cmphe<mode>_cc"): ... these.
35416 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35417 mode constant instead of magic value.
35419 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35421 * config/s390/s390.c (s390_expand_vec_compare): Support other
35422 vector floating point modes than just V2DF.
35423 (s390_expand_vcond): Likewise.
35424 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35425 (s390_cannot_change_mode_class): Prevent mode changes between TF
35426 and V1TF in vector registers.
35427 * config/s390/s390.md (DF, SF): New mode attributes.
35428 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35429 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35430 SFmode support for VRs.
35431 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35433 (VFT, VF_HW): New mode iterators.
35434 (vw, sdx): New mode attributes.
35435 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35436 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35437 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35438 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35439 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35440 also the new vector floating point modes. Renaming to ...
35442 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35443 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35444 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35445 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35446 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35447 ("vec_unordered<mode>"): ... these.
35449 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35450 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35451 ("*vec_extendv2df"): New insn definitions.
35453 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35455 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35456 ("mulditi3_2", "*muldi3_sign"): New patterns.
35457 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35458 rename the pattern definition.
35460 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35462 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35464 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35466 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35468 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35469 instruction if possible.
35470 * config/s390/vector.md (vec_halfnumelts): New mode
35472 ("*vec_vllezlf<mode>"): New pattern.
35474 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35476 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35477 ("popcountv4si2", "popcountv2di2"): Rename to ...
35478 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35479 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35481 ("popcount<mode>2_vxe"): New pattern.
35483 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35485 * common/config/s390/s390-common.c (processor_flags_table): Add
35487 * config.gcc: Add arch12.
35488 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35489 Default to arch12 for unknown CPU model numbers.
35490 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35491 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35492 PROCESSOR_max sanity check.
35493 * config/s390/s390-opts.h (enum processor_type): Add
35495 * config/s390/s390.c (processor_table): Add arch12.
35496 (s390_expand_builtin): Add check for B_VXE flag.
35497 (s390_issue_rate): Add PROCESSOR_ARCH12.
35498 (s390_get_sched_attrmask): Likewise.
35499 (s390_get_unit_mask): Likewise.
35500 (s390_sched_score): Enable z13 scheduling for arch12.
35501 (s390_sched_reorder): Likewise.
35502 (s390_sched_variable_issue): Likewise.
35503 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35505 (s390_tune_attr): Use z13 scheduling also for arch12.
35506 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35507 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35508 (TARGET_VXE_P): New macros.
35509 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35510 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35511 * config/s390/s390.opt: Add arch12 as processor_type.
35513 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35515 * config/s390/s390.md
35516 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35517 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35518 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35520 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35521 Rename expanders to ...
35523 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35524 ("fixuns_truncdddi2_emu"): ... these.
35526 ("fixuns_trunc<mode>si2_emu"): New expander.
35528 ("*fixuns_truncdfdi2_z13"): Rename to ...
35529 ("*fixuns_truncdfdi2_vx"): ... this.
35531 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35533 * config/s390/2964.md: Remove the single element vector compare
35534 instructions which are no longer used.
35535 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35537 (s390_canonicalize_comparison): Remove handling of DFmode
35539 (s390_expand_vec_compare_scalar): Remove function.
35540 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35541 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35543 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35545 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35547 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35549 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35550 will anyway by matched by mov<mode>_64dfp.
35552 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35554 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35555 vlef/vstef. Add missing operand to vleif.
35557 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35559 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35560 pair for all vector types with 64 bit elements.
35561 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35562 * config/s390/vector.md (V_HW_64): ... here.
35563 (V_128_NOSINGLE): New mode iterator.
35564 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35565 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35566 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35567 ("*vec_load_pairv2di"): Change to ...
35568 ("*vec_load_pair<mode>"): ... this one.
35570 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35572 * config/s390/constraints.md: Add comments.
35573 (jKK): Reject element sizes > 8 bytes.
35574 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35576 * config/s390/s390.md: Add the s_operand checks formerly in
35577 s390_split_ok_p to various splitters where they are still
35579 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35580 for 128 bit vectors. Plus two splitters.
35582 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35584 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35587 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35590 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35591 error if the boundary argument is not constant.
35593 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35595 PR rtl-optimization/80112
35596 * loop-doloop.c (doloop_condition_get): Don't check condition
35597 if cmp isn't SET with IF_THEN_ELSE src.
35599 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35601 PR tree-optimization/80158
35602 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35603 replacing a candidate statement, also replace it for the
35604 candidate's alternate interpretation.
35605 (replace_rhs_if_not_dup): Likewise.
35606 (replace_one_candidate): Likewise.
35608 2017-03-24 Richard Biener <rguenther@suse.de>
35610 PR tree-optimization/80167
35611 * graphite-isl-ast-to-gimple.c
35612 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35614 (translate_isl_ast_to_gimple::get_rename): Likewise.
35616 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35618 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35619 handling of certain combinations of target options, including the
35620 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35621 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35623 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35626 * config/arm/arm.md (*load_multiple): Add reload_completed to
35627 matching condition.
35629 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35630 Richard Biener <rguenth@suse.de>
35632 PR tree-optimization/79908
35633 PR tree-optimization/80136
35634 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35635 been cast away, gimplify_and_add suffices.
35637 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35639 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35641 2017-03-23 Richard Biener <rguenther@suse.de>
35643 PR tree-optimization/80032
35644 * gimplify.c (gimple_push_cleanup): Forced unconditional
35645 cleanups still have to go to the conditional_cleanups
35648 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35650 PR tree-optimization/80072
35651 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35653 (next_operand_entry_id): Change type to unsigned int.
35654 (sort_by_operand_rank): Make sure to return the right return value
35655 even if unsigned fields are bigger than INT_MAX.
35656 (struct oecount): Change cnt and id type to unsigned int.
35657 (oecount_hasher::equal): Formatting fix.
35658 (oecount_cmp): Make sure to return the right return value
35659 even if unsigned fields are bigger than INT_MAX.
35660 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35663 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35664 TREE_READONLY on result if writing it more than once.
35667 * doc/invoke.texi (-fsanitize=thread): Document that with
35668 -fnon-call-exceptions atomics are not able to throw
35672 * tsan.c: Include tree-eh.h.
35673 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35674 maybe_clean_or_replace_eh_stmt where needed.
35675 (instrument_memory_accesses): Add cfg_changed argument.
35676 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35677 if it returned true.
35678 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35680 PR rtl-optimization/63191
35681 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35682 wrapper function, moved the whole old content into ...
35683 (ix86_delegitimize_address_1): ... this. New inline function.
35684 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35685 true as last argument instead of ix86_delegitimize_address.
35687 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35689 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35690 cortexa57_branch_cost.
35692 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35694 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35696 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35699 * doc/md.texi (Constraints): Document wA constraint.
35700 * config/rs6000/constraints.md (wA): New.
35701 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35702 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35703 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35704 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35706 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35709 * gimplify.c (is_oacc_declared): New function.
35710 (oacc_default_clause): Use it to set default flags for acc declared
35711 variables inside parallel regions.
35712 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35713 declared variables.
35714 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35715 declare attribute to any decl as necessary.
35717 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35720 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35721 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35722 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35723 (arm_arch_lpae): This.
35724 * config/arm/arm.c (arm_arch7ve): Rename into ...
35725 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35726 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35729 2017-03-22 Martin Liska <mliska@suse.cz>
35732 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35733 error message instead of an ICE.
35735 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35737 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35739 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35741 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35743 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35746 2017-03-21 Martin Sebor <msebor@redhat.com>
35748 * doc/extend.texi: Use "cannot" instead of "can't."
35749 * doc/hostconfig.texi: Same.
35750 * doc/install.texi: Same.
35751 * doc/invoke.texi: Same.
35752 * doc/loop.texi: Same.
35753 * doc/md.texi: Same.
35754 * doc/objc.texi: Same.
35755 * doc/rtl.texi: Same.
35756 * doc/tm.texi: Same.
35757 * doc/tm.texi.in: Same.
35758 * doc/trouble.texi: Same.
35760 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35763 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35764 (collect_checksum_attributes): Set it.
35765 (die_checksum_ordered): Use it.
35767 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35769 PR tree-optimization/79908
35770 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35771 change: For a VA_ARG whose LHS has been cast away, use
35772 force_gimple_operand to construct the side effects.
35774 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35776 PR translation/80001
35777 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35778 more amenable to translation.
35779 (oacc_loop_auto_partitions): Likewise.
35781 2017-03-21 Marek Polacek <polacek@redhat.com>
35782 Martin Sebor <msebor@redhat.com>
35784 PR tree-optimization/80109
35785 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35786 on INTEGRAL_TYPE_P.
35788 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35789 Segher Boessenkool <segher@kernel.crashing.org>
35792 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35793 check reg_used_between_p between insn and one of succ or succ2
35794 depending on if succ is artificial insn not inserted into insn
35797 2017-03-21 Martin Liska <mliska@suse.cz>
35799 PR gcov-profile/80081
35800 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
35801 * doc/gcc.texi: Include gcov-dump stuff.
35802 * doc/gcov-dump.texi: New file.
35804 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
35806 PR rtl-optimization/79150
35807 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
35808 conditional jump, if the jump is the last insn of the loop.
35810 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35811 Richard Biener <rguenth@suse.de>
35813 PR tree-optimization/79908
35814 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35815 been cast away, use force_gimple_operand to construct the side
35818 2017-03-21 Martin Liska <mliska@suse.cz>
35820 PR libfortran/79956
35821 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
35824 2017-03-21 Brad Spengler <spender@grsecurity.net>
35827 * plugin.c (htab_hash_plugin): New function.
35828 (add_new_plugin): Use it and adjust.
35829 (parse_plugin_arg_opt): Adjust.
35830 (init_one_plugin): Likewise.
35832 2017-03-21 Richard Biener <rguenther@suse.de>
35834 PR tree-optimization/80032
35835 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35836 if set force the cleanup to happen unconditionally.
35837 (gimplify_target_expr): Push inserted clobbers with force_uncond
35838 to avoid them being removed by control-dependent DCE.
35840 2017-03-21 Richard Biener <rguenther@suse.de>
35842 PR tree-optimization/80122
35843 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35844 va_arg_pack_len when the inlined call stmt requires pack
35846 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35848 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35851 * tsan.c (instrument_builtin_call): If the memory model argument
35852 is not a constant, assume it is valid.
35855 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
35858 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35860 PR rtl-optimization/79910
35861 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35862 if its dest is used by an insn before I2 (other than the combined
35863 insns themselves, which are properly handled already).
35865 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35868 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35870 * combine.c (record_used_regs): New static function.
35871 (try_combine): Handle situations where there is an additional
35872 instruction between I2 and I3 which needs to have a LOG_LINK
35876 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35878 * combine.c (try_combine): Delete redundant i1 test. Call
35879 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35881 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35884 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35885 alternatives 13/14.
35887 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35889 PR tree-optimization/80054
35890 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35891 the optimization if a PHI or any of its arguments is not dominated
35892 by the candidate's basis. Use gphi* rather than gimple* as
35894 (replace_profitable_candidates): Clean up a gimple* variable that
35895 should be a gphi* variable.
35897 2017-03-20 Martin Sebor <msebor@redhat.com>
35900 * doc/extend.texi (attribute constructor): Document present limitation.
35902 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35905 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35906 __POWER9_VECTOR__ #ifdef control, change template definition to
35907 use Power9-specific built-in function.
35908 (vec_any_eq): Likewise.
35909 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35910 to control outcomes from this test.
35911 (vector_ae_<mode>p): For VEC_F modes, likewise.
35913 2017-03-20 Ian Lance Taylor <iant@google.com>
35915 * config/i386/i386.c (ix86_function_regparm): Save an extra
35916 register for -fsplit-stack with DECL_STATIC_CHAIN.
35918 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35921 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35922 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35924 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35926 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35928 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35931 2017-03-20 Marek Polacek <polacek@redhat.com>
35934 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35936 2017-03-20 Richard Biener <rguenther@suse.de>
35938 PR tree-optimization/80113
35939 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35940 allocate extra SSA name for PHI def.
35941 (add_close_phis_to_outer_loops): Likewise.
35942 (add_close_phis_to_merge_points): Likewise.
35943 (copy_loop_close_phi_args): Likewise.
35944 (copy_cond_phi_nodes): Likewise.
35946 2017-03-20 Martin Liska <mliska@suse.cz>
35948 PR middle-end/79753
35949 * tree-chkp.c (chkp_build_returned_bound): Do not build
35950 returned bounds for a LHS that's not a BOUNDED_P type.
35952 2017-03-20 Martin Liska <mliska@suse.cz>
35956 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35957 COMPLEX_CST and VECTOR_CST.
35959 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35962 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35963 target operand. A new splitter adds the clobber statement in case
35964 the target operand is dead anyway.
35966 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35968 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35969 to age-old versions of binutils and glibc.
35971 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35973 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35975 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35977 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35979 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35981 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35982 requirement for binutils 2.13.
35984 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35986 * combine.c (try_combine): Delete redundant i1 test. Call
35987 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35989 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35991 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35992 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35994 <riscv64-*-elf>: Re-arrange section
35995 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35996 <riscv32-*-linux>: Likewise.
35997 <riscv64-*-elf>: Likewise
35998 <riscv64-*-linux>: Likewise.
36000 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
36003 * aarch64.opt(verbose-cost-dump): Fix typo.
36005 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
36008 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36009 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36011 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36013 * reload.c (find_reloads): When reloading a nonoffsettable address,
36014 use RELOAD_OTHER for it and its address reloads.
36016 PR rtl-optimization/79910
36017 * combine.c (record_used_regs): New static function.
36018 (try_combine): Handle situations where there is an additional
36019 instruction between I2 and I3 which needs to have a LOG_LINK
36022 2017-03-17 Jeff Law <law@redhat.com>
36024 PR tree-optimization/71437
36025 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36026 conditional in the hash table first.
36027 (vrp_dom_walker::before_dom_children): Extract condition from
36028 ASSERT_EXPR. Record condition, its inverion and any implied
36029 conditions as well.
36031 2017-03-17 Marek Polacek <polacek@redhat.com>
36032 Markus Trippelsdorf <markus@trippelsdorf.de>
36034 PR tree-optimization/80079
36035 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36038 2017-03-17 Richard Biener <rguenther@suse.de>
36040 PR middle-end/80075
36041 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36042 Properly verify the LHS before the RHS possibly claims to be
36044 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36047 2017-03-17 Martin Jambor <mjambor@suse.cz>
36049 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36050 (List of -O2 options): Likewise.
36051 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36054 2017-03-17 Tom de Vries <tom@codesourcery.com>
36056 * gcov-dump.c (print_usage): Print bug_report_url.
36058 2017-03-17 Richard Biener <rguenther@suse.de>
36060 PR middle-end/80050
36061 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36062 (parser::peek): Likewise.
36064 2017-03-17 Richard Biener <rguenther@suse.de>
36066 PR tree-optimization/80048
36067 * sese.c (free_sese_info): Properly release rename_map and
36068 copied_bb_map elements.
36070 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36072 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36073 Add linked-list forward and backlinks. Insert on
36074 construction, remove on destruction.
36075 (class pass_store_merging): Add m_stores_head field.
36076 (pass_store_merging::terminate_and_process_all_chains):
36077 Iterate over m_stores_head list.
36078 (pass_store_merging::terminate_all_aliasing_chains):
36080 (pass_store_merging::execute): Check for debug stmts first.
36081 Push new chains onto the m_stores_head stack.
36083 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36086 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36087 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36088 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36090 2017-03-16 Jeff Law <law@redhat.com>
36092 PR tree-optimization/71437
36093 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36094 member function. Implementation moved into after_dom_children
36095 member function and into the threader's thread_outgoing_edges
36097 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36098 some code into new thread_outgoing_edges.
36099 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36100 definition. Simplify marker handling (do it here). Assume we always
36101 have the available expression and the const/copies tables.
36102 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36104 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36105 * tree-vrp.c (equiv_stack): No longer file scoped.
36106 (vrp_dom_walker): New class.
36107 (vrp_dom_walker::before_dom_children): New member function.
36108 (vrp_dom_walker::after_dom_children): Likewise.
36109 (identify_jump_threads): Setup domwalker. Use it rather than
36110 walking edges in a random order by hand. Simplify setup/finalization.
36111 (finalize_jump_threads): Remove.
36112 (vrp_finalize): Do not call identify_jump_threads here.
36113 (execute_vrp): Do it here instead and call thread_through_all_blocks
36116 PR tree-optimization/71437
36117 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36119 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36121 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36122 (dom_opt_dom_walker::thread_across_edge): Remove
36123 handle_dominating_asserts argument. All callers changed.
36124 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36125 changes. Remove calls to lhs_of_dominating_assert. Other
36126 uses of handle_dominating_asserts turn into unconditional code
36127 (simplify_control_stmt_condition_1): Likewise.
36128 (simplify_control_stmt_condition): Likewise.
36129 (thread_through_normal_block, thread_across_edge): Likewise.
36130 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36131 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36132 object if it is not an SSA_NAME.
36133 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36134 before calling into the VRP specific simplifiers.
36135 (identify_jump_threads): Remove handle_dominating_asserts
36138 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36141 * tree-diagnostic.c (default_tree_printer): No longer static.
36142 * tree-diagnostic.h (default_tree_printer): New prototype.
36144 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36146 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36147 Change ins into fmov.
36149 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36151 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36152 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36153 Use h_con constraint for operand 1.
36154 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36155 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36157 2017-03-15 Jeff Law <law@redhat.com>
36159 PR tree-optimization/71437
36160 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36161 (record_temporary_equivalences): Use it.
36163 PR tree-optimization/71437
36164 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36165 tree-ssa-scopedtables.
36166 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36167 (record_conditions, record_cond, vuse_eq): Likewise.
36168 (record_edge_info): Adjust to API tweak of record_conditions.
36169 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36170 (record_temporary_equivalences, optimize_stmt): Likewise.
36171 (eliminate_redundant_computations): Likewise.
36172 (record_equivalences_from_stmt): Likewise.
36173 * tree-ssa-scopedtables.c: Include options.h and params.h.
36174 (vuse_eq): New function, moved from tree-ssa-dom.c
36175 (build_and_record_new_cond): Likewise.
36176 (record_conditions): Likewise. Accept vector of conditions rather
36177 than edge_equivalence structure for first argument.
36178 for the first argument.
36179 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36180 from tree-ssa-dom.c.
36181 (avail_exprs_stack::record_cond): Likewise.
36182 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36183 from tree-ssa-dom.c.
36184 (avail_exprs_stack): Add new member functions lookup_avail_expr
36186 (record_conditions): Declare.
36188 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36191 * lra-constraints.c (process_alt_operands): Increase reject for
36192 reloading an input/output operand.
36194 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36197 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36198 insns to convert from signed/unsigned char/short to IEEE 128-bit
36200 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36202 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36205 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36206 subreg of inner mode for values already in registers.
36208 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36210 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36211 iteration reg is used after the loop.
36213 2017-03-14 Martin Sebor <msebor@redhat.com>
36215 PR tree-optimization/79800
36216 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36217 precision in negative-positive range.
36218 (format_floating): Call non-const overload with adjusted precision.
36220 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36223 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36226 2017-03-14 Martin Sebor <msebor@redhat.com>
36228 PR middle-end/80020
36229 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36230 * builtins.def (aligned_alloc): Use it.
36233 * Makefile.in (GTFILES): Add calls.c.
36234 * calls.c: Include "gt-calls.h".
36236 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36238 PR rtl-optimization/79728
36239 * regs.h (struct target_regs): New field
36240 x_contains_allocatable_regs_of_mode.
36241 (contains_allocatable_regs_of_mode): New macro.
36242 * reginfo.c (init_reg_sets_1): Initialize it, and change
36243 contains_reg_of_mode so it includes global regs as well.
36244 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36245 rather than contains_regs_of_mode.
36247 2017-03-14 Martin Liska <mliska@suse.cz>
36249 * doc/invoke.texi: Document options that can't be combined with
36250 -fcheck-pointer-bounds.
36252 2017-03-14 Martin Liska <mliska@suse.cz>
36254 PR middle-end/79831
36255 * doc/invoke.texi (-Wchkp): Document the option.
36257 2017-03-14 Martin Liska <mliska@suse.cz>
36259 * Makefile.in: Install gcov-dump.
36261 2017-03-14 Martin Liska <mliska@suse.cz>
36263 * multiple_target.c (expand_target_clones): Bail out for
36264 an invalid attribute.
36266 2017-03-14 Richard Biener <rguenther@suse.de>
36268 * alias.c (struct alias_set_entry): Pack properly.
36269 * cfgloop.h (struct loop): Likewise.
36270 * cse.c (struct set): Likewise.
36271 * ipa-utils.c (struct searchc_env): Likewise.
36272 * loop-invariant.c (struct invariant): Likewise.
36273 * lra-remat.c (struct cand): Likewise.
36274 * recog.c (struct change_t): Likewise.
36275 * rtl.h (struct address_info): Likewise.
36276 * symbol-summary.h (function_summary): Likewise.
36277 * tree-loop-distribution.c (struct partition): Likewise.
36278 * tree-object-size.c (struct object_size_info): Likewise.
36279 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36280 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36281 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36282 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36283 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36284 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36285 (struct _stmt_vec_info): Likewise.
36287 2017-03-14 Martin Liska <mliska@suse.cz>
36290 * multiple_target.c (create_dispatcher_calls): Check that
36291 a target can create a function dispatcher.
36293 2017-03-14 Martin Liska <mliska@suse.cz>
36296 * multiple_target.c (expand_target_clones): Drop local.local
36297 flag for default implementation.
36299 2017-03-14 Richard Biener <rguenther@suse.de>
36301 PR tree-optimization/80030
36302 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36304 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36306 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36307 gcc_fallthrough() instead of __attribute__((fallthrough));
36309 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36311 * doc/gcc.texi: Remove "up" link to (DIR).
36312 * doc/gccint.texi: Ditto.
36314 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36316 * doc/install.texi (Specific) <avr>: Remove reference to
36319 2017-03-13 Jeff Law <law@redhat.com>
36321 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36322 attribute rather than comments.
36324 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36325 match_scratch operand is highest.
36327 2017-03-13 Martin Liska <mliska@suse.cz>
36329 PR middle-end/78339
36330 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36331 is a CHKP clone, use original declaration.
36333 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36335 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36336 (arc_conditional_register_usage): Use a different allocation order
36337 when optimizing for size.
36338 * common/config/arc/arc-common.c (arc_option_optimization_table):
36339 Section anchors default on when optimizing for size.
36341 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36343 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36345 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36347 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36348 * config/arc/arc.md (cpu_facility): Add cd variant.
36349 (*movqi_insn): Add code density variant.
36350 (*movhi_insn): Likewise.
36351 (*movqi_insn): Likewise.
36352 (*addsi3_mixed): Likewise.
36353 (subsi3_insn): Likewise.
36355 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36357 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36359 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36361 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36362 expressions with MINUS and UNARY ops.
36364 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36367 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36369 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36370 between vec_select and vector argument.
36371 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36372 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36373 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36374 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36375 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36376 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36378 2017-03-13 Richard Biener <rguenther@suse.de>
36381 * params.def (vect-max-peeling-for-alignment): Fix typo.
36383 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36385 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36386 issue that only occurred with binutils below 2.18.
36388 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36390 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36391 refer to binutils 2.11/2.12 minimum.
36393 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36395 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36396 ftp.kernel.org and simplify binutils requirement.
36398 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36400 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36402 (Optimize Options): Ditto. Also remove redundancy.
36404 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36406 PR translation/79848
36407 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36409 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36410 to G_ to avoid double translation.
36412 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36414 PR translation/79923
36415 * auto-profile.c (get_combined_location): Convert leading
36416 character of diagnostics to lower case and remove trailing period.
36417 (read_profile): Likewise for various diagnostics.
36418 * config/arm/arm.c (arm_option_override): Remove trailing period
36419 from various diagnostics.
36420 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36421 (msp430_expand_delay_cycles): Likewise.
36423 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36426 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36427 full command-line argument, rather than just "str".
36428 (aarch64_validate_march): Likewise.
36429 (aarch64_validate_mtune): Likewise.
36431 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36433 PR rtl-optimization/78911
36434 * lra-assigns.c (must_not_spill_p): New function.
36435 (spill_for): Use it.
36437 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36439 PR tree-optimization/79981
36440 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36441 ATOMIC_COMPARE_EXCHANGE ifn result.
36442 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36443 IFN_ATOMIC_COMPARE_EXCHANGE.
36445 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36448 * opts.c (parse_sanitizer_options): Add missing question mark to
36449 "did you mean" message.
36451 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36453 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36455 (VMULEUH_UNS): Likewise.
36456 (VMULOUB_UNS): Likewise.
36457 (VMULOUH_UNS): Likewise.
36458 * config/rs6000/rs6000.c (builtin_function_type): Remove
36459 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36461 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36464 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36465 x with result of extra_parsing_for_operand_code_0.
36466 (function_reader::extra_parsing_for_operand_code_0): Convert
36467 return type from void to rtx, returning x. When reading
36468 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36469 larger size containing struct block_symbol.
36471 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36473 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36474 -mfloat128-hardware without -m64.
36476 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36479 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36480 entries to the case statement that marks unsigned arguments to
36481 overloaded functions.
36483 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36485 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36486 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36488 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36491 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36492 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36494 2017-03-10 Martin Liska <mliska@suse.cz>
36498 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36499 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36502 2017-03-10 Olivier Hainque <hainque@adacore.com>
36504 * tree-switch-conversion (array_value_type): Start by resetting
36505 candidate type to it's main variant.
36507 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36509 PR rtl-optimization/79909
36510 * combine.c (try_combine): Use simplify_replace_rtx on individual
36511 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36512 of the whole CALL_INSN_FUNCTION_USAGE.
36514 PR tree-optimization/79972
36515 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36516 get_range_info on SSA_NAMEs. Formatting fixes.
36518 2017-03-10 Richard Biener <rguenther@suse.de>
36519 Jakub Jelinek <jakub@redhat.com>
36521 PR tree-optimization/77975
36522 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36523 edge to be constant.
36524 (get_val_for): For constant x return it. Formatting fix.
36525 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36526 would use the same bases as the current one.
36528 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36530 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36531 instead of vec_select for V1TImode.
36532 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36534 (VSX_LE_128): Add V1TI to this mode iterator.
36535 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36536 (*vsx_le_perm_store_<mode>): Likewise.
36537 (pre-reload splitter for VSX stores): Likewise.
36538 (post-reload splitter for VSX stores): Likewise.
36539 (*vsx_xxpermdi2_le_<mode>): Likewise.
36540 (*vsx_lxvd2x2_le_<mode>): Likewise.
36541 (*vsx_stxvd2x2_le_<mode>): Likewise.
36543 2017-03-09 Michael Eager <eager@eagercon.com>
36545 Correct failures with --enable-checking=yes,rtl.
36547 * config/microblaze/microblaze.c (microblaze_expand_shift):
36548 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36549 test for const0_rtx.
36550 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36551 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36553 2017-03-09 Richard Biener <rguenther@suse.de>
36555 PR tree-optimization/79977
36556 * graphite-scop-detection.c (scop_detection::merge_sese):
36557 Handle the case of extra exits to blocks dominating the entry.
36559 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36561 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36564 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36566 PR rtl-optimization/79949
36567 * lra-constraints.c (process_alt_operands): Check memory when
36568 trying to predict a cycle. Print about the overall increase.
36570 2017-03-09 Richard Biener <rguenther@suse.de>
36572 PR middle-end/79971
36573 * gimple-expr.c (useless_type_conversion_p): Preserve
36574 TYPE_SATURATING for fixed-point types.
36576 2017-03-09 Richard Biener <rguenther@suse.de>
36579 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36580 alignment of BLKmode params.
36582 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36585 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36586 (VALL_NO_V2Q): Likewise.
36588 * config/aarch64/aarch64-simd.md
36589 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36591 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36592 VALL_NO_V2Q mode iterator.
36593 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36595 2017-03-09 Martin Liska <mliska@suse.cz>
36597 PR tree-optimization/79631
36598 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36599 tree_int_cst_sign_bit just for INTEGER constants.
36601 2017-03-09 Martin Liska <mliska@suse.cz>
36605 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36608 2017-03-09 Marek Polacek <polacek@redhat.com>
36611 * tree.c (inchash::add_expr): Handle TREE_VEC.
36613 2017-03-09 Martin Liska <mliska@suse.cz>
36616 (chkp_narrow_size_and_offset): New function.
36617 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36618 (void chkp_parse_bit_field_ref): New function.
36619 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36620 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36622 2017-03-09 Martin Liska <mliska@suse.cz>
36625 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36626 (chkp_find_bounds_1): Remove gcc_unreachable.
36628 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36631 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36632 BUILT_IN_SYNC*, determine the access type from the size suffix and
36633 always build a MEM_REF with that type. Handle forgotten
36634 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36637 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36638 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36639 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36640 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36641 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36642 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36643 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36644 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36645 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36646 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36647 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36648 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36649 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36650 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36651 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36652 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36653 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36654 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36655 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36656 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36657 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36658 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36659 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36660 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36661 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36662 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36663 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36664 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36665 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36666 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36667 definitions outside of __OPTIMIZE__ guarded section.
36670 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36671 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36672 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36673 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36676 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36678 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36679 ("vfenez<mode>"): Add missing constraints.
36681 2017-03-08 Martin Sebor <msebor@redhat.com>
36684 * config/nds32/nds32.c (nds32_option_override):
36685 Fix misspelled diagnostic.
36687 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36690 * gimplify.c (gimplify_omp_for): Replace index var in outer
36691 taskloop statement with an artificial variable and add
36692 OMP_CLAUSE_PRIVATE clause for it.
36694 2017-03-08 Richard Biener <rguenther@suse.de>
36696 PR tree-optimization/79955
36697 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36698 for accesses that are completely outside of the variable.
36700 2017-03-08 Andrew Haley <aph@redhat.com>
36702 PR tree-optimization/79943
36703 * tree-ssa-loop-split.c (compute_new_first_bound): When
36704 calculating the new upper bound, (END-BEG) should be added, not
36707 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36709 * config/avr/avr.md (setmemhi): Make sure match_dup
36710 operand number comes before match_scratch.
36712 2017-03-08 Richard Biener <rguenther@suse.de>
36714 PR tree-optimization/79920
36715 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36716 with ncopies == 1 to ...
36717 (vect_transform_slp_perm_load): ... here. Properly compute
36718 all element loads by iterating VF times over the group. Do
36719 not handle ncopies (computed in a broken way) in
36720 vect_create_mask_and_perm.
36722 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36725 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36726 is a uniform vector, use uniform_vector_p return value instead of
36727 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36729 2017-03-07 Marek Polacek <polacek@redhat.com>
36731 PR middle-end/79809
36732 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36733 (alloca_call_type): Likewise.
36735 2017-03-07 Martin Liska <mliska@suse.cz>
36737 * gcov.c (process_args): Put comment to correct location.
36739 2017-03-07 Martin Liska <mliska@suse.cz>
36741 PR middle-end/68270
36742 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36743 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36744 (chkp_narrow_bounds_for_field): Likewise.
36745 (chkp_parse_array_and_component_ref): Pass one more argument to
36748 2017-03-07 Richard Biener <rguenther@suse.de>
36750 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36753 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36755 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36756 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36758 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36761 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36762 to end of description.
36763 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36765 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36767 PR rtl-optimization/79901
36768 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36770 (*avx512f_<code><mode>3<mask_name>): ... this.
36771 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36772 iterator instead of VI8_AVX2_AVX512BW.
36774 PR rtl-optimization/79901
36775 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36776 min/max expander, expand it using expand_vec_cond_expr.
36779 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36782 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36785 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36786 to void * for PCH reasons.
36787 * dwarf2out.c (output_loc_operands, output_die): Cast
36788 v.val_vec.array to unsigned char *.
36790 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36793 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36796 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36798 PR rtl-optimization/79571
36799 * lra-constraints.c (process_alt_operands): Calculate static
36800 reject and subtract it from overall when only addresses will be
36803 2017-03-06 Julia Koval <julia.koval@intel.com>
36806 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
36807 incoming stack boundary to 128 for 64-bit targets.
36809 2017-03-06 Richard Biener <rguenther@suse.de>
36811 PR tree-optimization/79894
36812 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
36813 to NULL after folding it.
36815 2017-03-06 Richard Biener <rguenther@suse.de>
36817 PR tree-optimization/79824
36818 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
36819 check disabling peeling for gaps.
36821 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
36823 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
36824 attributes): Document gettimeofday.
36826 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36828 * config/s390/s390.c (s390_option_override_internal): Set
36829 PARAM_MIN_VECT_LOOP_BOUND
36831 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36833 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36834 * config/s390/s390.md: Likewise.
36836 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36839 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36840 (<avx2_avx512>_perm<mode>): Rename to ...
36841 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36843 (<avx512>_perm<mode>_mask): Rename to ...
36844 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36846 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36847 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36848 instead of VI8F_256_512.
36849 (avx512f_perm<mode>): New define_expand.
36850 (avx512f_perm<mode>_mask): Likewise.
36851 (avx512f_perm<mode>_1<mask_name>): New define_insn.
36852 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
36854 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36856 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
36857 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
36859 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36861 2017-03-06 Martin Liska <mliska@suse.cz>
36864 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
36865 when having a SSA NAME w/o VAR_DECL assigned to it.
36867 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36869 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
36870 msa_dpsub_<su>_d): Fix MODE for vec_select.
36872 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36874 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
36876 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
36878 2017-03-06 Richard Biener <rguenther@suse.de>
36880 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
36881 * plugin.c (register_plugin_info): Likewise.
36882 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36884 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36886 * config/i386/sse.md (sse_storehps, sse_storelps,
36887 avx_<castmode><avxsizesuffix>_<castmode>,
36888 avx512f_<castmode><avxsizesuffix>_<castmode>,
36889 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36890 in condition that at least one operand is not a MEM.
36892 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36894 PR middle-end/79805
36895 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36896 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36898 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36899 gimple_call_nothrow_p flag based on whether original builtin can throw.
36900 If it can, emit following stmts on the fallthrough edge.
36901 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36902 don't create new bb if inserting just debug stmts on the edge, try to
36903 insert them on the fallthru bb or just reset debug stmts.
36905 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36908 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36909 restore recog_data (including the operand rtxes inside it) around
36910 the call to get_insn_template.
36912 2017-03-03 Martin Sebor <msebor@redhat.com>
36914 PR tree-optimization/79699
36915 * context.c (context::~context): Free MPFR caches to avoid
36916 a memory leak on program exit.
36918 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36920 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36921 Use wide_int::ulow () instead of .elt (0).
36923 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36925 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36926 (*pushxf): Limit oF constraint to 32bit targets and add oC
36927 constraint for 64bit targets.
36928 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36929 (*pushdf): Change rmF constraint to rmC.
36931 2017-03-03 Martin Liska <mliska@suse.cz>
36933 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36934 Remove unused variable.
36936 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36939 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36940 is a memory operand, increase num_memory.
36941 (ix86_expand_args_builtin): Likewise.
36943 2017-03-03 Jan Hubicka <jh@suse.cz>
36946 * ipa-devirt.c (maybe_record_node): Properly handle
36947 __cxa_pure_virtual visibility.
36949 2017-03-03 Martin Liska <mliska@suse.cz>
36951 PR tree-optimization/79803
36952 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36954 (pass_loop_prefetch::execute): Disabled optimization if an
36955 assumption about L1 cache size is not met.
36957 2017-03-03 Martin Liska <mliska@suse.cz>
36959 PR rtl-optimization/79574
36960 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36961 (hash_scan_set): Likewise.
36962 (dump_hash_table): Likewise.
36963 (hoist_code): Likewise.
36965 2017-03-03 Richard Biener <rguenther@suse.de>
36967 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36968 in place of uhigh ().
36969 (fixed_convert_from_real): Likewise.
36971 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36974 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36976 2017-03-03 Richard Biener <rguenther@suse.de>
36978 PR middle-end/79818
36979 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36980 TYPE_OVERFLOW_UNDEFINED check.
36982 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36984 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36986 (vector_ae_<mode>_p): Likewise.
36987 (vector_nez_<mode>_p): Likewise.
36988 (vector_ne_v2di_p): Likewise.
36989 (vector_ae_v2di_p): Likewise.
36990 (vector_ne_<mode>_p): Likewise.
36991 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36993 (vsx_tsqrt<mode>2_fe): Likewise.
36995 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36998 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
37000 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37002 PR rtl-optimization/79780
37003 * cprop.c (one_cprop_pass): When second and further conditional trap
37004 in a single basic block is turned into an unconditional trap, turn it
37005 into a deleted note to avoid RTL verification failures.
37007 2017-03-02 Richard Biener <rguenther@suse.de>
37009 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37011 2017-03-02 Richard Biener <rguenther@suse.de>
37013 PR tree-optimization/79345
37015 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37016 param and abort the walk, returning -1 if it is hit.
37017 (walk_aliased_vdefs): Take a limit param and pass it on.
37018 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37019 defaulting to 0 and return a signed int.
37020 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37021 (check_defs): New helper.
37022 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37023 about uninitialized memory.
37024 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37025 bogus uninitialized warning.
37026 (fixed_convert_from_real): Likewise.
37028 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37030 PR tree-optimization/66768
37031 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37032 iv_use if base object can't be determined.
37034 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37036 PR tree-optimization/79345
37037 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37038 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37039 (get_pattern_stats): Initialize it.
37040 * genemit.c (gen_expand): Verify match_scratch numbers come after
37041 match_operand/match_dup numbers.
37042 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37043 match_scratch numbers.
37044 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37046 * config/s390/s390.md (trunctdsd2): Likewise.
37048 2017-03-02 Richard Biener <rguenther@suse.de>
37050 * wide-int.h (wide_int_storage::operator=): Implement in terms
37053 2017-03-02 Richard Biener <rguenther@suse.de>
37055 PR tree-optimization/79777
37056 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37057 the to insert expression to sth existing.
37059 2017-03-01 Martin Sebor <msebor@redhat.com>
37061 PR middle-end/79692
37062 * gimple-ssa-sprintf.c
37063 (directive::known_width_and_precision): New function.
37064 (format_integer): Use it.
37065 (get_mpfr_format_length): Consider the full range of precision
37066 when computing %g output with the # flag. Set the likely byte
37067 count to 3 rather than 1 when precision is indeterminate.
37068 (format_floating): Correct the lower bound of precision.
37070 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37072 * doc/invoke.texi: Document default code model for 64-bit Linux.
37074 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37077 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37078 udiv rather than div since input pattern is unsigned.
37080 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37082 * config/i386/i386.c (print_reg): Warn for values of
37083 unsupported size in integer register.
37085 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37088 * config/rs6000/predicates.md (current_file_function_operand): Do
37089 not allow self calls to be local if the function is replaceable.
37091 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37094 * config/rs6000/altivec.h (vec_ctz and others): Change the
37095 preprocessor macro that controls conditional compilation from
37096 _ARCH_PWR9 to __POWER9_VECTOR__.
37097 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37098 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37099 control (instead of _ARCH_PWR9 control) so that template
37100 definition uses power9-specific function.
37101 (vec_any_eq): Likewise.
37102 (vec_all_ne): Change macro definition to use a power9-specific
37103 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37104 _ARCH_PWR9 control).
37105 (vec_any_eq) Likewise.
37106 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37107 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37108 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37109 support for xvcmpnedp instruction.
37110 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37111 macro expansion so that Power9 implementation of vec_all_ne does
37112 not use the AltiVec predicate framework.
37113 (VCMPNEH_P): Likewise.
37114 (VCMPNEW_P): Likewise.
37115 (VCMPNED_P): Likewise.
37116 (VCMPNEFP_P): Likewise.
37117 (VCMPNEDP_P): Likewise.
37118 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37119 implementation of vec_any_eq to not use AltiVec predicate
37121 (VCMPAEH_P): Likewise.
37122 (VCMPAEW_P): Likewise.
37123 (VCMPAED_P): Likewise.
37124 (VCMPAEFP_P): Likewise.
37125 (VCMPAEDP_P): Likewise.
37126 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37127 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37128 not use the AltiVec predicate framework.
37129 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37130 of vec_any_eq to not use AltiVec predicate framework.
37131 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37132 support for predefined __POWER9_VECTOR__ macro to indicate that
37133 Power9 instruction selection is enabled.
37134 (altivec_overloaded_builtins): Remove extraneous
37135 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37136 function argument types RS6000_BTI_bool_V16QI and
37137 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37138 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37139 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37140 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37141 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37142 Power9 for implementations of vec_cmpne. Change the signature for
37143 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37144 (representing vec_all_ne) to remove the previously described first
37145 argument of type RS6000_BTI_INTSI, as this was an artifact of
37146 reliance on the AltiVec predicate framework, which is no longer
37147 used in the implementation of these functions. Add
37148 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37149 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37150 since, unlike the AltiVec predicate framework implementation, we
37151 do not share function descriptors between vec_alle and vec_anyeq.
37152 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37153 set of modes that receive special treatment even when
37154 TARGET_P9_VECTOR is true. The special treatment emits code that
37155 does not depend on Power9 instructions.
37156 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37157 define_expand to not rely on AltiVec predicate framework.
37158 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37160 (vector_ne_v2di_p): Change this define_expand to not rely on
37161 AltiVec predicate framework.
37162 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37164 (vector_ne_<mode>_p): Change this define_expand to not rely on
37165 AltiVec predicate framework.
37166 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37168 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37169 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37170 define_insn pattern.
37171 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37172 define_insn pattern because the xvcmpne<VSs>. instruction is not
37174 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37175 instruction is not supported.
37177 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37179 * config/nvptx/nvptx.c: Include intl.h.
37181 2017-03-01 Martin Jambor <mjambor@suse.cz>
37184 * ipa-prop.h (ipa_bits): Removed field known.
37185 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37186 to pointers. Adjusted their comments to warn about their sharing.
37187 (ipcp_transformation_summary): Change bits to a vector of pointers.
37188 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37189 (ipa_get_ipa_bits_for_value): Declare.
37190 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37191 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37192 (ipa_bits_hash_table): Likewise.
37193 (ipa_vr_ggc_hash_traits): Likewise.
37194 (ipa_vr_hash_table): Likewise.
37195 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37196 being pointers and vr_known being removed.
37197 (ipa_set_jf_unknown): Likewise.
37198 (ipa_get_ipa_bits_for_value): New function.
37199 (ipa_set_jfunc_bits): Likewise.
37200 (ipa_get_value_range): New overloaded functions.
37201 (ipa_set_jfunc_vr): Likewise.
37202 (ipa_compute_jump_functions_for_edge): Use the above functions to
37203 construct bits and vr parts of jump functions.
37204 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37205 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37207 (ipcp_grow_transformations_if_necessary): Also allocate
37208 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37210 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37211 them. Fix too long lines.
37212 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37213 vr_known being removed.
37214 (ipa_read_jump_function): Use new setter functions to construct bits
37215 and vr parts of jump functions or set them to NULL.
37216 (write_ipcp_transformation_info): Adjust for bits being pointers.
37217 (read_ipcp_transformation_info): Likewise.
37218 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37220 Include gt-ipa-prop.h.
37221 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37223 (ipcp_store_bits_results): Likewise.
37224 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37225 Do not write to existing jump functions but use a temporary instead.
37227 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37230 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37231 attempt to use its first operand as BIT_FIELD_REF base.
37233 2017-03-01 Richard Biener <rguenther@suse.de>
37235 PR middle-end/79721
37236 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37237 interpolating formula in wrapping arithmetic.
37238 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37240 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37242 PR tree-optimization/79734
37243 * tree-vect-generic.c (expand_vector_condition): Optimize
37244 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37245 Handle VEC_COND_EXPR where comparison has different inner width from
37246 type's inner width.
37248 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37250 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37251 markup, and similar issues. Remove @opindex entries for things
37252 that aren't options. Add missing -mmpy-option entries.
37254 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37256 PR tree-optimization/79737
37257 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37258 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37259 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37260 instead of byte_size. Formatting fix.
37261 (shift_bytes_in_array_right): Formatting fix.
37263 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37266 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37267 condition on optimize for the leaf function test.
37269 2017-02-28 Martin Liska <mliska@suse.cz>
37272 * read-rtl-function.c (function_reader::handle_unknown_directive):
37273 Bail out when one uses -flto.
37275 2017-02-28 Martin Liska <mliska@suse.cz>
37277 * common.opt: Replace space with tabular for options of <number>
37279 * config/i386/i386.opt: Show <number> value for
37280 -mlarge-data-threshold.
37281 * opts.c (print_filtered_help): Do not display number in hexadecimal
37284 2017-02-28 Martin Liska <mliska@suse.cz>
37286 * common.opt: Fix --help=option -Q for options which are of
37289 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37291 * config/i386/i386.c (print_reg): Error out for values
37292 of 8-bit size in invalid integer register.
37294 2017-02-28 Martin Sebor <msebor@redhat.com>
37296 PR tree-optimization/79691
37297 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37299 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37302 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37303 gcc_unreachable with output_operand_lossage.
37305 2017-02-28 Richard Biener <rguenther@suse.de>
37307 PR tree-optimization/79740
37308 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37310 (visit_nary_op): Insert the nary into the hashtable if we
37311 pattern-matched sth.
37312 * tree-ssa-pre.c (eliminate_insert): Robustify.
37314 2017-02-28 Richard Biener <rguenther@suse.de>
37316 PR middle-end/79731
37317 * fold-const.c (decode_field_reference): Reject out-of-bound
37320 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37322 * config/i386/i386.c: Include intl.h.
37323 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37324 instead of just cond ? "..." : "...".
37325 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37326 * coverage.c (read_counts_file): Likewise.
37327 * omp-offload.c: Include intl.h.
37328 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37329 of just cond ? "..." : "...".
37330 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37331 of just cond ? "..." : "...".
37333 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37336 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37338 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37339 'tune for' CPU name.
37340 * config/arm/arm-cpu-data.h: Regenerated.
37342 2017-02-28 Richard Biener <rguenther@suse.de>
37344 PR tree-optimization/79732
37345 * tree-inline.c (expand_call_inline): Do not shadow var.
37347 2017-02-28 Richard Biener <rguenther@suse.de>
37349 PR tree-optimization/79723
37350 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37351 address-space properly.
37353 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37355 * doc/optinfo.texi (Optimization groups): Fix option used for
37357 * doc/invoke.texi (-fopt-info): Document "omp".
37358 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37359 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37360 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37361 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37362 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37364 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37366 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37367 OPTGROUP_OMP with "omp".
37369 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37372 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37373 for arithmetic shift of unsigned V2DI.
37375 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37377 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37378 arc/linux.h headers.
37379 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37380 (LINK_SPEC): Likewise.
37381 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37382 (EXTRA_SPECS): Likewise.
37383 (STARTFILE_SPEC): Likewise.
37384 (ENDFILE_SPEC): Likewise.
37385 (LIB_SPEC): Likewise.
37386 (TARGET_SDATA_DEFAULT): Likewise.
37387 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37388 (MULTILIB_DEFAULTS): Likewise.
37389 (DWARF2_UNWIND_INFO): Likewise.
37390 * config/arc/big.h: New file.
37391 * config/arc/elf.h: Likewise.
37392 * config/arc/linux.h: Likewise.
37393 * config/arc/t-uClibc: Remove.
37395 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37397 PR tree-optimization/77536
37398 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37399 (tree_transform_and_unroll_loop): Use above function to compute the
37400 estimated niter of unrolled loop and use it when scaling profile.
37401 Also use count info rather than frequency if it's non-zero.
37402 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37403 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37404 (vect_transform_loop): Call above function.
37406 2017-02-27 Richard Biener <rguenther@suse.de>
37408 PR tree-optimization/45397
37409 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37410 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37411 (visit_nary_op): Add pattern matching for CSEing sign-changed
37412 or truncated operations with wider ones.
37414 2017-02-27 Richard Biener <rguenther@suse.de>
37416 PR tree-optimization/79690
37417 * tree-vect-stmts.c (vectorizable_store): Use vector type
37418 built from the DR with address-space.
37420 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37422 * doc/invoke.texi (Optimize Options): Refine the description
37423 of asan-use-after-return.
37425 2017-02-25 Alan Modra <amodra@gmail.com>
37427 PR rtl-optimization/79584
37428 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37429 base, not ad->base_term, the reg within base. Remove assertion
37430 that ad->base == ad->base_term. Replace gen_int_mode using
37431 bogus mode with const0_rtx.
37433 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37435 PR middle-end/79396
37436 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37437 FMA_EXPR like tcc_binary or tcc_unary.
37439 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37442 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37444 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37445 (output_loc_operands): Handle DW_OP_call_ref and
37446 DW_OP_GNU_variable_value.
37447 (struct variable_value_struct): New type.
37448 (struct variable_value_hasher): Likewise.
37449 (variable_value_hash): New variable.
37450 (string_types): Remove.
37451 (copy_loc_descr): New function.
37452 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37453 (prepend_loc_descr_to_each): New function.
37454 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37455 instead of add_loc_descr_to_each if the first argument is single
37456 location list and the second has multiple.
37457 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37458 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37459 when looking for variable value which doesn't have other location info.
37460 (loc_list_from_tree): Formatting fix.
37461 (gen_array_type_die): Simplify DW_AT_string_length handling.
37462 (adjust_string_types): Remove.
37463 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37464 string_types. Call resolve_variable_values.
37465 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37466 (resolve_addr_in_expr): Likewise. Add A argument.
37467 (copy_deref_exprloc): Remove deref argument. Adjust for the
37468 original expression being DW_OP_GNU_variable_value with optionally
37469 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37470 optionally after it.
37471 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37472 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37473 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37474 (variable_value_hasher::hash, variable_value_hasher::equal): New
37476 (resolve_variable_value_in_expr, resolve_variable_value,
37477 resolve_variable_values, note_variable_value_in_expr,
37478 note_variable_value): New functions.
37479 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37482 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37485 * opts.h (handle_generated_option): Add GENERATED_P argument.
37486 * opts-common.c (handle_option): Adjust function comment.
37487 (handle_generated_option): Add GENERATED_P argument, pass it to
37489 (control_warning_option): Pass false to handle_generated_option
37491 * opts.c (maybe_default_option): Pass true to handle_generated_option
37493 * optc-gen.awk: Likewise.
37495 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37497 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37498 a REG, look at the REG it is a SUBREG of.
37499 (splitter for cmpeqsi_t): Ditto.
37501 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37503 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37504 the special USEs with the pattern of the insn, not the insn itself.
37506 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37509 * doc/invoke.texi: Document -mload-store-pairs.
37511 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37512 Sandra Loosemore <sandra@codesourcery.com>
37514 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37515 argument isn't a CONST_INT.
37516 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37517 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37518 (nios2_validate_compare): Bypass alternate compare logic if *op2
37519 is not a CONST_INT.
37520 (ldstwm_operation_p): Return false if first_base is not a REG or
37521 if first_offset is not a CONST_INT.
37523 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37525 * config/cris/cris.md: Use correct operand in a define_peephole2.
37527 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37529 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37531 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37533 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37534 this_insn if it is an INSN or JUMP_INSN.
37535 (force_offsettable): Look at base, not at addr.
37536 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37537 on things that aren't necessarily CONST_INTs.
37539 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37541 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37542 -mfpmath=sse is the default also for x86-32 targets with SSE2
37543 instruction set when @option{-ffast-math} is enabled
37545 2017-02-24 Jeff Law <law@redhat.com>
37547 PR rtl-optimizatoin/79286
37548 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37551 2017-02-24 Richard Biener <rguenther@suse.de>
37553 PR tree-optimization/79389
37554 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37557 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37559 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37560 function comment to reflect reality.
37561 (loop_exits_before_overflow): Fix typo in function description.
37563 2017-02-24 Richard Biener <rguenther@suse.de>
37565 PR tree-optimization/79389
37566 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37567 properly that a threading opportunity exists. Detect conditional
37568 copy/constant propagation opportunities.
37570 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37572 * config/visium/visium.md (type): Add trap.
37573 (b): New mode attribute.
37574 (*btst): Rename into...
37575 (*btst<mode>): ...this and adjust.
37576 (*cbranchsi4_btst_insn): Rename into...
37577 (*cbranch<mode>4_btst_insn): ...this and adjust.
37578 (trap): New define_insn.
37580 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37582 PR tree-optimization/79389
37583 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37584 (noce_reversed_cond_code): New function.
37585 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37586 reversed_comparison_code. Formatting fix.
37587 (noce_try_store_flag): Test rev_cond != NULL in addition to
37588 reversed_comparison_code.
37589 (noce_try_store_flag_constants): Likewise.
37590 (noce_try_store_flag_mask): Likewise.
37591 (noce_try_addcc): Use rev_cond if non-NULL instead of
37592 reversed_comparison_code.
37593 (noce_try_cmove_arith): Likewise. Formatting fixes.
37594 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37595 (noce_find_if_block): Initialize rev_cond.
37596 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37597 instead of false as last argument never attempt to reverse it
37600 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37602 PR tree-optimization/79663
37603 * tree-predcom.c (combine_chains): Process refs in reverse order
37604 only for ZERO length chains, and add explaining comment.
37606 2017-02-23 Jeff Law <law@redhat.com>
37608 PR tree-optimization/79578
37609 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37610 in call to operand_equal_p.
37612 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37615 * config/i386/cpuid.h: Fix another undefined behavior.
37617 2017-02-23 Richard Biener <rguenther@suse.de>
37619 PR tree-optimization/79683
37620 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37621 vector types for data-refs.
37623 2017-02-23 Martin Liska <mliska@suse.cz>
37625 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37627 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37629 PR middle-end/79665
37630 * internal-fn.c (get_range_pos_neg): Moved to ...
37631 * tree.c (get_range_pos_neg): ... here. No longer static.
37632 * tree.h (get_range_pos_neg): New prototype.
37633 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37634 are known to be in between 0 and signed maximum inclusive, try to
37635 expand both unsigned and signed divmod and use the cheaper one from
37638 2017-02-22 Jeff Law <law@redhat.com>
37640 PR tree-optimization/79578
37641 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37642 to compare base operands.
37644 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37647 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37648 gpc_reg_operand instead of fpr_reg_operand.
37650 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37652 * config/mips/mips.c (mips_return_in_memory): Force FP
37653 vector types to be returned in memory for o32 ABI.
37655 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37657 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37658 instead of DW_TAG_member for static data member declarations and don't
37659 set no_linkage_name for static inline data members.
37660 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37663 2017-02-22 Martin Liska <mliska@suse.cz>
37665 * doc/invoke.texi: Replace inequality signs with square brackets
37668 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37671 * lra-constraints.c (simplify_operand_subreg): Handle
37672 WORD_REGISTER_OPERATIONS targets.
37674 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37677 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37678 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37679 elimination by swapping fld*.
37681 2017-02-22 Richard Biener <rguenther@suse.de>
37683 PR tree-optimization/79673
37684 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37685 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37686 irrelevant address-space qualifiers and avoiding a
37687 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37689 2017-02-22 Richard Biener <rguenther@suse.de>
37691 PR tree-optimization/79666
37692 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37693 to not symbolically negate if that may introduce undefined
37696 2017-02-22 Martin Liska <mliska@suse.cz>
37699 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37700 * data-streamer-out.c (streamer_write_gcov_count_stream):
37702 * value-prof.c (stream_out_histogram_value): Make assert more
37703 precise based on type of counter.
37705 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37708 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37709 Use nonimmediate_operand instead of memory_operand for operand 1.
37710 (float-extend standard_x87sse_constant_load splitter): Ditto.
37712 2017-02-21 Jeff Law <law@redhat.com>
37714 PR tree-optimization/79621
37715 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37716 blocks with edges to themselves.
37718 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37721 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37722 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37723 Use gimple_call_builtin_p.
37726 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37727 on temporarily removed DEBUG_INSNs.
37729 PR tree-optimization/79649
37730 * tree-loop-distribution.c (classify_partition): Give up on
37731 non-generic address space loads/stores.
37733 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37735 * doc/loop.texi (Loop manipulation): Remove nonexistent
37736 tree_ssa_loop_version from the documentation.
37737 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37739 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37742 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37743 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37744 * config/rs6000/rs6000.c: Include except.h.
37745 (rs6000_expand_split_stack_prologue): Call
37746 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37748 2017-02-21 Martin Jambor <mjambor@suse.cz>
37751 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37752 have been analyzed.
37754 2017-02-21 Martin Jambor <mjambor@suse.cz>
37756 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37757 for backward compatibility only.
37758 * doc/invoke.texi (Option Summary): Remove all references to
37759 -fipa-cp-alignment.
37761 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37765 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37767 * lra-constraints.c (curr_insn_transform): Handle
37768 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37770 2017-02-21 Martin Liska <mliska@suse.cz>
37772 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37774 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37776 PR translation/79638
37777 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37779 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37782 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37783 (arm_function_ok_for_sibcall): Return false for an indirect call by
37784 descriptor if all the argument registers are used.
37785 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37786 alignment of the function.
37788 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37790 PR tree-optimization/61441
37791 * simplify-rtx.c (simplify_const_unary_operation): For
37792 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37793 the sNaN unmodified.
37795 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37797 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37798 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
37799 instead of SYSTEM_HEADER_DIR.
37801 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
37802 Martin Liška <mliska@suse.cz>
37804 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
37805 Fix typos and grammar, use active voice, and clarify.
37807 2017-02-20 Marek Polacek <polacek@redhat.com>
37809 PR middle-end/79537
37810 * gimplify.c (gimplify_expr): Handle unused *&&L;.
37813 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
37815 2017-02-20 Jakub Jelinek <jakub@redhat.com>
37818 * config/i386/i386.c (ix86_expand_builtin): Handle
37819 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
37820 ix86_builtins_isa[fcode].isa as a requirement of those
37821 flags and any other flag in the bitmask.
37822 (ix86_init_mmx_sse_builtins): Use 0 instead of
37823 ~OPTION_MASK_ISA_64BIT as mask.
37824 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
37825 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
37826 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37827 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37829 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37832 * lra-constraints.c (split_reg): Check requested split mode
37833 is supported by the register.
37835 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37837 * lra-constraints.c (simplify_operand_subreg): Remove early
37840 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37843 * lra-constraints.c (curr_insn_transform): Tighten condition
37844 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37846 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37849 * lra-constraints.c (curr_insn_transform): Handle
37850 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37852 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
37855 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
37857 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
37859 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37862 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
37865 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37867 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
37870 2017-02-18 Jakub Jelinek <jakub@redhat.com>
37872 * final.c (last_columnnum, override_columnnum): New variables.
37873 (final_start_function): Set last_columnnum, pass it to begin_prologue
37874 hook and pass 0 to dwarf2out_begin_prologue.
37875 (final_scan_insn): Update override_columnnum. Pass last_columnnum
37876 to source_line debug hook.
37877 (notice_source_line): Compute last_columnnum and for debug_column_info
37878 return true on column changes.
37879 * debug.h (struct gcc_debug_hooks): Add column argument to
37880 source_line and begin_prologue hooks.
37881 (debug_nothing_int_charstar_int_bool): Remove prototype.
37882 (debug_nothing_int_int_charstar,
37883 debug_nothing_int_int_charstar_int_bool): New prototypes.
37884 (dwarf2out_begin_prologue): Add column argument.
37885 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37886 begin_prologue hooks.
37887 (debug_nothing_int_charstar_int_bool): Remove.
37888 (debug_nothing_int_int_charstar,
37889 debug_nothing_int_int_charstar_int_bool): New functions.
37890 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37891 through to dwarf2out_source_line.
37892 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37893 (dwarf2out_source_line): Add column argument, emit it if requested.
37894 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37896 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37897 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37898 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37899 through to dwarf2out_begin_prologue.
37900 (vmsdbgout_source_line): Add column argument, pass it through to
37901 dwarf2out_source_line.
37902 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37903 dbxout_source_line caller.
37904 (dbxout_source_line): Add column argument.
37906 * common.opt (gno-column-info, gcolumn-info): New options.
37907 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37908 (check_die): Also test for multiple DW_AT_decl_column attributes.
37909 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37910 DW_AT_decl_column if requested.
37911 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37913 (gen_variable_die): Likewise.
37914 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37915 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37918 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37919 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37920 (ix86_handle_option): Handle OPT_m3dnowa.
37921 * doc/invoke.texi (-m3dnowa): Document.
37922 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37923 -m3dnowa instead of -m3dnow -march=athlon.
37926 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37927 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37929 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37932 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37933 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37934 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37935 generator for vsx_xxpermdi_<mode>_be.
37936 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37937 force big-endian semantics.
37938 (vsx_xxpermdi_<mode>_be): New define_expand with same
37939 implementation as previous version of vsx_xxpermdi_<mode>.
37941 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37943 PR tree-optimization/79327
37944 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37945 variable, its initialization and use.
37947 2017-02-17 Julia Koval <julia.koval@intel.com>
37949 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37950 (OPTION_MASK_ISA_PKU_UNSET): New.
37951 (ix86_handle_option): Handle -mrdpid.
37952 * config/i386/cpuid.h (bit_RDPID): New.
37953 * config/i386/driver-i386.c (host_detect_local_cpu):
37954 Detect RDPID feature.
37955 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37956 * config/i386/i386-c.c (ix86_target_macros_internal):
37958 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37959 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37960 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37961 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37962 * config/i386/i386.md (define_insn "rdpid"): New.
37963 * config/i386/i386.opt Add -mrdpid.
37964 * config/i386/immintrin.h (_rdpid_u32): New.
37966 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37968 PR rtl-optimization/79541
37969 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37970 instead of transforming it into USE.
37972 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37974 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37975 If HONOR_SNANS (SFmode) force the input to a register.
37976 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37977 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37978 an frsp or similar insn.
37980 2017-02-17 Martin Liska <mliska@suse.cz>
37982 PR rtl-optimization/79577
37983 * params.def (selsched-max-sched-times): Increase minimum to 1.
37985 2017-02-17 Martin Liska <mliska@suse.cz>
37987 PR rtl-optimization/79574
37988 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37990 2017-02-17 Martin Liska <mliska@suse.cz>
37992 PR tree-optimization/79529
37993 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37994 ssa_defined_default_def_p to handle cases which are implicitly
37996 * tree-ssa.c (ssa_defined_default_def_p): New function.
37997 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37998 which are implicitly defined.
37999 * tree-ssa.h (ssa_defined_default_def_p): Declare.
38001 2017-02-17 Richard Biener <rguenther@suse.de>
38003 PR middle-end/79576
38004 * params.def (max-ssa-name-query-depth): Limit to 10.
38006 2017-02-17 Richard Biener <rguenther@suse.de>
38008 PR tree-optimization/79552
38009 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38012 2017-02-17 Richard Biener <rguenther@suse.de>
38015 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38017 2017-02-17 Marek Polacek <polacek@redhat.com>
38019 PR middle-end/79536
38020 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38021 (fold_negate_expr): New wrapper.
38023 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38025 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38026 Correct terminology and de-emphasize pre-standard behavior.
38028 2017-02-16 Alan Modra <amodra@gmail.com>
38030 PR rtl-optimization/79286
38031 * ira.c (def_dominates_uses): New function.
38032 (update_equiv_regs): Don't create an equivalence for insns that
38033 may trap where the register def does not dominate the use.
38035 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38037 PR rtl-optimization/78127
38038 * lra.c (lra): Call lra_eliminate before finish the loop after
38041 2017-02-16 Richard Biener <rguenther@suse.de>
38043 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38045 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38046 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38047 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38048 (isl_val_int_from_wi): New function.
38049 (extract_affine_gmp): Rename to ...
38050 (extract_affine_wi): ... this, take a widest_int.
38051 (extract_affine_int): Just wrap extract_affine_wi.
38052 (add_param_constraints): Use isl_val_int_from_wi.
38053 (add_loop_constraints): Likewise, and extract_affine_wi.
38055 2017-02-15 Jeff Law <law@redhat.com>
38057 PR middle-end/79521
38058 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38059 ira_init_register_move_cost_if_necessary.
38061 2017-02-15 Martin Sebor <msebor@redhat.com>
38063 PR middle-end/32003
38064 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38065 removed in a prior commit.
38067 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38069 PR tree-optimization/79347
38070 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38071 counters during peeling.
38073 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38075 * Makefile.in (site.exp): Remove "set ISLVER".
38077 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38080 * real.c (real_from_integer): Call real_convert even for decimal.
38082 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38085 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38087 2017-02-14 Andrew Pinski <apinski@cavium.com>
38089 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38090 cores and change the partno/implementer to be correct.
38091 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38092 the 'B" as the implementer.
38093 * config/aarch64/aarch64-tune.md: Regenerate.
38095 2017-02-14 Carl Love <cel@us.ibm.com>
38097 * config/rs6000/rs6000.c: Add case statement entry to make the
38098 xvcvuxdsp built-in argument unsigned.
38099 * config/rs6000/vsx.md: Fix the source and return operand types so they
38100 match the instruction definitions from the ISA document. Fix typo
38101 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38104 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38107 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38108 member early_clobber_alts.
38109 * lra-lives.c (reg_early_clobber_p): New.
38110 (process_bb_lives): Use it.
38111 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38112 (debug_operand_data): Initialize early_clobber_alts.
38113 (setup_operand_alternative): Set up early_clobber_alts.
38114 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38115 alternatives to new_insn_reg.
38116 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38118 (lra_update_insn_regno_info): Pass the new arg.
38120 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38122 PR middle-end/79505
38123 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38124 (new_oacc_loop_raw): Don't clear already cleared fields.
38127 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38128 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38129 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38131 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38134 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38136 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38139 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38140 the extra instruction to the right place to store 128-bit constant
38143 2017-02-14 Martin Sebor <msebor@redhat.com>
38145 PR middle-end/79448
38146 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38147 warning for strings of unknown length.
38149 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38151 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38153 2017-02-14 Jeff Law <law@redhat.com>
38156 * ira-costs.c (scan_one_insn): Initialize register move costs
38157 for pseudos seen in USE/CLOBBER insns.
38159 PR tree-optimization/79095
38160 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38161 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38162 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38163 if the operands are known to be not equal, then the resulting range
38165 (intersect_ranges): If the new range is ~[0,0] and the old range is
38166 wide, then prefer ~[0,0].
38167 * tree-vrp.c (overflow_comparison_p_1): New function.
38168 (overflow_comparison_p): New function.
38169 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38170 if NAME is used in an overflow test.
38171 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38172 overflow check that can be expressed as an equality test, then adjust
38173 ops to be that equality test.
38175 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38177 * config/s390/s390-builtin-types.def: Remove flags argument.
38178 * config/s390/s390.c (s390_init_builtins): Likewise.
38180 2017-02-14 Martin Liska <mliska@suse.cz>
38182 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38183 vector. Fix trailing white spaces.
38185 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38187 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38190 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38192 PR rtl-optimization/68664
38193 * config/arm/arm.c (arm_sched_can_speculate_insn):
38194 New function. Declare prototype.
38195 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38197 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38199 PR rtl-optimization/68664
38200 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38202 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38204 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38206 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38207 max skip bytes for function, loop and jump.
38209 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38211 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38212 ABS_EXPR for gimple dump.
38214 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38217 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38219 PR tree-optimization/79408
38220 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38221 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38222 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38223 also if rhs1 is INTEGER_CST.
38225 2017-02-14 Richard Biener <rguenther@suse.de>
38227 PR middle-end/79432
38228 * tree-into-ssa.c (insert_phi_nodes): When the function can
38229 have abnormal edges rewrite SSA names with broken use-def
38230 dominance out of SSA and register them for PHI insertion.
38232 2017-02-13 Martin Sebor <msebor@redhat.com>
38234 PR middle-end/79496
38235 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38236 clearing info.nowrite flag when snprintf size argument is a range.
38238 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38240 * cprop.c (cprop_jump): Add missing space in string literal.
38241 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38242 (get_constraint_for_component_ref): Likewise.
38243 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38244 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38245 * lra-constraints.c (process_alt_operands): Likewise.
38246 * ipa-inline.c (inline_small_functions): Likewise.
38247 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38248 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38249 * trans-mem.c (diagnose_tm_1_op): Likewise.
38250 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38251 (grid_parallel_clauses_gridifiable): Likewise.
38253 * config/nvptx/mkoffload.c (process): Add space in between
38256 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38257 "MOD4_SSE_REGS" and "ALL_REGS".
38259 * spellcheck.c (test_data): Add , in between "foo" and "food".
38261 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38264 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38265 boundary crossing check and subsequent code generation agree.
38267 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38269 * config/aarch64/aarch64.c (has_memory_op): Delete.
38270 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38273 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38275 PR rtl-optimization/79388
38276 PR rtl-optimization/79450
38277 * combine.c (distribute_notes): When removing TEM_INSN for which
38278 corresponding dest has last value recorded, invalidate that last
38281 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38283 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38284 of explicit '@'. Add missing assembly comment marker on branch costs
38287 2017-02-13 Nathan Sidwell <nathan@acm.org>
38289 * gengtype-lex.l (<in_struct>): Add '/'.
38291 2017-02-13 Martin Liska <mliska@suse.cz>
38294 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38296 2017-02-13 Richard Biener <rguenther@suse.de>
38298 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38300 * configure: Re-generate.
38301 * config.in: Likewise.
38302 * graphite-dependences.c: Simplify as if
38303 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38304 * graphite-isl-ast-to-gimple.c: Likewise.
38305 * graphite-optimize-isl.c: Likewise.
38306 * graphite-poly.c: Likewise.
38307 * graphite-sese-to-poly.c: Likewise.
38308 * graphite.h: Likewise.
38309 * toplev.c: Include isl/version.h and use isl_version () for
38310 printing the ISL version.
38311 * doc/install.texi: Update ISL requirement.
38313 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38315 * doc/standards.texi (Standards): Update reference to
38318 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38320 * doc/extend.texi (Named Address Spaces): sourceware.org now
38322 * doc/install.texi (Binaries): Ditto.
38325 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38327 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38328 standard terminology "stringize"/"stringizing" throughout.
38329 * doc/cppinternals.texi: Likewise.
38331 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38333 * doc/extend.texi: Fix some spelling mistakes and typos.
38334 * doc/invoke.texi: Likewise.
38336 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38339 * params.def (inline-min-speedup) Change from 10 to 8.
38341 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38343 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38346 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38349 * ipa-inline-analysis.c (get_minimal_bb): New function.
38350 (record_modified): Use it.
38351 (remap_edge_change_prob): Handle also ancestor functions.
38353 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38355 * doc/contrib.texi (Contributors): Remove broken link into
38356 the Mauve CVS repository.
38358 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38360 PR middle-end/79454
38361 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38362 result computation whenever lhs doesn't have vector mode, not
38363 just when it has BLKmode.
38365 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38367 * doc/makefile.texi (profiledbootstrap): Refer to the
38368 installation instructions only in textual form.
38370 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38373 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38375 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38377 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38378 (Specific): Update mingw-w64 reference.
38380 (Specific): Remove broken link to Renesas RX processor.
38382 2017-02-10 Richard Biener <rguenther@suse.de>
38384 * toplev.c (process_options): Do not mention obsolete graphite
38385 options when printing sorry message about missing graphite support.
38386 Mention -floop-nest-optimize.
38388 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38390 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38391 (vtst_p16): Likewise.
38392 (vtstq_p8): Likewise.
38393 (vtstq_p16): Likewise.
38395 (vtstq_p64): Likewise.
38396 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38397 (vset_lane_p64): New.
38398 (vsetq_lane_p64): New.
38400 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38402 PR tree-optimization/79411
38403 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38404 stmt operands are SSA_NAMEs used in abnormal phis.
38405 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38408 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38411 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38414 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38416 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38418 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38420 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38424 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38425 not arbitrary TREE_CONSTANT.
38428 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38429 "omp declare target link" attribute unless is_global_var.
38430 * omp-offload.c (find_link_var_op): Likewise.
38432 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38433 Chung-Lin Tang <cltang@codesourcery.com>
38435 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38437 (gimplify_adjust_omp_clauses): Don't delete TILE.
38438 (gimplify_omp_for): Deal with TILE.
38439 * internal-fn.c (expand_GOACC_TILE): New function.
38440 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38442 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38443 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38445 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38446 avoid DIV for outermost collapse var.
38447 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38448 Remove out of date comments, fix whitespace.
38449 * omp-general.c (omp_extract_for_data): Deal with tiling.
38450 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38451 adjust OLF_DIM_BASE value.
38452 (struct omp_for_data): Add tiling field.
38453 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38454 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38455 for auto loops. Remove default auto determining, moved to
38456 oacc_loop_fixed_partitions.
38457 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38458 stmts, add e_mask field.
38459 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38460 (oacc_thread_numbers): Use oacc_dim_call.
38461 (oacc_xform_tile): New.
38462 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38463 (finish_oacc_loop): Adjust for ifns vector.
38464 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38465 add case for GOACC_TILE fns.
38466 (oacc_loop_xform_loop): Delete.
38467 (oacc_loop_process): Iterate over call list directly, and add
38468 handling for GOACC_TILE fns.
38469 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38471 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38472 vector partitioning to outer loops. Assign 2 partitions to loops
38473 when available. Add TILE handling.
38474 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38475 (execite_oacc_device_lower): Process GOACC_TILE fns,
38476 ignore unknown specs.
38477 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38478 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38479 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38481 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38483 * configure.ac (ACX_BUGURL): Update.
38484 * configure: Regenerate.
38486 2017-02-09 Richard Biener <rguenther@suse.de>
38488 PR tree-optimization/69823
38489 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38490 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38492 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38494 * config/arc/arc-c.def: Add __NPS400__ definition.
38495 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38496 (TARGET_NPS400): Define.
38498 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38500 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38502 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38503 pointer, arch_info.
38504 (arc_cpu_types): Fill the arch_info field with a pointer into the
38505 arc_arch_types table.
38506 (arc_selected_cpu): Declare.
38507 * config/arc/arc.c (arc_selected_cpu): Make global.
38508 (arc_selected_arch): Delete.
38509 (arc_base_cpu): Delete.
38510 (arc_override_options): Remove references to deleted variables,
38511 update access to arch information.
38512 (ARC_OPT): Update access to arch information.
38513 (ARC_OPTX): Likewise.
38514 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38515 (TARGET_ARC600): Update access to arch information.
38516 (TARGET_ARC601): Likewise.
38517 (TARGET_ARC700): Likewise.
38518 (TARGET_EM): Likewise.
38519 (TARGET_HS): Likewise.
38520 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38523 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38526 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38527 condition/operands for integer GE/LE/GEU/LEU operations.
38529 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38531 PR translation/79397
38532 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38535 2017-02-08 Martin Jambor <mjambor@suse.cz>
38538 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38539 whether allocation happened.
38540 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38541 nothing was allocated.
38543 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38545 PR tree-optimization/79408
38546 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38547 constant, but SSA_NAME with a known integer range, use the minimum
38548 of that range instead of op1 to determine if modulo can be replaced
38549 with its first operand.
38551 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38553 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38555 2017-02-08 Richard Biener <rguenther@suse.de>
38557 PR tree-optimization/71824
38558 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38559 Check all loops contained in the merged region.
38561 2017-02-07 Andrew Pinski <apinski@cavium.com>
38563 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38565 2017-02-07 Andrew Pinski <apinski@cavium.com>
38567 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38568 (thunderxt88): Likewise.
38569 (thunderxt81): Disable LSE and change v8.1 to v8.
38570 (thunderxt83): Likewise.
38572 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38573 Richard Biener <rguenther@suse.de>
38575 PR middle-end/79399
38576 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38577 type from int to size_t.
38578 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38580 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38582 PR rtl-optimization/79386
38583 * cprop.c (bypass_conditional_jumps): Initialize
38584 bypass_last_basic_block already before splitting bbs after
38585 unconditional traps...
38586 (bypass_conditional_jumps): ... rather than here.
38589 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38590 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38591 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38592 fix -masm=intel patterns.
38594 2017-02-07 Richard Biener <rguenther@suse.de>
38596 PR tree-optimization/79256
38597 PR middle-end/79278
38598 * builtins.c (get_object_alignment_2): Use min_align_of_type
38599 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38600 and ADJUST_FIELD_ALIGN.
38602 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38604 * doc/tm.texi: Regenerate.
38605 * stor-layout.c (layout_decl): Adjust.
38606 (update_alignment_for_field): Likewise.
38607 (place_field): Likewise.
38608 (min_align_of_type): Likewise.
38609 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38610 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38611 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38612 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38613 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38614 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38615 * config/i386/i386.c (x86_field_alignment): Likewise.
38616 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38617 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38618 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38619 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38620 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38621 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38625 2017-01-30 Richard Biener <rguenther@suse.de>
38627 PR tree-optimization/79256
38628 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38629 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38632 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38634 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38635 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38636 builtins to SImode and emit a zero-extend, if necessary.
38638 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38640 * docs/invoke.texi (RISC-V Options): Alphabetize.
38642 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38644 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38647 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38649 * config/riscv/riscv.c: New file.
38650 * common/config/riscv/riscv-common.c: Likewise.
38651 * config.gcc: Likewise.
38652 * config/riscv/constraints.md: Likewise.
38653 * config/riscv/elf.h: Likewise.
38654 * config/riscv/generic.md: Likewise.
38655 * config/riscv/linux.h: Likewise.
38656 * config/riscv/multilib-generator: Likewise.
38657 * config/riscv/peephole.md: Likewise.
38658 * config/riscv/pic.md: Likewise.
38659 * config/riscv/predicates.md: Likewise.
38660 * config/riscv/riscv-builtins.c: Likewise.
38661 * config/riscv/riscv-c.c: Likewise.
38662 * config/riscv/riscv-ftypes.def: Likewise.
38663 * config/riscv/riscv-modes.def: Likewise.
38664 * config/riscv/riscv-opts.h: Likewise.
38665 * config/riscv/riscv-protos.h: Likewise.
38666 * config/riscv/riscv.h: Likewise.
38667 * config/riscv/riscv.md: Likewise.
38668 * config/riscv/riscv.opt: Likewise.
38669 * config/riscv/sync.md: Likewise.
38670 * config/riscv/t-elf-multilib: Likewise.
38671 * config/riscv/t-linux: Likewise.
38672 * config/riscv/t-linux-multilib: Likewise.
38673 * config/riscv/t-riscv: Likewise.
38674 * configure.ac: Likewise.
38675 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38676 Waterman as RISC-V maintainers.
38677 * doc/install.texi: Add RISC-V entries.
38678 * doc/invoke.texi: Add RISC-V options section.
38679 * doc/md.texi: Add RISC-V constraints section.
38680 * configure: Regenerated.
38682 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38685 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38686 false values to be constant vectors with all 0 or all 1 bits set.
38687 (vcondu<mode><mode>): Likewise.
38688 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38690 (fpmask_comparison_operator): Update comment.
38691 (vecint_comparison_operator): New predicate.
38692 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38693 vector conditionals when the true and false values are constant
38694 vectors with all 0 bits or all 1 bits set.
38696 2017-02-06 Martin Sebor <msebor@redhat.com>
38698 PR tree-optimization/79376
38699 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38701 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38703 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38704 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38705 to simplify split condition.
38707 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38709 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38710 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38713 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38715 PR rtl-optimization/68664
38716 * target.def (can_speculate_insn): New hook.
38717 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38718 * doc/tm.texi: Regenerate.
38719 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38720 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38721 (rs6000_sched_can_speculate_insn): New function.
38723 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38725 PR tree-optimization/79284
38726 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38727 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38728 vectorizable_mask_load_store, vectorizable_operation,
38729 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38730 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38731 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38732 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38733 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38734 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38735 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38736 is_gimple_assign (stmt). Replace another such test with
38737 is_gimple_assign (stmt).
38739 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38742 * config/avr/avr.c (rtl-iter.h): Include it.
38743 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38744 (avr_legitimate_combined_insn): ...and implementation.
38746 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38748 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38749 * config/s390/s390.c (s390_const_operand_ok)
38750 (s390_canonicalize_comparison, s390_extract_part)
38751 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38752 (s390_contiguous_bitmask_p, s390_rtx_costs)
38753 (legitimize_pic_address): Likewise.
38754 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38755 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38756 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38757 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38758 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38760 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38762 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38763 REGNO($0) == REGNO($1).
38765 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38767 * config/s390/linux.h(SIZE_TYPE): Add comment.
38769 2017-02-06 Julian Brown <julian@codesourcery.com>
38770 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38771 Virendra Pathak <virendra.pathak@broadcom.com>
38773 * config/aarch64/aarch64-cores.def: Change the scheduler
38775 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38776 * config/aarch64/thunderx2t99.md: New file.
38778 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38780 * doc/standards.texi (Go Language): Update link to language
38783 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38785 * tree-eh.c (lower_resx): Sanitize profile.
38786 (cleanup_empty_eh_move_lp): Likewise.
38788 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38791 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38793 * cfgloopmanip.h (loop_version): Update prototype.
38794 * modulo-sched.c (sms_schedule): Update call of loop_version.
38795 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38796 * tree-parloops.c (gen_parallel_loop): Likewise.
38797 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38798 * tree-ssa-loop-split.c (split_loop): Likewise.
38799 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
38800 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
38802 2017-02-05 Martin Liska <mliska@suse.cz>
38805 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
38807 (print_operand_address): Initialize a struct to zero.
38809 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38811 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
38812 garbage collector only in textual form.
38814 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38816 * doc/extend.texi (x86 specific memory model extensions for
38817 transactional memory): Simplify a phrase.
38819 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
38822 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
38823 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
38824 (atomic_storedi_1): Likewise.
38826 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38828 PR tree-optimization/79338
38829 * tree-parloops.c (gather_scalar_reductions): Don't call
38830 vect_analyze_loop_form for loop->inner before destroying loop's
38833 2017-02-03 Martin Sebor <msebor@redhat.com>
38835 PR tree-optimization/79327
38836 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38837 when precision has resulted in leading zeros.
38838 (format_integer): Adjust the likely counter to assume an unknown
38839 argument that may be zero is non-zero.
38841 2017-02-03 Jason Merrill <jason@redhat.com>
38844 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38845 avoid copying non-taken branch.
38847 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38849 PR tree-optimization/79340
38850 * tree-vect-loop.c (vectorizable_reduction): Release
38851 vec_defs elements after safe_splicing them into other vectors.
38854 PR tree-optimization/79327
38855 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
38856 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
38858 (format_integer): Use wide_int_to_tree instead of build_int_cst
38859 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38860 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38861 of shortest and longest sequence.
38863 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
38865 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
38866 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
38868 2017-02-03 Walter Lee <walt@tilera.com>
38871 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
38872 after initial stackframe link reg save.
38873 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
38875 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38878 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
38879 wu for stxssp alternative.
38881 2017-02-03 Martin Sebor <msebor@redhat.com>
38883 PR tree-optimization/79352
38884 * gimple-fold.c (get_range_strlen): Add argument.
38885 (get_range_strlen): Change return type to bool.
38886 (get_maxval_strlen): Pass in a dummy argument.
38887 * gimple-fold.h (get_range_strlen): Change return type to bool.
38888 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38889 * tree.h (array_at_struct_end_p): Add argument.
38890 * tree.c (array_at_struct_end_p): Handle it.
38892 2017-02-03 Martin Liska <mliska@suse.cz>
38895 * multiple_target.c (create_dispatcher_calls): Redirect edge
38896 from a caller of a dispatcher.
38897 (expand_target_clones): Make the clones local.
38898 (ipa_target_clone): Do both target clones and resolvers.
38899 (ipa_dispatcher_calls): Remove the pass.
38900 (pass_dispatcher_calls::gate): Likewise.
38901 (make_pass_dispatcher_calls): Likewise.
38902 * passes.def (pass_target_clone): Put as very first IPA early
38905 2017-02-03 Martin Liska <mliska@suse.cz>
38907 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38908 in case of a function with ifunc attribute.
38910 2017-02-03 Martin Liska <mliska@suse.cz>
38912 * cgraph.c (cgraph_node::dump): Dump function version info.
38913 * symtab.c (symtab_node::dump_base): Add missing new line.
38915 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38917 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38918 (ifcombine_ifandif): Use it.
38920 2017-02-03 Martin Liska <mliska@suse.cz>
38922 * doc/invoke.texi: Document default value for
38923 use-after-scope-direct-emission-threshold.
38925 2017-02-03 Martin Liska <mliska@suse.cz>
38927 PR tree-optimization/79339
38928 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38929 (format_floating): Likewise.
38931 2017-02-03 Martin Liska <mliska@suse.cz>
38934 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38936 (ipa_node_params_t::remove): Likewise.
38937 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38938 initialization from removed ipa_node_params_t::insert.
38939 (ipa_node_params::~ipa_node_params): Move from removed
38940 ipa_node_params_t::release.
38941 * symbol-summary.h (symbol_summary::m_released): New member.
38942 Do not release a summary twice. Do not allow to call finalizer
38943 for types of a summary that live in GGC memory.
38945 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38947 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38950 2017-02-02 Martin Sebor <msebor@redhat.com>
38952 PR middle-end/79275
38953 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38954 (format_string): Tighten up the range of output for non-constant
38955 strings and correct the expected range for wide non-constant strings.
38957 2017-02-02 Martin Sebor <msebor@redhat.com>
38959 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38961 PR middle-end/32003
38962 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38964 (-fdump-tree-@var): Add to index and document how to come up
38965 with pass-specific option and dump file names.
38966 (-fdump-passes): Clarify where to look for output.
38968 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38970 PR middle-end/77445
38971 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38972 statistics of the analyzed path; allow threading for speed when
38973 any of BBs along the path are optimized for speed.
38975 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38977 PR middle-end/78468
38978 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38979 settings of the virtual registers.
38982 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38984 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38985 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38988 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38990 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38991 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38993 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38995 * config/s390/s390.md: Add missing comments with the expanded
38997 * config/s390/vector.md: Likewise.
38998 * config/s390/vx-builtins.md: Likewise.
39000 2017-02-02 Jakub Jelinek <jakub@redhat.com>
39003 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
39004 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
39005 conditions on a single line.
39007 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39009 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39010 __S390_VX__ to __VX__.
39012 2017-02-01 Andrew Pinski <apinski@cavium.com>
39014 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39015 stmt_info to record_stmt_cost.
39016 (vect_get_known_peeling_cost): Pass stmt_info if known to
39018 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39019 cpu_vector_cost field into
39020 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39021 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39022 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39023 splitting of scalar_stmt_cost and vec_stmt_cost.
39024 (thunderx_vector_cost): Likewise.
39025 (cortexa57_vector_cost): LIkewise.
39026 (exynosm1_vector_cost): Likewise.
39027 (xgene1_vector_cost): Likewise.
39028 (thunderx2t99_vector_cost): Improve after the splitting of the two
39030 (aarch64_builtin_vectorization_cost): Update for the splitting of
39031 scalar_stmt_cost and vec_stmt_cost.
39033 2017-02-01 Torvald Riegel <triegel@redhat.com>
39034 Richard Henderson <rth@redhat.com>
39036 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39037 conditional on existance of a fast atomic load.
39038 * optabs-query.c (can_atomic_load_p): New function.
39039 * optabs-query.h (can_atomic_load_p): Declare it.
39040 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39041 no fast atomic load is available for the particular size of access.
39042 (expand_atomic_compare_and_swap): Likewise.
39043 (expand_atomic_load): Likewise.
39044 (expand_atomic_store): Likewise.
39045 (expand_atomic_fetch_op): Likewise.
39046 * testsuite/lib/target-supports.exp
39047 (check_effective_target_sync_int_128): Remove x86 because it provides
39048 no fast atomic load.
39049 (check_effective_target_sync_int_128_runtime): Likewise.
39051 2017-02-01 Richard Biener <rguenther@suse.de>
39053 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39054 (graphite_transform_loops): Provide opt-info for optimized nests.
39055 * tree-parloop.c (parallelize_loops): Provide opt-info for
39056 parallelized loops.
39058 2017-02-01 Richard Biener <rguenther@suse.de>
39060 PR middle-end/79315
39061 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39062 was not set before.
39064 2017-02-01 Richard Biener <rguenther@suse.de>
39066 PR tree-optimization/71824
39067 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39068 Verify the loops are valid in the merged SESE region.
39069 (scop_detection::can_represent_loop_1): Check analyzing the
39070 evolution of the number of iterations in the region succeeds.
39072 2017-01-31 Ian Lance Taylor <iant@golang.org>
39074 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39075 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39077 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39079 PR preprocessor/79210
39080 * input.c (get_substring_ranges_for_loc): Replace line_width
39081 assertion with error-handling.
39083 2017-01-31 Richard Biener <rguenther@suse.de>
39085 PR tree-optimization/77318
39086 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39087 (create_pw_aff_from_tree): Take loop parameter.
39088 (add_condition_to_pbb): Pass loop of the condition to
39089 create_pw_aff_from_tree.
39091 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39093 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39094 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39096 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39100 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39102 (convert_int_to_float128): Likewise.
39103 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39104 (convert_int_to_float128): Likewise.
39105 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39106 (UNSPEC_IEEE128_CONVERT): Likewise.
39107 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39108 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39109 Use local variables for IBM extended format.
39110 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39111 (fix_trunc<mode>si2_fprs): Likewise.
39112 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39113 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39114 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39115 to know that we can now have integers of all sizes in vector
39117 (fix<uns>_<mode>di2_hw): Likewise.
39118 (float<uns>_<mode>si2_hw): Likewise.
39119 (fix_<mode>si2_hw): Likewise.
39120 (fixuns_<mode>si2_hw): Likewise.
39121 (float<uns>_<mode>di2_hw): Likewise.
39122 (float_<mode>di2_hw): Likewise.
39123 (float_<mode>si2_hw): Likewise.
39124 (floatuns_<mode>di2_hw): Likewise.
39125 (floatuns_<mode>si2_hw): Likewise.
39126 (xscvqp<su>wz_<mode>): Delete, no longer used.
39127 (xscvqp<su>dz_<mode>): Likewise.
39128 (xscv<su>dqp_<mode>): Likewise.
39129 (ieee128_mfvsrd_64bit): Likewise.
39130 (ieee128_mfvsrd_32bit): Likewise.
39131 (ieee128_mfvsrwz): Likewise.
39132 (ieee128_mtvsrw): Likewise.
39133 (ieee128_mtvsrd_64bit): Likewise.
39134 (ieee128_mtvsrd_32bit): Likewise.
39136 2017-01-31 Martin Liska <mliska@suse.cz>
39139 * ipa-prop.c (ipa_free_all_node_params): Call release method
39140 instead of ~sumbol_summary to not to trigger double times
39143 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39145 PR tree-optimization/71691
39146 * bitmap.h (class auto_bitmap): New.
39147 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39148 is_maybe_undefined instead of ssa_undefined_value_p.
39150 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39152 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39153 __S390_ARCH_LEVEL__ to __ARCH__.
39155 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39157 PR tree-optimization/79267
39158 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39159 if should_remove_lhs_p is true.
39161 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39164 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39165 (add_alignment_attribute): New.
39166 (base_type_die): Add alignment attribute.
39167 (subrange_type_die): Likewise.
39168 (modified_type_die): Likewise.
39169 (gen_array_type_die): Likewise.
39170 (gen_descr_array_type_die: Likewise.
39171 (gen_enumeration_type_die): Likewise.
39172 (gen_subprogram_die): Likewise.
39173 (gen_variable_die): Likewise.
39174 (gen_field_die): Likewise.
39175 (gen_ptr_to_mbr_type_die): Likewise.
39176 (gen_struct_or_union_type_die): Likewise.
39177 (gen_subroutine_type_die): Likewise.
39178 (gen_typedef_die): Likewise.
39179 (base_type_cmp): Compare alignment attribute.
39181 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39184 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39185 (setb_unsigned) New pattern for setb with CCUNS.
39186 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39187 subfc./subfe sequence to avoid overflow problems. Generate a
39188 shorter sequence with cmpld/setb for power9.
39189 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39190 for generating subfc. instruction.
39191 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39192 now uses this instruction.
39194 2017-01-30 Ian Lance Taylor <iant@google.com>
39197 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39198 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39200 2017-01-30 Martin Sebor <msebor@redhat.com>
39202 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39203 Move constant to the right of a relational operator.
39204 (get_mpfr_format_length, format_character, format_string): Ditto.
39205 (should_warn_p, maybe_warn): Same.
39207 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39209 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39212 * asan.c (get_translation_unit_decl): Remove function.
39213 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39215 2017-01-30 Martin Liska <mliska@suse.cz>
39217 PR gcov-profile/79259
39218 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39219 -fprofile-generate.
39221 2017-01-30 Martin Liska <mliska@suse.cz>
39224 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39225 Initialize variables with NULL value.
39227 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39230 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39232 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39234 2017-01-30 Richard Biener <rguenther@suse.de>
39236 PR tree-optimization/79276
39237 * tree-vrp.c (process_assert_insertions): Properly adjust common
39238 when removing a duplicate.
39240 2017-01-30 Richard Biener <rguenther@suse.de>
39242 PR tree-optimization/79256
39243 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39244 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39246 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39248 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39251 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39252 ("*r<noxa>sbg_<mode>_sll_bitmask")
39253 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39254 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39255 Use contiguous_bitmask_nowrap_operand.
39257 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39260 * config/rs6000/altivec.h (vec_xl): Revise #define.
39261 (vec_xst): Likewise.
39263 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39265 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39267 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39269 PR rtl-optimization/79194
39270 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39271 traps before call to bypass_conditional_jumps.
39273 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39275 PR tree-optimization/71374
39276 * lra-constraints.c (check_conflict_input_operands): New.
39277 (match_reload): Use it.
39279 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39282 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39283 account to calculate conflict_set.
39285 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39287 PR rtl-optimization/78559
39288 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39289 other_insn in combine.
39291 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39293 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39294 uint16_type_node for BT_UINT16.
39296 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39298 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39299 "RTL Tests" to menu.
39300 (GIMPLE Tests): New node.
39301 (RTL Tests): New node.
39303 2017-01-27 Richard Biener <rguenther@suse.de>
39305 PR tree-optimization/79245
39306 * tree-loop-distribution.c (distribute_loop): Apply cost
39307 modeling also to detected patterns.
39309 2017-01-27 Richard Biener <rguenther@suse.de>
39311 PR tree-optimization/71433
39312 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39313 (compare_assert_loc): New function.
39314 (process_assert_insertions): Sort and optimize assert locations
39315 to remove duplicates and push down identical assertions on
39316 edges to their destination block.
39318 2017-01-27 Richard Biener <rguenther@suse.de>
39320 PR tree-optimization/79244
39321 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39322 out SSA names even if abnormal.
39324 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39326 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39327 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39328 instead of MPFR_RNDN.
39330 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39333 * arm.c (arm_option_override): Don't call build_target_option_node
39334 until after doing all option overrides.
39335 (arm_valid_target_attribute_tree): Likewise.
39337 2017-01-27 Martin Liska <mliska@suse.cz>
39339 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39340 for {cd}tors and C++ {cd}tors.
39342 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39344 * config/s390/s390.md ("*setmem_long_and")
39345 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39347 2017-01-26 Martin Sebor <msebor@redhat.com>
39349 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39352 2017-01-26 Martin Sebor <msebor@redhat.com>
39354 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39355 HAVE_DFmode before using XFmode or DFmode.
39356 (parse_directive): Avoid using the z length modifier to avoid
39357 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39359 PR middle-end/78703
39360 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39361 to accept adjustment as an array.
39362 (get_int_range): New function.
39363 (struct directive): Make width and prec arrays.
39364 (directive::set_width, directive::set_precision): Call get_int_range.
39365 (format_integer, format_floating): Handle width and precision ranges.
39366 (format_string, parse_directive): Same.
39368 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39371 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39372 comdat_type_p set, just clone them, but keep the children in the
39376 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39377 which have direct callers with -fvar-tracking-assignments enabled
39379 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39380 inside of type units.
39382 2017-01-26 Martin Sebor <msebor@redhat.com>
39384 PR middle-end/78703
39385 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39387 (struct format_result): Replace number_chars, number_chars_min,
39388 and number_chars_max with a single member of struct result_range.
39390 (format_result::operator+=): Adjust.
39391 (struct fmtresult): Remove bounded. Handle likely and unlikely
39393 (fmtresult::adjust_for_width_or_precision): New function.
39394 (fmtresult:type_max_digits): New function.
39395 (bytes_remaining): Handle likely and unlikely counters.
39396 (min_bytes_remaining): Remove.
39397 (format_percent): Simplify.
39398 (format_integer, format_floating): Set likely and unlikely counters.
39399 (get_string_length, format_character, format_string): Same.
39400 (format_plain, should_warn_p): New function.
39401 (maybe_warn): Call should_warn_p. Update diagnostic messages
39402 and handle those for all directives, including plain strings.
39403 (format_directive): Handle likely and unlikely counters.
39404 Remove unnecessary quoting from diagnostics. Add an informational
39406 (add_bytes): Remove.
39407 (pass_sprintf_length::compute_format_length): Simplify.
39408 (try_substitute_return_value): Handle likely and unlikely counters.
39410 2017-01-26 Carl Love <cel@us.ibm.com>
39412 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39413 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39415 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39418 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39419 endianess for subregs into account.
39420 * lra-constraints.c (lra_constraints): Do risky transformations
39421 always on the first iteration.
39422 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39423 last_call_used_reg_set.
39424 (process_bb_lives): Define and use last_call_used_reg_set.
39425 * lra.c (lra): Always continue after lra_constraints on the first
39428 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39430 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39432 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39434 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39436 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39437 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39438 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39439 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39440 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39441 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39442 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39443 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39444 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39446 2017-01-26 Marek Polacek <polacek@redhat.com>
39449 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39450 for the third operand.
39452 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39454 PR middle-end/79236
39455 * omp-low.c (struct omp_context): Add simt_stmt field.
39456 (scan_omp_for): Return omp_context *.
39457 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39458 context to the _simt_ SIMD stmt.
39459 (lower_omp_for): For combined SIMD with sibling _simt_
39460 SIMD, make sure to use the same decls in _looptemp_
39461 clauses as in the sibling.
39463 2017-01-26 David Sherwood <david.sherwood@arm.com>
39465 PR middle-end/79212
39466 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39469 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39472 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39473 emit fld b; fld a; if possible.
39475 * brig-builtins.def: Update copyright years.
39476 * config/arm/arm_acle_builtins.def: Update copyright years.
39478 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39481 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39482 constraint instead of o for the stxsd instruction.
39484 2017-01-25 Carl Love <cel@us.ibm.com>
39486 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39487 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39489 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39491 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39493 2017-01-25 Richard Biener <rguenther@suse.de>
39495 PR tree-optimization/69264
39496 * target.def (vector_alignment_reachable): Improve documentation.
39497 * doc/tm.texi: Regenerate.
39498 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39500 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39501 earlier changes with respect to TYPE_USER_ALIGN.
39502 (vector_alignment_reachable_p): Likewise. Improve dumping.
39504 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39507 * config/arm/arm.md (xordi3): Force constant operand into a register
39510 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39512 * doc/invoke.texi (-fstore-merging): Correct default optimization
39513 levels at which it is enabled.
39514 (-O): Move -fstore-merging from list to...
39517 2017-01-25 Richard Biener <rguenther@suse.de>
39520 * omp-expand.c: Include debug.h.
39521 (expand_omp_taskreg): Make sure to generate early debug before
39522 outlining anything from a function.
39523 (expand_omp_target): Likewise.
39524 (grid_expand_target_grid_body): Likewise.
39526 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39529 * asan.c (get_translation_unit_decl): New function.
39530 (asan_add_global): Extract modules file name from globals
39531 TRANSLATION_UNIT_DECL name.
39533 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39536 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39537 for long calls with APCS frame and VFP.
39539 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39541 * cfg.c (original_copy_tables_initialized_p): New function.
39542 * cfg.h (original_copy_tables_initialized_p): New decl.
39543 * cfgrtl.c (relink_block_chain): Guard the call to
39544 free_original_copy_tables with a call to
39545 original_copy_tables_initialized_p.
39546 * cgraph.h (symtab_node::native_rtl_p): New decl.
39547 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39548 (symtab_node::needed_p): Don't assert for early assembly output
39549 for __RTL functions.
39550 (cgraph_node::finalize_function): Set "force_output" for __RTL
39552 (cgraph_node::analyze): Bail out early for __RTL functions.
39553 (analyze_functions): Update assertion to support __RTL functions.
39554 (cgraph_node::expand): Bail out early for __RTL functions.
39555 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39557 * function.h (struct function): Update comment for field
39559 * gimple-expr.c: Include "tree-pass.h".
39560 (gimple_has_body_p): Return false for __RTL functions.
39561 * Makefile.in (OBJS): Add run-rtl-passes.o.
39562 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39564 (gcc::pass_manager::get_clean_slate): New accessor.
39565 * passes.c: Include "insn-addr.h".
39566 (should_skip_pass_p): Add logging. Update logic for running
39567 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39568 property-provider override so it is only done for gimple passes.
39570 (skip_pass): New function.
39571 (execute_one_pass): Call skip_pass when skipping passes.
39572 * read-md.c (md_reader::read_char): Support filtering
39573 the input to a subset of line numbers.
39574 (md_reader::md_reader): Initialize fields
39575 m_first_line and m_last_line.
39576 (md_reader::read_file_fragment): New function.
39577 * read-md.h (md_reader::read_file_fragment): New decl.
39578 (md_reader::m_first_line): New field.
39579 (md_reader::m_last_line): New field.
39580 * read-rtl-function.c (function_reader::create_function): Only
39581 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39582 curr_properties. Set DECL_INITIAL to a dummy block.
39583 (read_rtl_function_body_from_file_range): New function.
39584 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39586 * run-rtl-passes.c: New file.
39587 * run-rtl-passes.h: New file.
39589 2017-01-24 Jeff Law <law@redhat.com>
39591 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39594 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39596 PR tree-optimization/79159
39597 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39598 (record_nonwrapping_iv): Improve boundary using above function if no
39599 value range information.
39601 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39602 Martin Jambor <mjambor@suse.cz>
39604 * brig-builtins.def: New file.
39605 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39606 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39607 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39608 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39609 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39610 * builtin-types.def (BT_INT8): New.
39611 (BT_INT16): Likewise.
39612 (BT_UINT8): Likewise.
39613 (BT_UINT16): Likewise.
39614 (BT_FN_ULONG): Likewise.
39615 (BT_FN_UINT_INT): Likewise.
39616 (BT_FN_UINT_ULONG): Likewise.
39617 (BT_FN_UINT_LONG): Likewise.
39618 (BT_FN_UINT_PTR): Likewise.
39619 (BT_FN_ULONG_PTR): Likewise.
39620 (BT_FN_INT8_FLOAT): Likewise.
39621 (BT_FN_INT16_FLOAT): Likewise.
39622 (BT_FN_UINT32_FLOAT): Likewise.
39623 (BT_FN_UINT16_FLOAT): Likewise.
39624 (BT_FN_UINT8_FLOAT): Likewise.
39625 (BT_FN_UINT64_FLOAT): Likewise.
39626 (BT_FN_UINT16_UINT32): Likewise.
39627 (BT_FN_UINT32_UINT16): Likewise.
39628 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39629 (BT_FN_INT_PTR_INT): Likewise.
39630 (BT_FN_UINT_PTR_UINT): Likewise.
39631 (BT_FN_LONG_PTR_LONG): Likewise.
39632 (BT_FN_ULONG_PTR_ULONG): Likewise.
39633 (BT_FN_VOID_UINT64_UINT64): Likewise.
39634 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39635 (BT_FN_INT8_INT8_INT8): Likewise.
39636 (BT_FN_INT16_INT16_INT16): Likewise.
39637 (BT_FN_INT_INT_INT): Likewise.
39638 (BT_FN_UINT_FLOAT_UINT): Likewise.
39639 (BT_FN_FLOAT_UINT_UINT): Likewise.
39640 (BT_FN_ULONG_UINT_UINT): Likewise.
39641 (BT_FN_ULONG_UINT_PTR): Likewise.
39642 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39643 (BT_FN_UINT_UINT_UINT): Likewise.
39644 (BT_FN_VOID_UINT_PTR): Likewise.
39645 (BT_FN_UINT_UINT_PTR: Likewise.
39646 (BT_FN_UINT32_UINT64_PTR): Likewise.
39647 (BT_FN_INT_INT_UINT_UINT): Likewise.
39648 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39649 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39650 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39651 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39652 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39653 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39654 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39655 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39656 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39657 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39658 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39659 * doc/frontends.texi: List BRIG FE.
39660 * doc/install.texi (Testing): Add BRIG tesring requirements.
39661 * doc/invoke.texi (Overall Options): Mention BRIG.
39662 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39664 2017-01-24 Richard Biener <rguenther@suse.de>
39666 PR translation/79208
39667 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39669 2017-01-24 Martin Jambor <mjambor@suse.cz>
39672 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39673 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39674 and known_contexts.
39676 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39678 PR middle-end/79123
39679 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39680 casts from signed to unsigned really don't have a range.
39682 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39684 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39685 GMP_RNDx for compatiblity.
39687 2017-01-24 Martin Liska <mliska@suse.cz>
39690 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39691 that would prevent us to call alloca with -1 as argument.
39693 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39695 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39696 Avoid -Wformat-security warning.
39698 2017-01-23 Andrew Pinski <apinski@cavium.com>
39700 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39703 2017-01-23 Martin Sebor <msebor@redhat.com>
39705 PR middle-end/78703
39706 * gimple-ssa-sprintf.c (warn_level): New global.
39707 (format_integer): Use it here and throughout the rest of the file.
39708 Use the same switch to compute sign as base.
39709 (maybe_warn): New function.
39710 (format_directive): Factor out warnings into maybe_warn.
39711 Add debugging output. Use warn_level.
39712 (add_bytes): Use warn_level.
39713 (pass_sprintf_length::compute_format_length): Add debugging output.
39714 (try_substitute_return_value): Same.
39715 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39717 PR middle-end/78703
39718 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39719 (struct fmtresult, format_integer, format_floating): Adjust.
39720 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39721 (get_string_length, format_string,format_directive): Same.
39722 (pass_sprintf_length::compute_format_length): Same.
39723 (try_substitute_return_value): Simplify slightly.
39725 PR middle-end/78703
39726 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39727 (fmtresult::operator+=): Outlined.
39728 (struct fmtresult): Add ctors.
39729 (struct conversion_spec): Rename...
39730 (struct directive): ...to this. Add and remove data members.
39731 (directive::set_width, directive::set_precision): New functions.
39732 (format_percent): Use fmtresult ctor.
39733 (get_width_and_precision): Remove.
39734 (format_integer): Make naming changes. Avoid computing width and
39736 (format_floating): Same. Adjust indentation.
39737 (format_character, format_none): New functions.
39738 (format_string): Moved character handling to format_character.
39739 (format_directive): Remove arguments, change return type.
39740 (parse_directive): New function.
39741 (pass_sprintf_length::compute_format_length): Move directive
39742 parsing to parse_directive.
39744 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39746 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39747 (assign_assembler_name_if_needed): ... this.
39748 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39749 (assign_assembler_name_if_needed): ... this.
39750 (free_lang_data_in_cgraph): Adjust callers.
39751 * cgraphunit.c (cgraph_node::analyze): Likewise.
39752 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39755 2017-01-23 Richard Biener <rguenther@suse.de>
39757 PR tree-optimization/79088
39758 PR tree-optimization/79188
39759 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39760 resetting loop bounds after last path deletion. Reset loop
39761 bounds of the target loop, make code match the comments.
39762 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39763 Make sure loops need no fixups.
39765 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39767 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39768 exponent support with double type for first argument.
39769 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39770 type returned by __builtin_vec_extract_sig,
39771 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39772 functions from "vector int" to "vector unsigned int" or from
39773 "vector long long int" to "vector unsigned long long int".
39774 Changed type returned by __builtin_vec_extract_exp,
39775 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39776 functions from "vector int" to "vector unsigned int" or from
39777 "vector long long int" to "vector unsigned long long int".
39778 Changed return type of __builtin_vec_test_data_class,
39779 __builtin_vec_test_data_class_sp, and
39780 __builtin_vec_test_data_class_dp from "vector int" to
39781 "vector bool int" or from "vector long long int" to "vector bool
39782 long long int" and changed second argument type from "unsigned
39783 int" to "int". Added new overloaded function forms "vector float
39784 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39785 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39786 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39787 double, vector unsigned long long int)" and "vector double
39788 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39789 long int)". Changed return type of
39790 __builtin_scalar_test_data_class and
39791 __builtin_scalar_test_data_class_sp and
39792 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39793 int" and changed second argument from "unsigned int" to "int".
39794 Changed type returned by __builtin_scalar_test_neg,
39795 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39796 from "int" to "bool int". Added new overloaded function form
39797 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39798 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
39799 exponent double-precision with floating point first argument.
39800 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
39801 documentation of scalar_test_data_class, scalar_test_neg,
39802 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
39803 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
39804 vec_test_data_class built-in functions to reflect refinements in
39805 their type signatures.
39807 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
39809 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
39811 (aarch64_elf_asm_destructor): Likewise.
39813 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
39815 PR rtl-optimization/78634
39816 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
39817 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
39818 * ifcvt.c (noce_try_cmove): Add missing cost check.
39820 PR rtl-optimization/71724
39821 * combine.c (if_then_else_cond): Look for situations where it is
39822 beneficial to undo the work of one of the recursive calls.
39824 2017-01-23 Bin Cheng <bin.cheng@arm.com>
39826 PR tree-optimization/70754
39827 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39828 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39829 combined stmt before it if not NULL.
39830 (combine_chains): Process refs reversely and compute dominance point
39833 2017-01-23 Martin Liska <mliska@suse.cz>
39835 PR tree-optimization/79196
39836 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39837 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39839 (strlen_optimize_stmt): Call the renamed function.
39841 2017-01-23 Michael Matz <matz@suse.de>
39843 PR tree-optimization/78384
39844 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39846 2017-01-23 Richard Biener <rguenther@suse.de>
39848 PR tree-optimization/79186
39849 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39850 both incoming edges before moving an assert.
39852 2017-01-23 Martin Jambor <mjambor@suse.cz>
39854 * ipa-prop.c (load_from_param_1): Removed.
39855 (load_from_unmodified_param): Bits from load_from_param_1 put back
39857 (load_from_param): Removed.
39858 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
39859 with stmt. Reverted back to use of load_from_unmodified_param.
39861 2017-01-23 Martin Jambor <mjambor@suse.cz>
39864 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
39865 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
39866 field a pointer to garbage collected vector, mark lattices and
39867 ipcp_orig_node with GTY((skip)).
39868 (ipa_get_param_count): Adjust to descriptors being a pointer.
39869 (ipa_get_param): Likewise.
39870 (ipa_get_type): Likewise.
39871 (ipa_get_param_move_cost): Likewise.
39872 (ipa_set_param_used): Likewise.
39873 (ipa_get_controlled_uses): Likewise.
39874 (ipa_set_controlled_uses): Likewise.
39875 (ipa_is_param_used): Likewise.
39876 (ipa_node_params_t): Move into garbage collector. New methods insert
39878 (ipa_node_params_sum): Annotate wth GTY(()).
39879 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
39881 (ipa_load_from_parm_agg): Adjust declaration.
39882 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
39883 * ipa-profile.c (ipa_profile): Likewise.
39884 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39885 (ipa_populate_param_decls): Make descriptors parameter garbage
39887 (ipa_dump_param): Adjust to descriptors being a pointer.
39888 (ipa_alloc_node_params): Likewise.
39889 (ipa_initialize_node_params): Likewise.
39890 (load_from_param_1): Make descriptors parameter garbage collected.
39891 (load_from_unmodified_param): Likewise.
39892 (load_from_param): Likewise.
39893 (ipa_load_from_parm_agg): Likewise.
39894 (ipa_node_params::~ipa_node_params): Removed.
39895 (ipa_free_all_node_params): Remove call to delete operator.
39896 (ipa_node_params_t::insert): New.
39897 (ipa_node_params_t::remove): Likewise.
39898 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39899 copy known_csts and known_contexts vectors.
39900 (ipa_read_node_info): Adjust to descriptors being a pointer.
39901 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39903 (ipcp_transform_function): Make descriptors variable garbage
39906 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39908 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39909 * config/i386/avx512dqintrin.h: Ditto.
39910 * config/i386/avx512fintrin.h: Ditto.
39911 * config/i386/i386.c: Handle new builtins.
39912 * config/i386/i386-builtin.def: Add new builtins.
39913 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39914 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39916 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39917 Martin Liska <mliska@suse.cz>
39919 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39920 * asan.c (asan_expand_poison_ifn): Support stores and use
39921 appropriate ASAN report function.
39922 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39923 * internal-fn.def (ASAN_POISON_USE): Declare.
39924 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39925 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39926 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39927 ASAN_POISON calls w/o LHS.
39928 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39929 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39930 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39931 * gimplify.c (asan_poison_variables): Add attribute
39932 use_after_scope_memory to variables that really needs to live
39934 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39935 having the attribute.
39937 2017-01-23 Martin Liska <mliska@suse.cz>
39939 * asan.c (create_asan_shadow_var): New function.
39940 (asan_expand_poison_ifn): Likewise.
39941 * asan.h (asan_expand_poison_ifn): New declaration.
39942 * internal-fn.c (expand_ASAN_POISON): Likewise.
39943 * internal-fn.def (ASAN_POISON): New builtin.
39944 * sanopt.c (pass_sanopt::execute): Expand
39945 asan_expand_poison_ifn.
39946 * tree-inline.c (copy_decl_for_dup_finish): Make function
39948 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39949 * tree-ssa.c (is_asan_mark_p): New function.
39950 (execute_update_addresses_taken): Rewrite local variables
39951 (identified just by use-after-scope as addressable) into SSA.
39953 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39955 * doc/install.texi (Specific): opensource.apple.com uses https
39956 now. Remove trailing slash.
39958 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39960 * README.Portability: Remove note on an Irix compatibility issue.
39962 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39964 * gcov.c (INCLUDE_ALGORITHM): Define.
39965 (INCLUDE_VECTOR): Define.
39966 No longer include <vector> and <algorithm> directly.
39968 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39970 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39972 * doc/invoke.texi (Code Gen Options): Ditto.
39974 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39977 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39979 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39981 rtl-optimization/79125
39982 * cprop.c (local_cprop_pass): Handle cases where we make an
39983 unconditional trap.
39985 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39989 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39990 read from, for big endian.
39992 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39994 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39995 register pauth builtins for LP64 only.
39997 2017-01-20 Marek Polacek <polacek@redhat.com>
40000 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
40003 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40005 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
40007 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
40008 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40009 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40011 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40014 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40015 in big-endian mode when they are not a single duplicated value.
40017 2017-01-20 Richard Biener <rguenther@suse.de>
40019 * BASE-VER: Bump to 7.0.1.
40021 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40023 * omp-low.c (omplow_simd_context): New struct. Use it...
40024 (lower_rec_simd_input_clauses): ...here and...
40025 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40026 references to idx, lane, max_vf, is_simt.
40028 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40030 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40033 2017-01-20 Martin Jambor <mjambor@suse.cz>
40035 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40036 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40038 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40039 (GTFILES): Rename hsa.c to hsa-common.c.
40040 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40041 * hsa-dump.c: Likewise.
40042 * hsa-gen.c: Likewise.
40043 * hsa-regalloc.c: Likewise.
40044 * ipa-hsa.c: Likewise.
40045 * omp-expand.c: Likewise.
40046 * omp-low.c: Likewise.
40047 * toplev.c: Likewise.
40049 2017-01-20 Marek Polacek <polacek@redhat.com>
40052 * doc/invoke.texi: Document -Wduplicated-branches.
40053 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40054 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40055 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40056 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40057 return 0 only when not OEP_LEXICOGRAPHIC.
40058 (fold_build_cleanup_point_expr): Use the expression
40059 location when building CLEANUP_POINT_EXPR.
40060 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40061 * tree.c (add_expr): Handle error_mark_node.
40063 2017-01-20 Martin Liska <mliska@suse.cz>
40066 * tree-profile.c (init_ic_make_global_vars): Do not call
40068 (gimple_init_gcov_profiler): Likewise.
40070 2017-01-20 Martin Liska <mliska@suse.cz>
40073 * cgraph.h (maybe_create_reference): Remove argument and
40075 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40077 * ipa-cp.c (create_specialized_node): Likewise.
40078 * symtab.c (symtab_node::maybe_create_reference): Handle
40079 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40081 2017-01-20 Martin Liska <mliska@suse.cz>
40083 * read-rtl-function.c (function_reader::create_function): Use
40084 build_decl instread of build_decl_stat.
40086 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40088 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40089 * config/i386/avx512dqintrin.h: Ditto.
40090 * config/i386/avx512fintrin.h: Ditto.
40091 * config/i386/i386-builtin-types.def: Add new types.
40092 * config/i386/i386.c: Handle new types.
40093 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40094 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40095 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40096 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40097 (__builtin_ia32_kshiftridi): New.
40098 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40100 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40104 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40105 define to rs6000_init_stack_protect_guard.
40106 (rs6000_init_stack_protect_guard): New function.
40108 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40109 Yunqiang Su <yunqiang.su@imgtec.com>
40111 * config.gcc (supported_defaults): Add madd4.
40112 (with_madd4): Add validation.
40113 (all_defaults): Add madd4.
40114 * config/mips/mips.opt (mmadd4): New option.
40115 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40117 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40119 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40120 (ISA_HAS_FUSED_MADD4): Likewise.
40121 * doc/invoke.texi (-mmadd4): Document the new option.
40122 * doc/install.texi (--with-madd4): Document the new option.
40124 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40126 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40127 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40128 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40129 (aarch64_init_pauth_hint_builtins): New.
40130 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40131 (aarch64_expand_builtin): Expand new builtins.
40133 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40135 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40136 * combine-stack-adj.c (no_unhandled_cfa): Handle
40137 REG_CFA_TOGGLE_RA_MANGLE.
40138 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40139 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40140 info for return address signing.
40141 (aarch64_expand_epilogue): Likewise.
40143 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40145 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40146 * config/aarch64/aarch64-protos.h
40147 (aarch64_return_address_signing_enabled): New declaration.
40148 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40150 (aarch64_expand_prologue): Sign return address before it's pushed onto
40152 (aarch64_expand_epilogue): Authenticate return address fetched from
40154 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40155 (aarch64_attributes): New function attributes for "sign-return-address".
40156 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40157 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40158 ("*do_return"): Generate combined instructions according to key index.
40159 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40160 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40162 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40163 * config/aarch64/aarch64.opt (msign-return-address=): New.
40164 * doc/extend.texi (AArch64 Function Attributes): Documents
40165 "sign-return-address=".
40166 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40168 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40170 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40171 overall option summary.
40173 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40175 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40176 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40177 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40178 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40180 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40182 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40183 -mpower9-minmax by default for -mcpu=power9.
40184 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40185 128-bit floating point.
40187 2017-01-20 Alan Modra <amodra@gmail.com>
40189 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40190 optimizing for size.
40192 2017-01-20 Alan Modra <amodra@gmail.com>
40195 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40196 for strcmp and strncmp from corresponding builtin decl.
40198 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40200 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40201 instead of i386/rtems-64.h.
40202 * config/i386/rtems-64.h: Remove.
40204 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40208 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40210 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40212 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40214 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40215 Change int to HOST_WIDE_INT.
40216 * config/aarch64/aarch64-protos.h
40217 (aarch64_simd_gen_const_vector_dup): Likewise.
40218 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40220 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40222 * langhooks-def.h (lhd_type_for_size): New decl.
40223 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40224 * langhooks.c (lhd_type_for_size): New function, taken from
40227 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40229 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40230 define_bypass for CR latency.
40231 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40232 (power9-alu2): Add define_bypass for CR latency.
40234 (power9-mul): Update insn latency.
40235 (power9-mul-compare): Update insn latency, bypass latency and remove
40238 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40240 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40242 * config/aarch64/aarch64.md
40243 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40244 aarch64_nopcrelative_literal_loads.
40245 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40247 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40249 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40250 TARGET_LOONGSON_3A.
40251 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40253 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40256 * config.gcc (supported_defaults): Add lxc1-sxc1.
40257 (with_lxc1_sxc1): Add validation.
40258 (all_defaults): Add lxc1-sxc1.
40259 * config/mips/mips.opt (mlxc1-sxc1): New option.
40260 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40262 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40263 __mips_no_lxc1_sxc1.
40264 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40265 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40266 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40268 2017-01-19 Richard Biener <rguenther@suse.de>
40270 PR tree-optimization/72488
40271 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40272 sure to restore SSA info.
40273 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40275 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40277 PR rtl-optimization/79121
40278 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40279 of the inner type when shifting an extended value.
40281 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40284 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40285 interposable aliases.
40287 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40290 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40291 Use the evmergelohi instruction.
40292 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40293 (mov_sitf_e500_subreg8_2_be): Likewise.
40294 (mov_sitf_e500_subreg12_2_le): Likewise.
40295 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40296 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40297 (mov_sitf_e500_subreg8_2_le): Likewise.
40298 (mov_sitf_e500_subreg12_2_be): Likewise.
40300 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40302 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40303 attribute from vecsimple to vecperm.
40304 (altivec_vbpermq2): Likewise.
40306 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40309 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40311 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40312 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40313 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40314 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40315 case where N arg is SIZE_MAX.
40316 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40317 (cmpstrsi): Add pattern.
40319 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40321 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40322 __builtin_vec_revb builtins.
40323 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40324 built-in functions to support generation of the ISA 3.0 XXBR<x>
40325 vector byte reverse instructions.
40326 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40327 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40328 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40329 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40330 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40331 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40332 (P9V_BUILTIN_VEC_REVB): Likewise.
40333 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40334 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40335 (p9_xxbrq_v16qi): Likewise.
40336 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40337 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40338 (p9_xxbrh_v8hi): Likewise.
40339 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40340 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40341 vec_revb built-in functions.
40343 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40345 PR rtl-optimization/78952
40346 * config/i386/i386.md (any_extract): New code iterator.
40347 (*insvqi_2): Use any_extract for source operand.
40348 (*insvqi_3): Use any_shiftrt for source operand.
40350 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40352 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40354 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40356 2017-01-18 Matthias Klose <doko@ubuntu.com>
40358 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40360 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40362 * config/rs6000/altivec.h (vec_bperm): Change #define.
40363 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40364 (altivec_vbpermq2): New define_insn.
40365 (altivec_vbpermd): Likewise.
40366 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40367 function interface.
40368 (VBPERMD): Likewise.
40369 (VBPERM): New polymorphic function interface.
40370 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40371 Add entries for P9V_BUILTIN_VEC_VBPERM.
40372 * doc/extend.texi: Add interfaces for vec_bperm.
40374 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40376 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40377 first letter of error messages.
40378 (s390_resolve_overloaded_builtin): Likewise.
40379 * config/s390/s390.c (s390_expand_builtin): Likewise.
40380 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40381 (s390_valid_target_attribute_inner_p): Likewise.
40382 * config/s390/s390.md ("tabort"): Likewise.
40384 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40386 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40387 (ISA_AVOID_DIV_HILO): New macro.
40388 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40389 (ISA_HAS_DDIV): Likewise.
40391 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40393 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40395 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40397 * doc/invoke.texi (fabi-version): Spelling fix.
40399 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40402 * doc/invoke.texi (fabi-version): Mention mangling fix for
40405 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40408 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40410 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40413 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40414 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40416 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40418 (stack_protect_test): Ditto.
40419 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40420 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40422 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40423 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40424 -mstack-protector-guard-offset=.
40425 (RS/6000 and PowerPC Options): Ditto.
40427 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40429 * config/i386/i386.h (MASK_CLASS_P): New define.
40430 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40431 there are no registers from different register sets also when
40432 mask registers are used. Update function comment.
40433 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40434 to (*k/*r) and (*k/*km) alternatives.
40436 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40438 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40439 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40440 (EH_RETURN_HANDLER_RTX): New define.
40441 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40442 Force frame pointer in EH return functions.
40443 (aarch64_expand_epilogue): Add barrier for eh_return.
40444 (aarch64_final_eh_return_addr): Remove.
40445 (aarch64_eh_return_handler_rtx): New function.
40446 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40448 (aarch64_eh_return_handler_rtx): New prototype.
40450 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40452 * config/rs6000/altivec.h (vec_rlmi): New #define.
40453 (vec_vrlnm): Likewise.
40454 (vec_rlnm): Likewise.
40455 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40456 (UNSPEC_VRLNM): Likewise.
40457 (VIlong): New mode iterator.
40458 (altivec_vrl<VI_char>mi): New define_insn.
40459 (altivec_vrl<VI_char>nm): Likewise.
40460 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40462 (VRLDNM): Likewise.
40463 (RLNM): New polymorphic function entry.
40464 (VRLWMI): New monomorphic function entry.
40465 (VRLDMI): Likewise.
40466 (RLMI): New polymorphic function entry.
40467 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40468 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40469 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40472 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40475 * dwarf2out.c (field_byte_offset): Restore the
40476 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40477 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40478 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40481 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40484 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40486 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40489 * dwarf2out.c (add_data_member_location_attribute): For constant
40490 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40491 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40492 DW_AT_byte_size attributes.
40494 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40496 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40497 after forcing to constant memory when the code model is medium.
40499 2017-01-17 Julia Koval <julia.koval@intel.com>
40502 * config/i386/avx512fintrin.h
40503 (_mm512_i32gather_ps): Change __addr type to void const*.
40504 (_mm512_mask_i32gather_ps): Ditto.
40505 (_mm512_i32gather_pd): Ditto.
40506 (_mm512_mask_i32gather_pd): Ditto.
40507 (_mm512_i64gather_ps): Ditto.
40508 (_mm512_mask_i64gather_ps): Ditto.
40509 (_mm512_i64gather_pd): Ditto.
40510 (_mm512_mask_i64gather_pd): Ditto.
40511 (_mm512_i32gather_epi32): Ditto.
40512 (_mm512_mask_i32gather_epi32): Ditto.
40513 (_mm512_i32gather_epi64): Ditto.
40514 (_mm512_mask_i32gather_epi64): Ditto.
40515 (_mm512_i64gather_epi32): Ditto.
40516 (_mm512_mask_i64gather_epi32): Ditto.
40517 (_mm512_i64gather_epi64): Ditto.
40518 (_mm512_mask_i64gather_epi64): Ditto.
40519 (_mm512_i32scatter_ps): Change __addr type to void*.
40520 (_mm512_mask_i32scatter_ps): Ditto.
40521 (_mm512_i32scatter_pd): Ditto.
40522 (_mm512_mask_i32scatter_pd): Ditto.
40523 (_mm512_i64scatter_ps): Ditto.
40524 (_mm512_mask_i64scatter_ps): Ditto.
40525 (_mm512_i64scatter_pd): Ditto.
40526 (_mm512_mask_i64scatter_pd): Ditto.
40527 (_mm512_i32scatter_epi32): Ditto.
40528 (_mm512_mask_i32scatter_epi32): Ditto.
40529 (_mm512_i32scatter_epi64): Ditto.
40530 (_mm512_mask_i32scatter_epi64): Ditto.
40531 (_mm512_i64scatter_epi32): Ditto.
40532 (_mm512_mask_i64scatter_epi32): Ditto.
40533 (_mm512_i64scatter_epi64): Ditto.
40534 (_mm512_mask_i64scatter_epi64): Ditto.
40535 * config/i386/avx512pfintrin.h
40536 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40537 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40538 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40539 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40540 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40541 (_mm512_prefetch_i32scatter_ps): Ditto.
40542 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40543 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40544 (_mm512_prefetch_i64scatter_pd): Ditto.
40545 (_mm512_prefetch_i64scatter_ps): Ditto.
40546 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40547 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40548 * config/i386/avx512vlintrin.h
40549 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40550 (_mm_mmask_i32gather_ps): Ditto.
40551 (_mm256_mmask_i32gather_pd): Ditto.
40552 (_mm_mmask_i32gather_pd): Ditto.
40553 (_mm256_mmask_i64gather_ps): Ditto.
40554 (_mm_mmask_i64gather_ps): Ditto.
40555 (_mm256_mmask_i64gather_pd): Ditto.
40556 (_mm_mmask_i64gather_pd): Ditto.
40557 (_mm256_mmask_i32gather_epi32): Ditto.
40558 (_mm_mmask_i32gather_epi32): Ditto.
40559 (_mm256_mmask_i32gather_epi64): Ditto.
40560 (_mm_mmask_i32gather_epi64): Ditto.
40561 (_mm256_mmask_i64gather_epi32): Ditto.
40562 (_mm_mmask_i64gather_epi32): Ditto.
40563 (_mm256_mmask_i64gather_epi64): Ditto.
40564 (_mm_mmask_i64gather_epi64): Ditto.
40565 (_mm256_i32scatter_ps): Change __addr type to void*.
40566 (_mm256_mask_i32scatter_ps): Ditto.
40567 (_mm_i32scatter_ps): Ditto.
40568 (_mm_mask_i32scatter_ps): Ditto.
40569 (_mm256_i32scatter_pd): Ditto.
40570 (_mm256_mask_i32scatter_pd): Ditto.
40571 (_mm_i32scatter_pd): Ditto.
40572 (_mm_mask_i32scatter_pd): Ditto.
40573 (_mm256_i64scatter_ps): Ditto.
40574 (_mm256_mask_i64scatter_ps): Ditto.
40575 (_mm_i64scatter_ps): Ditto.
40576 (_mm_mask_i64scatter_ps): Ditto.
40577 (_mm256_i64scatter_pd): Ditto.
40578 (_mm256_mask_i64scatter_pd): Ditto.
40579 (_mm_i64scatter_pd): Ditto.
40580 (_mm_mask_i64scatter_pd): Ditto.
40581 (_mm256_i32scatter_epi32): Ditto.
40582 (_mm256_mask_i32scatter_epi32): Ditto.
40583 (_mm_i32scatter_epi32): Ditto.
40584 (_mm_mask_i32scatter_epi32): Ditto.
40585 (_mm256_i32scatter_epi64): Ditto.
40586 (_mm256_mask_i32scatter_epi64): Ditto.
40587 (_mm_i32scatter_epi64): Ditto.
40588 (_mm_mask_i32scatter_epi64): Ditto.
40589 (_mm256_i64scatter_epi32): Ditto.
40590 (_mm256_mask_i64scatter_epi32): Ditto.
40591 (_mm_i64scatter_epi32): Ditto.
40592 (_mm_mask_i64scatter_epi32): Ditto.
40593 (_mm256_i64scatter_epi64): Ditto.
40594 (_mm256_mask_i64scatter_epi64): Ditto.
40595 (_mm_i64scatter_epi64): Ditto.
40596 (_mm_mask_i64scatter_epi64): Ditto.
40597 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40598 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40599 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40600 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40601 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40602 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40603 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40604 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40605 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40606 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40607 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40608 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40609 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40610 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40611 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40612 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40613 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40614 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40615 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40616 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40617 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40618 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40619 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40620 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40621 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40622 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40623 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40624 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40625 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40626 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40627 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40628 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40629 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40630 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40631 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40632 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40633 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40634 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40635 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40636 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40637 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40638 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40639 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40640 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40641 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40642 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40643 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40644 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40645 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40646 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40647 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40648 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40649 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40650 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40651 definitions accordingly.
40653 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40654 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40657 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40660 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40663 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40664 hard regs for inner regno.
40666 2017-01-17 Martin Liska <mliska@suse.cz>
40669 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40670 assumption and add comment.
40672 2017-01-17 Nathan Sidwell <nathan@acm.org>
40674 * ipa-visibility.c (localize_node): New function, broken out of ...
40675 (function_and_variable_visibility): ... here. Call it.
40677 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40679 PR middle-end/77445
40680 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40681 correctly set frequency of oudgoing edge.
40682 (duplicate_thread_path): Fix profile updating.
40684 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40687 * configure.ac: Add GCC_BASE_VER.
40688 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40689 version from BASE-VER file.
40690 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40691 (gcc.o): Depend on $(BASEVER).
40692 * common.opt (dumpfullversion): New option.
40693 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40694 * doc/invoke.texi: Document -dumpfullversion.
40695 * doc/install.texi: Document --with-gcc-major-version-only.
40696 * configure: Regenerated.
40698 2017-01-17 Richard Biener <rguenther@suse.de>
40700 PR tree-optimization/71433
40701 * tree-vrp.c (register_new_assert_for): Merge same asserts
40702 on all incoming edges.
40703 (process_assert_insertions_for): Handle insertions at the
40706 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40708 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40709 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40711 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40714 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40717 2017-01-17 Alan Modra <amodra@gmail.com>
40720 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40721 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40722 symbolic stack limit when pic.
40724 2017-01-16 Martin Sebor <msebor@redhat.com>
40726 PR tree-optimization/78608
40727 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40729 2017-01-16 Jeff Law <law@redhat.com>
40732 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40733 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40734 for several include directories that may be relative to sysroot.
40735 * config/i386/x-mingw32 (gplus_includedir): Define.
40736 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40737 (native_system_includedir): Likewise.
40738 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40739 override if TARGET_SYSTEM_ROOT is defined.
40740 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40742 PR tree-optimization/79090
40743 PR tree-optimization/33562
40744 PR tree-optimization/61912
40745 PR tree-optimization/77485
40746 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40747 and computed trims into the dump file.
40749 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40751 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40753 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40756 * gimplify.c (gimplify_init_constructor): If want_value and
40757 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40761 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40762 sequence. Formatting fixes.
40763 (doloop_optimize): Formatting fixes.
40766 * gcc.c (debug_level_greater_than_spec_func): New function.
40767 (static_spec_functions): Add debug-level-gt spec function.
40768 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40770 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40771 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40772 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40773 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40774 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40775 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40777 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40779 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40780 QImode fixups to general and mask registers only.
40782 2017-01-16 Carl Love <cel@us.ibm.com>
40784 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40785 for built-in functions
40786 vector signed char vec_nabs (vector signed char)
40787 vector signed short vec_nabs (vector signed short)
40788 vector signed int vec_nabs (vector signed int)
40789 vector signed long long vec_nabs (vector signed long long)
40790 vector float vec_nabs (vector float)
40791 vector double vec_nabs (vector double)
40792 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40794 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40795 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40796 * doc/extend.texi: Update the documentation file for the new built-in
40799 2017-01-16 Martin Sebor <msebor@redhat.com>
40801 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
40804 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40806 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
40807 UNSPEC_VSX__XXSPLTD to require special splat handling.
40809 2017-01-16 David Malcolm <dmalcolm@redhat.com>
40812 * system.h: Poison strndup.
40814 2017-01-16 Alan Modra <amodra@gmail.com>
40817 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
40820 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
40822 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
40824 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
40826 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40827 call recog here. Assert that INSN_CODE (insn) is non-negative.
40829 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40832 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40834 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40835 in the currently scheduled RTL fragment.
40837 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40839 PR rtl-optimization/78751
40840 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40843 2017-01-14 Jeff Law <law@redhat.com>
40845 PR tree-optimization/79090
40846 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40847 variable length stores.
40848 (compute_trims): Delete dead assignment to *trim_tail.
40849 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
40852 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
40854 PR rtl-optimization/78626
40855 PR rtl-optimization/78727
40856 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
40857 of a block, and split such blocks after everything else is finished.
40859 2017-01-14 Alan Modra <amodra@gmail.com>
40862 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40863 target legitimate_combined_insn.
40864 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
40865 (rs6000_legitimate_combined_insn): New function.
40866 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
40868 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
40869 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
40870 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
40872 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
40874 * doc/frontends.texi (G++ and GCC): Remove references to Java.
40876 2017-01-13 Jeff Law <law@redhat.com>
40878 PR tree-optimization/33562
40879 PR tree-optimization/61912
40880 PR tree-optimization/77485
40881 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
40883 (delete_dead_assignment): Likewise.
40884 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40885 statement to delete_dead_call and delete_dead_assignment.
40887 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40890 * substring-locations.c (format_warning_va): Strengthen case 1 so
40891 that both endpoints of the substring must be within the format
40892 range for just the substring to be printed.
40894 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40896 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40897 * config/i386/i386.c (ix86_target_string): Add missing options
40898 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40899 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40900 flags_other and ix86_target_other to flags2_other. Display unknown
40902 (ix86_valid_target_attribute_inner_p): Add missing options and
40903 reorder options by implied ISAs, as in ix86_target_string.
40905 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40907 * hash-table.h (hash_table::too_empty_p): New function.
40908 (hash_table::expand): Use it.
40909 (hash_table::traverse): Likewise.
40910 (hash_table::empty_slot): Use sizeof (value_type) instead of
40911 sizeof (PTR) to convert bytes to elements. Shrink the table
40912 if the current size is excessive for the current number of
40915 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40917 * ira-costs.c (record_reg_classes): Break from the inner loop
40918 early once alt_fail is known to be true. Update outer loop
40919 handling accordingly.
40921 2017-01-13 Jeff Law <law@redhat.com>
40923 * tree-ssa-dse.c (decrement_count): New function.
40924 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40925 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40926 when we know the partially dead statement is a mem* function.
40928 PR tree-optimization/61912
40929 PR tree-optimization/77485
40930 * tree-ssa-dse.c: Include expr.h.
40931 (maybe_trim_constructor_store): New function.
40932 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40934 PR tree-optimization/33562
40935 PR tree-optimization/61912
40936 PR tree-optimization/77485
40937 * doc/invoke.texi: Document new dse-max-object-size param.
40938 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40939 * tree-ssa-dse.c: Include params.h.
40940 (dse_store_status): New enum.
40941 (initialize_ao_ref_for_dse): New, partially extracted from
40943 (valid_ao_ref_for_dse, normalize_ref): New.
40944 (setup_live_bytes_from_ref, compute_trims): Likewise.
40945 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40946 (maybe_trim_partially_dead_store): Likewise.
40947 (maybe_trim_complex_store): Likewise.
40948 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40949 Track what bytes live from the original store. Return tri-state
40950 for dead, partially dead or live.
40951 (dse_dom_walker): Add constructor, destructor and new private members.
40952 (delete_dead_call, delete_dead_assignment): New extracted from
40954 (dse_optimize_stmt): Make a member of dse_dom_walker.
40955 Use initialize_ao_ref_for_dse.
40957 PR tree-optimization/33562
40958 PR tree-optimization/61912
40959 PR tree-optimization/77485
40960 * sbitmap.h (bitmap_count_bits): Prototype.
40961 (bitmap_clear_range, bitmap_set_range): Likewise.
40962 * sbitmap.c (bitmap_clear_range): New function.
40963 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40965 2017-01-13 Martin Liska <mliska@suse.cz>
40968 * function.c (set_cfun): Add new argument force.
40969 * function.h (set_cfun): Likewise.
40970 * ipa-inline-transform.c (inline_call): Use the function when
40971 strict alising from is dropped for function we inline to.
40973 2017-01-13 Richard Biener <rguenther@suse.de>
40975 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40976 for dumping GIMPLE INTEGER_CSTs.
40978 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40980 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40981 to 201112L since C++17.
40983 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40986 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40987 if -fsanitize=kernel-address is present.
40989 2017-01-13 Richard Biener <rguenther@suse.de>
40991 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40992 as _Literal ( type ) number in case usual suffixes do not
40993 preserve all information.
40995 2017-01-13 Richard Biener <rguenther@suse.de>
40997 PR tree-optimization/77283
40998 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40999 and ssa-iterators.h.
41000 (is_feasible_trace): Implement a cost model based on joiner
41003 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
41006 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
41007 char or short to __float128/_Float128 directly.
41009 2017-01-12 Martin Sebor <msebor@redhat.com>
41011 to -Wformat-overflow.
41012 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41013 (min_bytes_remaining): Same.
41014 (get_string_length): Same.
41015 (format_string): Same.
41016 (format_directive): Same.
41018 (pass_sprintf_length::handle_gimple_call): Same.
41020 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41022 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41023 info.nowrite calls with no lhs that can't throw. Return bool
41024 whether gsi_remove has been called or not.
41025 (pass_sprintf_length::handle_gimple_call): Return bool whether
41026 try_substitute_return_value called gsi_remove. Formatting fix.
41027 (pass_sprintf_length::execute): Don't use gsi_remove if
41028 handle_gimple_call returned true.
41031 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41032 be removed due to side-effects, don't remove following barrier nor
41033 turn the successor edge into fallthru edge.
41035 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41038 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41039 element-reversing loads and stores as not swappable.
41041 2017-01-12 Nathan Sidwell <nathan@acm.org>
41042 Nicolai Stange <nicstange@gmail.com>
41044 * combine.c (try_combine): Don't ignore result of overlap checking
41045 loop. Combine overlap & asm check into single loop.
41047 2017-01-12 Richard Biener <rguenther@suse.de>
41049 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41050 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41052 2017-01-12 Richard Biener <rguenther@suse.de>
41054 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41055 and TS_TARGET_OPTION directly derive from TS_BASE.
41056 * tree-core.h (tree_optimization_option): Derive from tree_base.
41057 (tree_target_option): Likewise.
41059 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41061 * config/i386/i386.c (memory_address_length): Increase len
41062 only when rip_relative_addr_p returns false.
41064 2017-01-11 Julia Koval <julia.koval@intel.com>
41066 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41067 (OPTION_MASK_ISA_SGX_SET): New.
41068 (ix86_handle_option): Handle OPT_msgx.
41069 * config.gcc: Added sgxintrin.h.
41070 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41071 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41072 * config/i386/i386.c (ix86_target_string): Add -msgx.
41074 (ix86_option_override_internal): Handle new options.
41075 (ix86_valid_target_attribute_inner_p): Add sgx.
41076 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41077 * config/i386/i386.opt: Add msgx.
41078 * config/i386/sgxintrin.h: New file.
41079 * config/i386/x86intrin.h: Add sgxintrin.h.
41081 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41084 * fold-const.c (maybe_nonzero_address): Return 1 for function
41086 (tree_single_nonzero_warnv_p): Don't handle function local objects
41090 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41093 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41096 * opts.c: Include "spellcheck.h"
41097 (struct string_fragment): New struct.
41098 (struct edit_distance_traits<const string_fragment &>): New
41100 (get_closest_sanitizer_option): New function.
41101 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41103 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41105 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41107 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41108 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41109 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41110 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41111 for initial die_offset if dwarf_split_debug_info.
41112 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41113 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41115 (output_skeleton_debug_sections): Formatting fix. Use
41116 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41117 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41119 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41121 * config/arm/cortex-a53.md: Add bypasses for
41122 cortex_a53_r2f_cvt.
41123 (cortex_a53_r2f): Only use for transfers.
41124 (cortex_a53_f2r): Likewise.
41125 (cortex_a53_r2f_cvt): Add reservation for conversions.
41126 (cortex_a53_f2r_cvt): Likewise.
41128 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41130 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41131 to all inlined functions, change static to extern.
41133 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41136 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41138 (arm_assemble_integer): Likewise.
41140 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41142 * config.gcc: Use new awk script to check CPU, FPU and architecture
41143 parameters for --with-... options.
41144 * config/arm/parsecpu.awk: New file
41145 * config/arm/arm-cpus.in: New file.
41146 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41148 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41150 * config/arm/t-arm: Update dependency rules.
41151 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41152 of processing .def files.
41153 * config/arm/genopt.sh: Deleted.
41154 * config/arm/gentune.sh: Deleted.
41155 * config/arm/arm-cores.def: Deleted.
41156 * config/arm/arm-arches.def: Deleted.
41157 * config/arm/arm-fpus.def: Deleted.
41158 * config/arm/arm-tune.md: Regenerated.
41159 * config/arm/arm-tables.opt: Regenerated.
41160 * config/arm/arm-cpu.h: New generated file.
41161 * config/arm/arm-cpu-data.h: New generated file.
41162 * config/arm/arm-cpu-cdata.h: New generated file.
41164 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41167 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41169 (input_varpool_node): Unpack dynamically_initialized bit.
41171 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41173 PR rtl-optimization/79032
41174 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41175 the alignment of the adjusted memory reference against that of MODE,
41176 instead of the alignment of the original memory reference.
41178 2017-01-11 Martin Jambor <mjambor@suse.cz>
41180 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41182 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41183 decorated functions.
41185 2017-01-11 Richard Biener <rguenther@suse.de>
41187 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41188 set range/nonnull info for PHI results. Do not set it on
41189 stmts marked for removal.
41191 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41193 * expr.c (store_field): In the bitfield case, fetch the return value
41194 from the registers before applying a single big-endian adjustment.
41195 Always do a final load for a BLKmode value not larger than a word.
41197 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41200 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41201 that we correctly handle column numbers greater than
41202 LINE_MAP_MAX_COLUMN_NUMBER.
41204 2017-01-10 Martin Sebor <msebor@redhat.com>
41206 PR middle-end/78245
41207 * gimple-ssa-sprintf.c (get_destination_size): Call
41208 {init,fini}object_sizes.
41209 * tree-object-size.c (addr_object_size): Adjust.
41210 (pass_through_call): Adjust.
41211 (pass_object_sizes::execute): Adjust.
41212 * tree-object-size.h (fini_object_sizes): Declare.
41214 2017-01-10 Martin Sebor <msebor@redhat.com>
41216 PR tree-optimization/78775
41217 * builtins.c (get_size_range): Move...
41218 * calls.c: ...to here.
41219 (alloc_max_size): Accept zero argument.
41220 (operand_signed_p): Remove.
41221 (maybe_warn_alloc_args_overflow): Call get_size_range.
41222 * calls.h (get_size_range): Declare.
41224 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41226 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41227 from TI's devices.csv file as of September 2016.
41228 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41230 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41232 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41233 * doc/invoke.texi: Likewise.
41234 * doc/md.texi: Likewise.
41235 * doc/objc.texi: Likewise.
41237 2017-01-10 Joshua Conner <joshconner@google.com>
41239 * config/arm/fuchsia-elf.h: New file.
41240 * config/fuchsia.h: New file.
41241 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41242 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41244 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41246 2016-01-10 Richard Biener <rguenther@suse.de>
41248 PR tree-optimization/79034
41249 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41250 Propagate out degenerate PHIs in the joiner.
41252 2017-01-10 Martin Liska <mliska@suse.cz>
41254 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41255 (sort_congruence_classes_by_decl_uid): Likewise.
41256 (sort_congruence_class_groups_by_decl_uid): Likewise.
41257 (sem_item_optimizer::merge_classes): Sort class, groups in these
41258 classes and members in the groups by DECL_UID of declarations.
41259 This would make merge operations stable.
41261 2017-01-10 Martin Liska <mliska@suse.cz>
41263 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41264 usage of m_classes_vec.
41265 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41266 (sem_item_optimizer::get_group_by_hash): Likewise.
41267 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41268 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41269 (sem_item_optimizer::verify_classes): Likewise.
41270 (sem_item_optimizer::process_cong_reduction): Likewise.
41271 (sem_item_optimizer::dump_cong_classes): Likewise.
41272 (sem_item_optimizer::merge_classes): Likewise.
41273 * ipa-icf.h (congruence_class_hash): Rename from
41274 congruence_class_group_hash. Remove declaration of m_classes_vec.
41276 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41278 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41279 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41280 * config.gcc: Add avx512vpopcntdqintrin.h.
41281 * config/i386/avx512vpopcntdqintrin.h: New.
41282 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41283 * config/i386/i386-builtin-types.def: Add new types.
41284 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41285 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41286 __builtin_ia32_vpopcountq_v8di_mask): New.
41287 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41288 __AVX512VPOPCNTDQ__.
41289 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41290 (PTA_AVX512VPOPCNTDQ): Define.
41291 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41292 TARGET_AVX512VPOPCNTDQ_P): Define.
41293 * config/i386/i386.opt: Add mavx512vpopcntdq.
41294 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41295 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41297 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41299 PR middle-end/77484
41300 * predict.def (PRED_CALL): Set to 67.
41302 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41304 * expr.c (store_field): In the bitfield case, if the value comes from
41305 a function call and is of an aggregate type returned in registers, do
41306 not modify the field mode; extract the value in all cases if the mode
41307 is BLKmode and the size is not larger than a word.
41309 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41312 * config/i386/cpuid.h: Fix undefined behavior.
41314 2017-01-04 Jeff Law <law@redhat.com>
41316 PR tree-optimization/79007
41317 PR tree-optimization/67955
41318 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41319 conservative for pt.null when flag_non_call_exceptions is on.
41321 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41323 PR translation/79019
41324 PR translation/79020
41325 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41326 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41327 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41329 * config/avr/avr.opt (maccumulate-args): Likewise.
41330 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41331 * common.opt (freport-bug): Likewise.
41332 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41333 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41334 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41335 translatable string.
41336 * config/i386/i386.c (function_value_32): Likewise.
41337 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41338 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41340 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41341 * common/config/msp430/msp430-common.c (msp430_handle_option):
41343 * symtab.c (symtab_node::verify_base): Likewise.
41344 * opts.c (set_debug_level): Likewise.
41345 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41346 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41347 missing whitespace to translatable strings.
41348 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41349 * config/sh/superh.h: Likewise.
41350 * config/i386/xopintrin.h: Likewise.
41351 * config/i386/znver1.md: Likewise.
41352 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41353 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41354 * double-int.h (struct double_int): Likewise.
41355 * double-int.c (div_and_round_double): Likewise.
41356 * wide-int.cc: Likewise.
41357 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41358 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41359 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41360 (crossjumps_occurred): ... this.
41361 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41364 PR tree-optimization/78899
41365 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41366 returning bool return struct loop *, NULL for failure and the new
41368 (versionable_outer_loop_p): Don't version outer loop if it has
41369 dont_vectorized bit set.
41370 (tree_if_conversion): When versioning outer loop, ensure
41371 tree_if_conversion is performed also on the inner loop of the
41372 non-vectorizable outer loop copy.
41373 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41374 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41375 prevent vectorization of it.
41376 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41377 the outer loop vectorization of the non-scalar version is attempted
41378 before vectorization of the inner loop in scalar version. If
41379 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41380 vectorization of its inner loop.
41381 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41382 has 2 inner loops, rename also on edges from bb whose single pred
41383 is outer_loop->header. Fix typo in function comment.
41385 2017-01-09 Martin Sebor <msebor@redhat.com>
41388 * asan.c (asan_emit_stack_protection): Increase local buffer size
41389 to avoid snprintf truncation warning.
41391 2017-01-09 Andrew Pinski <apinski@cavium.com>
41393 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41394 to reference thunderx2t99 for the tuning structure
41395 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41397 (thunderx2t99_extra_costs): This.
41398 * config/aarch64/aarch64-tune.md: Regenerate.
41399 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41400 (vulcan_addrcost_table): This.
41401 (vulcan_regmove_cost): Rename to ...
41402 (thunderx2t99_regmove_cost): This.
41403 (vulcan_vector_cost): Rename to ...
41404 (thunderx2t99_vector_cost): this.
41405 (vulcan_branch_cost): Rename to ...
41406 (thunderx2t99_branch_cost): This.
41407 (vulcan_tunings): Rename to ...
41408 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41409 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41411 2017-01-09 Martin Jambor <mjambor@suse.cz>
41415 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41416 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41417 (propagate_vr_accross_jump_function): Use the above function for all
41418 value range computations for pass-through jump functions and type
41419 converasion from explicit value range values.
41420 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41421 parameters from TYPE_ARG_TYPES.
41422 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41423 (ipa_write_node_info): Stream type of the actual argument.
41424 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41426 2017-01-09 Martin Liska <mliska@suse.cz>
41429 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41430 (lookup_compiler): Do not show error message with have_E.
41432 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41434 PR tree-optimization/78938
41435 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41436 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41437 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41438 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41441 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41443 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41446 2017-01-09 Richard Biener <rguenther@suse.de>
41448 PR tree-optimization/78997
41449 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41450 name condition properly.
41452 2017-01-09 Richard Biener <rguenther@suse.de>
41455 * dwarf2out.c (is_cxx): New overload with context.
41456 (is_naming_typedef_decl): Use it.
41458 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41460 * invoke.texi (Option Summary): Correct spacing in option lists
41461 and add line breaks to fix over-long lines.
41463 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41465 PR middle-end/17660
41467 * extend.texi (Common Variable Attributes): Add xref to GCC
41468 Internals manual to explain mode attribute keywords.
41470 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41473 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41474 and Preprocessor Options.
41475 (Options for Linking): Document -pthread here....
41476 (RS/6000 and PowerPC Options): ...not here.
41477 (Solaris 2 Options): ...or here.
41478 * doc/cppopts.texi: Document -pthread.
41480 2017-01-08 Martin Sebor <msebor@redhat.com>
41482 PR middle-end/77708
41483 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41484 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41485 New member functions.
41486 (format_directive): Used them.
41488 (pass_sprintf_length::handle_gimple_call): Same.
41489 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41490 to avoid truncation for any argument.
41491 (extract_affine_mul): Same.
41492 * tree.c (get_file_function_name): Same.
41494 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41496 PR middle-end/77484
41497 * predict.def (PRED_INDIR_CALL): Set to 86.
41499 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41501 PR preprocessor/54124
41502 * doc/cppopts.texi: Reformat -d subtable to list the full name
41503 of the options. Add cross-reference to the docs for the general
41504 compiler -d options.
41505 * doc/invoke.texi (Developer Options): Add cross-reference to the
41506 preprocessor-specific -d option documentation.
41508 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41510 PR preprocessor/13498
41511 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41512 redudant material, and reflect new command-line options.
41513 (System Headers): Likewise.
41515 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41517 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41518 -isystem, and -idirafter. Copy-edit.
41519 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41520 default for -ftrack-macro-expansion. Delete obsolete and
41521 badly-formatted implementation details about -fdebug-cpp output.
41522 * doc/cppwarnopts.texi: Copy-edit.
41524 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41527 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41528 that the transition from a max line width >= 1<<10 to narrower
41529 lines works correctly.
41531 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41533 * doc/options.texi (PerFunction): New.
41534 * opt-functions.awk (switch_flags): Map both Optimization and
41535 PerFunction to CL_OPTIMIZATION.
41536 * opth-gen.awk: Test for PerFunction flag along with
41538 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41539 it only when the latter is present. Skip those that don't in
41540 the hash function generator.
41541 * common.opt (fvar-tracking): Mark as PerFunction instead of
41543 (fvar-tracking-assignments): Likewise.
41544 (fvar-tracking-assignments-toggle): Likewise.
41545 (fvar-tracking-uninit): Likewise.
41547 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41549 PR translation/79018
41550 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41553 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41556 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41557 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41558 TARGET_LONG_JUMP_TABLE_OFFSETS.
41559 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41560 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41561 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41562 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41563 * config/m68k/m68k.md (tablejump expander): Likewise.
41564 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41565 TARGET_LONG_JUMP_TABLE_OFFSETS.
41566 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41567 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41569 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41570 David Holsgrove <david.holsgrove@xilinx.com>
41572 * common/config/microblaze/microblaze-common.c
41573 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41574 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41576 * config/microblaze/microblaze.c (microblaze_must_save_register)
41577 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41579 (microblaze_eh_return): New function.
41580 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41581 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41582 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41583 * config/microblaze/microblaze.md (eh_return): New pattern.
41585 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41587 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41588 GCC_DIAGNOSTIC_STRINGIFY): Define.
41590 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41592 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41594 * config/arm/arm.md (<mcrr>): New.
41596 * config/arm/arm.c (arm_arch5te): New.
41597 (arm_option_override): Set arm_arch5te.
41598 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41600 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41601 (arm_mcrr_qualifiers): ... this. New.
41602 (MRRC_QUALIFIERS): Define to...
41603 (arm_mrrc_qualifiers): ... this. New.
41604 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41606 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41607 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41608 (MRRCI, mrrc, MRRC): New.
41609 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41610 VUNSPEC_MRRC2): New.
41612 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41614 * config/arm/arm.md (<mcr>): New.
41616 * config/arm/arm.c (arm_coproc_builtin_available): Add
41617 support for mcr, mrc, mcr2 and mrc2.
41618 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41619 (arm_mcr_qualifiers): ... this. New.
41620 (MRC_QUALIFIERS): Define to ...
41621 (arm_mrc_qualifiers): ... this. New.
41622 (MCR_QUALIFIERS): Define to ...
41623 (arm_mcr_qualifiers): ... this. New.
41624 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41626 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41627 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41628 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41629 VUNSPEC_MRC2): New.
41631 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41633 * config/arm/arm.md (*ldc): New.
41637 * config/arm/arm.c (arm_coproc_builtin_available): Add
41638 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41639 (arm_coproc_ldc_stc_legitimate_address): New.
41640 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41641 'qualifier_const_pointer'.
41642 (LDC_QUALIFIERS): Define to...
41643 (arm_ldc_qualifiers): ... this. New.
41644 (STC_QUALIFIERS): Define to...
41645 (arm_stc_qualifiers): ... this. New.
41646 * config/arm/arm-protos.h
41647 (arm_coproc_ldc_stc_legitimate_address): New.
41648 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41649 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41650 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41651 stc2, stcl, stc2l): New.
41652 * config/arm/constraints.md (Uz): New.
41653 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41654 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41655 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41656 VUNSPEC_STC2L): New.
41658 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41660 * config/arm/arm.md (<cdp>): New.
41661 * config/arm/arm.c (neon_const_bounds): Rename this ...
41662 (arm_const_bounds): ... this.
41663 (arm_coproc_builtin_available): New.
41664 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41665 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41666 (CDP_QUALIFIERS): Define to...
41667 (arm_cdp_qualifiers): ... this. New.
41669 (arm_expand_builtin_args): Add case for 6 arguments.
41670 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41671 (arm_const_bounds): ... this.
41672 (arm_coproc_builtin_available): New.
41673 * config/arm/arm_acle.h (__arm_cdp): New.
41675 * config/arm/arm_acle_builtins.def (cdp): New.
41677 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41678 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41679 'arm_const_bounds'.
41680 * config/arm/types.md (coproc): New.
41681 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41682 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41683 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41684 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41686 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41688 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41689 (UBINOP_QUALIFIERS): New.
41691 (acle_builtin_data): New. Change comment.
41692 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41693 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41694 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41695 arm_acle_builtins.def.
41696 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41697 (arm_init_acle_builtins): New.
41698 (CRC32_BUILTIN): Remove.
41699 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41700 crc32cb, crc32ch and crc32cw.
41701 (arm_init_crc32_builtins): Remove.
41702 (arm_init_builtins): Use arm_init_acle_builtins rather
41703 than arm_init_crc32_builtins.
41704 (arm_expand_acle_builtin): New.
41705 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41706 * config/arm/arm_acle_builtins.def: New.
41708 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41710 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41711 (arm_builtin_datum): ... this.
41712 (arm_init_neon_builtin): Rename to ...
41713 (arm_init_builtin): ... this. Add a new parameters PREFIX
41714 and USE_SIG_IN_NAME.
41715 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41716 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41717 'arm_builtin_datum'.
41718 (arm_init_vfp_builtins): Likewise.
41719 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41720 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41721 (arm_expand_neon_args): Rename to ...
41722 (arm_expand_builtin_args): ... this. Rename builtin_arg
41723 enum values and differentiate between ARG_BUILTIN_MEMORY
41724 and ARG_BUILTIN_NEON_MEMORY.
41725 (arm_expand_neon_builtin_1): Rename to ...
41726 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41727 values, arm_expand_builtin_args and add bool parameter NEON.
41728 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41729 (arm_expand_vfp_builtin): Likewise.
41730 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41732 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41734 PR middle-end/77484
41735 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41736 * predict.c (tree_estimate_probability_bb): Reverse direction of
41737 polymorphic call predictor.
41739 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41741 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41742 (determine_pass_name_match): ...this new function and...
41743 (should_skip_pass_p): ...this new function.
41745 2017-01-06 Nathan Sidwell <nathan@acm.org>
41747 * ipa-visibility.c (function_and_variable_visibility): Reformat
41748 comments and long lines. Remove extrneous if.
41749 * symtab.c (symtab_node::make_decl_local): Fix code format.
41750 (symtab_node::set_section_for_node): Fix comment typo.
41752 2017-01-06 Martin Liska <mliska@suse.cz>
41755 * lra-constraints.c: Rename invariant to lra_invariant.
41756 * predict.c (set_even_probabilities): Initialize e to NULL.
41758 2017-01-05 Martin Sebor <msebor@redhat.com>
41760 PR tree-optimization/78910
41761 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41762 (format_integer): Correct off-by-one error in the handling
41763 of precision with negative numbers in signed conversions..
41765 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41767 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41769 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41771 PR tree-optimization/71016
41772 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41773 factor_out_conditional_conversion. Formatting fix.
41774 (factor_out_conditional_conversion): Add cond_stmt argument.
41775 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41776 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41779 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41781 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41782 read-rtl-function.o, and selftest-rtl.o.
41783 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41784 (selftest::aarch64_test_loading_full_dump): New function.
41785 (selftest::aarch64_run_selftests): New function.
41786 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41787 selftest::aarch64_run_selftests.
41788 * config/i386/i386.c
41789 (selftest::ix86_test_loading_dump_fragment_1): New function.
41790 (selftest::ix86_test_loading_call_insn): New function.
41791 (selftest::ix86_test_loading_full_dump): New function.
41792 (selftest::ix86_test_loading_unspec): New function.
41793 (selftest::ix86_run_selftests): Call the new functions.
41794 * emit-rtl.c (maybe_set_max_label_num): New function.
41795 * emit-rtl.h (maybe_set_max_label_num): New decl.
41796 * function.c (instantiate_decls): Guard call to
41797 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41798 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
41800 * gensupport.c (gen_reader::gen_reader): Pass "false"
41801 for new "compact" param of rtx_reader.
41802 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
41803 rather than an empty string for NULL strings.
41804 * read-md.c: Potentially include config.h rather than bconfig.h.
41805 Wrap include of errors.h with #ifdef GENERATOR_FILE.
41806 (have_error): New global, copied from errors.c.
41807 (md_reader::read_name): Rename to...
41808 (md_reader::read_name_1): ...this, adding "out_loc" param,
41809 and converting "missing name or number" to returning false, rather
41811 (md_reader::read_name): Reimplement in terms of read_name_1.
41812 (md_reader::read_name_or_nil): New function.
41813 (md_reader::read_string): Handle "(nil)" by returning NULL.
41814 (md_reader::md_reader): Add new param "compact".
41815 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
41816 (md_reader::read_file): New method.
41817 * read-md.h (md_reader::md_reader): Add new param "compact".
41818 (md_reader::read_file): New method.
41819 (md_reader::is_compact): New accessor.
41820 (md_reader::read_name): Convert return type from void to file_location.
41821 (md_reader::read_name_or_nil): New decl.
41822 (md_reader::read_name_1): New decl.
41823 (md_reader::m_compact): New field.
41824 (noop_reader::noop_reader): Pass "false" for new "compact" param
41826 (rtx_reader::rtx_reader): Add new "compact" param.
41827 (rtx_reader::read_rtx_operand): Make virtual and convert return
41828 type from void to rtx.
41829 (rtx_reader::read_until): New decl.
41830 (rtx_reader::handle_any_trailing_information): New virtual function.
41831 (rtx_reader::postprocess): New virtual function.
41832 (rtx_reader::finalize_string): New virtual function.
41833 (rtx_reader::m_in_call_function_usage): New field.
41834 (rtx_reader::m_reuse_rtx_by_id): New field.
41835 * read-rtl-function.c: New file.
41836 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41837 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41838 (selftest::verify_three_block_rtl_cfg): New decl.
41839 * read-rtl-function.h: New file.
41840 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41841 For host, include function.h, memmodel.h, and emit-rtl.h.
41842 (one_time_initialization): New function.
41843 (struct compact_insn_name): New struct.
41844 (compact_insn_names): New array.
41845 (find_code): Handle insn codes in compact dumps.
41846 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41847 (bind_subst_iter_and_attr): Likewise.
41848 (add_condition_to_string): Likewise.
41849 (add_condition_to_rtx): Likewise.
41850 (apply_attribute_uses): Likewise.
41851 (add_current_iterators): Likewise.
41852 (apply_iterators): Likewise.
41853 (initialize_iterators): Guard usage of apply_subst_iterator with
41854 #ifdef GENERATOR_FILE.
41855 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
41856 (md_reader::read_mapping): Likewise.
41857 (add_define_attr_for_define_subst): Likewise.
41858 (add_define_subst_attr): Likewise.
41859 (read_subst_mapping): Likewise.
41860 (check_code_iterator): Likewise.
41861 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41863 (one_time_initialization): New function.
41864 (rtx_reader::read_until): New method.
41865 (read_flags): New function.
41866 (parse_reg_note_name): New function.
41867 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
41868 Handle reuse_rtx ids.
41869 Wrap iterator lookup within #ifdef GENERATOR_FILE.
41870 Add parsing support for RTL dumps, mirroring the special-cases in
41871 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
41872 values, and calling handle_any_trailing_information.
41873 (rtx_reader::read_rtx_operand): Convert return type from void
41874 to rtx, returning return_rtx. Handle case 'e'. Call
41875 finalize_string on XSTR and XTMPL fields.
41876 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
41877 "(nil)" values were omitted. Call the postprocess vfunc on the
41879 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
41880 class ctor. Initialize m_in_call_function_usage. Call
41881 one_time_initialization.
41882 * rtl-tests.c (selftest::test_uncond_jump): Call
41883 set_new_first_and_last_insn.
41884 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41885 * selftest-rtl.c: New file.
41886 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41887 (selftest::get_insn_by_uid): New decl.
41888 * selftest-run-tests.c (selftest::run_tests): Call
41889 read_rtl_function_c_tests.
41890 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41891 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41894 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41896 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41897 operands in a special way. Assert that pos+len <= mode precision.
41899 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41901 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41902 3 argument Alias with unlimited for the negative form.
41903 (fno-vect-cost-model): Removed.
41905 2017-01-05 Martin Liska <mliska@suse.cz>
41907 * hsa-gen.c (gen_hsa_divmod): New function.
41908 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41910 2017-01-05 Martin Liska <mliska@suse.cz>
41913 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41916 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41918 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41919 small constant length operands.
41921 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41923 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41924 between loop iterations.
41926 2017-01-05 Martin Liska <mliska@suse.cz>
41929 * gimplify.c (gimplify_decl_expr): Compare to
41930 asan_poisoned_variables instread of checking flags.
41931 (gimplify_target_expr): Likewise.
41932 (gimplify_expr): Likewise.
41933 (gimplify_function_tree): Conditionally initialize
41934 asan_poisoned_variables.
41936 2017-01-04 Jeff Law <law@redhat.com>
41938 PR tree-optimizatin/78812
41939 * rtl.h (contains_mem_rtx_p): Prototype.
41940 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41941 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41942 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41943 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41944 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41946 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41948 * input.c (assert_char_at_range): Default-initialize actual_range.
41950 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41952 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41953 to match the caller.
41955 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41957 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41958 insns after final jump in test to emit dummy move.
41960 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41962 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41963 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41965 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41967 * multiple_target.c (create_dispatcher_calls): Init e_next.
41968 * tree-ssa-loop-split.c (split_loop): Init border.
41969 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41972 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41977 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41978 (altivec_register_operand): Do not return true if the operand
41979 contains a SUBREG mixing SImode and SFmode.
41980 (vsx_register_operand): Likewise.
41981 (vsx_reg_sfsubreg_ok): New predicate.
41982 (vfloat_operand): Do not return true if the operand contains a
41983 SUBREG mixing SImode and SFmode.
41984 (vint_operand): Likewise.
41985 (vlogical_operand): Likewise.
41986 (gpc_reg_operand): Likewise.
41987 (int_reg_operand): Likewise.
41988 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41989 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41990 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41992 (rs6000_emit_move_si_sf_subreg): New helper function.
41993 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41994 fixup SUBREGs involving SImode and SFmode.
41995 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41996 numbers for the new peephole2 optimization.
41997 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41998 the GLIBC math library that do AND/IOR/XOR operations on single
41999 precision floating point.
42000 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
42001 target macros to say whether we need to avoid SUBREGs mixing
42003 (TARGET_ALLOW_SF_SUBREG): Likewise.
42004 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
42005 (UNSPEC_SI_FROM_SF): Likewise.
42006 (iorxor): Change spacing.
42007 (and_ior_xor): New iterator for AND, IOR, and XOR.
42008 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42009 (movdi_from_sf_zero_ext): Likewise.
42010 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42011 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42012 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42013 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42014 (fms<mode>4): Likewise.
42015 (fnma<mode>4): Likewise.
42016 (fnms<mode>4): Likewise.
42017 (nfma<mode>4): Likewise.
42018 (nfms<mode>4): Likewise.
42020 2017-01-04 Marek Polacek <polacek@redhat.com>
42023 * doc/invoke.texi: Document -Wpointer-compare.
42025 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42027 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42030 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42031 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42034 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42037 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42038 documentation of the powerpc_popcntb_ok attribute.
42039 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42040 code to issue warning messages if a requested CPU configuration is
42041 not supported by the binary (assembler and loader) toolchain.
42042 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42043 made to define a built-in function that has been disabled.
42044 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42045 made to define a built-in function that has been disabled.
42046 (altivec_init_builtins): Add comment explaining why definition
42047 of the DST built-in functions is not preceded by an assertion
42048 check. Add assertions to prevent ICE if attempts are made to
42049 define an altivec predicate or an abs* built-in function that has
42051 (htm_init_builtins): Add comment explaining why definition of the
42052 htm built-in functions is not preceded by an assertion check.
42054 2017-01-04 Jeff Law <law@redhat.com>
42056 PR tree-optimizatin/67955
42057 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42058 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42059 the points-to solution does not include pt_null. Use DECL_PT_UID
42062 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42064 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42065 Use gen_int_mode instead of gen_lopwart for const_int operands.
42067 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42069 PR tree-optimization/71563
42070 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42071 out of range value - has low bits known to be zero.
42073 2017-01-04 Alan Modra <amodra@gmail.com>
42075 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42076 * configure: Regenerate.
42077 * config.in: Regenerate.
42079 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42082 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42083 a substring of the message, but strcmp with the whole message. Ifdef
42084 ENABLE_NLS, translate the message first using dgettext.
42086 2017-01-03 Jeff Law <law@redhat.com>
42088 PR tree-optimizatin/78856
42089 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42090 (mark_threaded_blocks): Remove code to truncate thread paths that
42091 cross multiple loop headers. Instead invalidate the cached loop
42092 iteration information and handle case of a thread path walking
42093 into an irreducible region.
42095 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42098 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42099 assertions. Add support for doing the signbit if the IEEE 128-bit
42100 floating point value is in a GPR.
42101 * config/rs6000/rs6000.md (Fsignbit): Delete.
42102 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42103 Update the length attribute if the value is in a GPR.
42104 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42105 the sign or zero extension instruction, since the value is always 0/1.
42106 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42109 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42110 extracting SImode to a GPR register so that we can generate a
42111 store, limit the vector to be in a traditional Altivec register
42112 for the vextuwrx instruction.
42114 2017-01-03 Ian Lance Taylor <iant@google.com>
42116 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42118 2017-01-03 Martin Sebor <msebor@redhat.com>
42120 PR tree-optimization/78696
42121 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42122 precision. Use MPFR for %f for greater fidelity. Correct handling
42124 (pass_sprintf_length::compute_format_length): Set width and precision
42125 specified by asrerisk to void_node for vararg functions.
42126 (try_substitute_return_value): Adjust dump output.
42128 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42130 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42132 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42134 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42135 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42136 -mlra/-mno-lra was passed to the compiler.
42138 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42140 PR rtl-optimization/65618
42141 * emit-rtl.c (try_split): Move initialization of "before" and
42142 "after" to just before the call to emit_insn_after_setloc.
42144 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42146 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42148 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42150 * dwarf2out.c (gen_enumeration_type_die): When
42151 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42153 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42155 PR tree-optimization/78965
42156 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42157 Change first argument from const call_info & to call_info &. For %n
42158 set info.nowrite to false.
42160 PR middle-end/78901
42161 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42162 possibly throwing calls.
42164 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42165 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42166 and fns handling, rather than in a separate case SSA_NAME.
42168 2017-01-02 Jeff Law <law@redhat.com>
42170 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42171 fixes for first_period and second_period variables.
42173 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42176 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42177 (*insvqi_1): New insn pattern.
42178 (*insvqi_1_mem_rex64): Ditto.
42179 (*insvqi_2): Ditto.
42180 (*insvqi_3): Rename from *insvqi.
42182 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42184 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42186 * doc/cfg.texi (Edges): Remove reference to Java.
42187 (Maintaining the CFG): Ditto.
42189 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42191 PR middle-end/77674
42192 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42193 transparent aliases.
42195 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42197 PR middle-end/77484
42198 * predict.def (PRED_CALL): Update hitrate.
42199 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42200 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42201 into direct/indirect/polymorphic variants.
42203 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42205 Update copyright years.
42207 * gcc.c (process_command): Update copyright notice dates.
42208 * gcov-dump.c (print_version): Ditto.
42209 * gcov.c (print_version): Ditto.
42210 * gcov-tool.c (print_version): Ditto.
42211 * gengtype.c (create_file): Ditto.
42212 * doc/cpp.texi: Bump @copying's copyright year.
42213 * doc/cppinternals.texi: Ditto.
42214 * doc/gcc.texi: Ditto.
42215 * doc/gccint.texi: Ditto.
42216 * doc/gcov.texi: Ditto.
42217 * doc/install.texi: Ditto.
42218 * doc/invoke.texi: Ditto.
42220 Copyright (C) 2017 Free Software Foundation, Inc.
42222 Copying and distribution of this file, with or without modification,
42223 are permitted in any medium without royalty provided the copyright
42224 notice and this notice are preserved.