1 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
3 * expr.c (fixup_args_size_notes): Check that any existing
4 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
5 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
6 (emit_single_push_insn): ...here.
8 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
10 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
11 (const_vector_encoded_nelts): New function.
12 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
13 (const_vector_int_elt, const_vector_elt): Declare.
14 * emit-rtl.c (const_vector_int_elt_1): New function.
15 (const_vector_elt): Likewise.
16 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
19 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
21 * expr.c: Include rtx-vector-builder.h.
22 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
23 directly on the tree encoding.
24 (const_vector_from_tree): Likewise.
25 * optabs.c: Include rtx-vector-builder.h.
26 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
27 sequence of "u" values.
28 * vec-perm-indices.c: Include rtx-vector-builder.h.
29 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
30 directly on the vec_perm_indices encoding.
32 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34 * doc/rtl.texi (const_vector): Describe new encoding scheme.
35 * Makefile.in (OBJS): Add rtx-vector-builder.o.
36 * rtx-vector-builder.h: New file.
37 * rtx-vector-builder.c: Likewise.
38 * rtl.h (rtx_def::u2): Add a const_vector field.
39 (CONST_VECTOR_NPATTERNS): New macro.
40 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
41 (CONST_VECTOR_DUPLICATE_P): Likewise.
42 (CONST_VECTOR_STEPPED_P): Likewise.
43 (CONST_VECTOR_ENCODED_ELT): Likewise.
44 (const_vec_duplicate_p): Check for a duplicated vector encoding.
45 (unwrap_const_vec_duplicate): Likewise.
46 (const_vec_series_p): Check for a non-duplicated vector encoding.
47 Say that the function only returns true for integer vectors.
48 * emit-rtl.c: Include rtx-vector-builder.h.
49 (gen_const_vec_duplicate_1): Delete.
50 (gen_const_vector): Call gen_const_vec_duplicate instead of
51 gen_const_vec_duplicate_1.
52 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
53 (gen_const_vec_duplicate): Use rtx_vector_builder.
54 (gen_const_vec_series): Likewise.
55 (gen_rtx_CONST_VECTOR): Likewise.
56 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
57 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
58 Build a new vector rather than modifying a CONST_VECTOR in-place.
59 (handle_special_swappables): Update call accordingly.
60 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
61 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
62 Build a new vector rather than modifying a CONST_VECTOR in-place.
63 (handle_special_swappables): Update call accordingly.
65 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
67 * simplify-rtx.c (simplify_const_binary_operation): Use
68 CONST_VECTOR_ELT instead of XVECEXP.
70 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
72 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
73 the selector elements to be different from the data elements
74 if the selector is a VECTOR_CST.
75 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
76 ssizetype for the selector.
78 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
80 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
81 before testing each element individually.
82 * tree-vect-generic.c (lower_vec_perm): Likewise.
84 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
86 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
87 * selftest-run-tests.c (selftest::run_tests): Call it.
88 * vector-builder.h (vector_builder::operator ==): New function.
89 (vector_builder::operator !=): Likewise.
90 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
91 (vec_perm_indices::all_from_input_p): New function.
92 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
93 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
94 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
95 instead of reading the VECTOR_CST directly. Detect whether both
96 vector inputs are the same before constructing the vec_perm_indices,
97 and update the number of inputs argument accordingly. Use the
98 utility functions added above. Only construct sel2 if we need to.
100 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
102 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
103 the broadcast of the low byte.
104 (expand_mult_highpart): Use an explicit encoding for the permutes.
105 * optabs-query.c (can_mult_highpart_p): Likewise.
106 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
107 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
108 (vectorizable_bswap): Likewise.
109 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
110 explicit encoding for the power-of-2 permutes.
111 (vect_permute_store_chain): Likewise.
112 (vect_grouped_load_supported): Likewise.
113 (vect_permute_load_chain): Likewise.
115 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
117 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
118 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
119 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
120 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
121 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
122 (vect_gen_perm_mask_any): Likewise.
124 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
126 * int-vector-builder.h: New file.
127 * vec-perm-indices.h: Include int-vector-builder.h.
128 (vec_perm_indices): Redefine as an int_vector_builder.
129 (auto_vec_perm_indices): Delete.
130 (vec_perm_builder): Redefine as a stand-alone class.
131 (vec_perm_indices::vec_perm_indices): New function.
132 (vec_perm_indices::clamp): Likewise.
133 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
134 (vec_perm_indices::new_vector): New function.
135 (vec_perm_indices::new_expanded_vector): Update for new
136 vec_perm_indices class.
137 (vec_perm_indices::rotate_inputs): New function.
138 (vec_perm_indices::all_in_range_p): Operate directly on the
139 encoded form, without computing elided elements.
140 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
141 encoding. Update for new vec_perm_indices class.
142 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
143 the given vec_perm_builder.
144 (expand_vec_perm_var): Update vec_perm_builder constructor.
145 (expand_mult_highpart): Use vec_perm_builder instead of
146 auto_vec_perm_indices.
147 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
148 vec_perm_indices instead of auto_vec_perm_indices. Use a single
149 or double series encoding as appropriate.
150 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
151 vec_perm_indices instead of auto_vec_perm_indices.
152 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
153 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
154 (vect_permute_store_chain): Likewise.
155 (vect_grouped_load_supported): Likewise.
156 (vect_permute_load_chain): Likewise.
157 (vect_shift_permute_load_chain): Likewise.
158 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
159 (vect_transform_slp_perm_load): Likewise.
160 (vect_schedule_slp_instance): Likewise.
161 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
162 (vectorizable_mask_load_store): Likewise.
163 (vectorizable_bswap): Likewise.
164 (vectorizable_store): Likewise.
165 (vectorizable_load): Likewise.
166 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
167 vec_perm_indices instead of auto_vec_perm_indices. Use
168 tree_to_vec_perm_builder to read the vector from a tree.
169 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
170 vec_perm_builder instead of a vec_perm_indices.
171 (have_whole_vector_shift): Use vec_perm_builder and
172 vec_perm_indices instead of auto_vec_perm_indices. Leave the
173 truncation to calc_vec_perm_mask_for_shift.
174 (vect_create_epilog_for_reduction): Likewise.
175 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
176 from auto_vec_perm_indices to vec_perm_indices.
177 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
178 instead of changing individual elements.
179 (aarch64_vectorize_vec_perm_const): Use new_vector to install
180 the vector in d.perm.
181 * config/arm/arm.c (expand_vec_perm_d::perm): Change
182 from auto_vec_perm_indices to vec_perm_indices.
183 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
184 instead of changing individual elements.
185 (arm_vectorize_vec_perm_const): Use new_vector to install
186 the vector in d.perm.
187 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
188 Update vec_perm_builder constructor.
189 (rs6000_expand_interleave): Likewise.
190 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
191 (rs6000_expand_interleave): Likewise.
193 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
195 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
196 to qimode could truncate the indices.
197 * optabs.c (expand_vec_perm_var): Likewise.
199 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
201 * Makefile.in (OBJS): Add vec-perm-indices.o.
202 * vec-perm-indices.h: New file.
203 * vec-perm-indices.c: Likewise.
204 * target.h (vec_perm_indices): Replace with a forward class
206 (auto_vec_perm_indices): Move to vec-perm-indices.h.
207 * optabs.h: Include vec-perm-indices.h.
208 (expand_vec_perm): Delete.
209 (selector_fits_mode_p, expand_vec_perm_var): Declare.
210 (expand_vec_perm_const): Declare.
211 * target.def (vec_perm_const_ok): Replace with...
212 (vec_perm_const): ...this new hook.
213 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
214 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
215 * doc/tm.texi: Regenerate.
216 * optabs.def (vec_perm_const): Delete.
217 * doc/md.texi (vec_perm_const): Likewise.
218 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
219 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
220 expand_vec_perm for constant permutation vectors. Assert that
221 the mode of variable permutation vectors is the integer equivalent
222 of the mode that is being permuted.
223 * optabs-query.h (selector_fits_mode_p): Declare.
224 * optabs-query.c: Include vec-perm-indices.h.
225 (selector_fits_mode_p): New function.
226 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
227 is defined, instead of checking whether the vec_perm_const_optab
228 exists. Use targetm.vectorize.vec_perm_const instead of
229 targetm.vectorize.vec_perm_const_ok. Check whether the indices
230 fit in the vector mode before using a variable permute.
231 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
232 vec_perm_indices instead of an rtx.
233 (expand_vec_perm): Replace with...
234 (expand_vec_perm_const): ...this new function. Take the selector
235 as a vec_perm_indices rather than an rtx. Also take the mode of
236 the selector. Update call to shift_amt_for_vec_perm_mask.
237 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
238 Use vec_perm_indices::new_expanded_vector to expand the original
239 selector into bytes. Check whether the indices fit in the vector
240 mode before using a variable permute.
241 (expand_vec_perm_var): Make global.
242 (expand_mult_highpart): Use expand_vec_perm_const.
243 * fold-const.c: Includes vec-perm-indices.h.
244 * tree-ssa-forwprop.c: Likewise.
245 * tree-vect-data-refs.c: Likewise.
246 * tree-vect-generic.c: Likewise.
247 * tree-vect-loop.c: Likewise.
248 * tree-vect-slp.c: Likewise.
249 * tree-vect-stmts.c: Likewise.
250 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
252 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
253 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
254 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
255 (aarch64_vectorize_vec_perm_const): ...this new function.
256 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
257 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
258 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
259 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
260 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
261 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
262 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
264 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
265 check for NEON modes.
266 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
267 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
268 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
269 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
271 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
272 the old VEC_PERM_CONST conditions.
273 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
274 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
275 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
276 (ia64_vectorize_vec_perm_const_ok): Merge into...
277 (ia64_vectorize_vec_perm_const): ...this new function.
278 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
279 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
280 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
281 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
282 * config/mips/mips.c (mips_expand_vec_perm_const)
283 (mips_vectorize_vec_perm_const_ok): Merge into...
284 (mips_vectorize_vec_perm_const): ...this new function.
285 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
286 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
287 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
288 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
289 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
290 (rs6000_expand_vec_perm_const): Delete.
291 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
293 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
294 (altivec_expand_vec_perm_const_le): Take each operand individually.
295 Operate on constant selectors rather than rtxes.
296 (altivec_expand_vec_perm_const): Likewise. Update call to
297 altivec_expand_vec_perm_const_le.
298 (rs6000_expand_vec_perm_const): Delete.
299 (rs6000_vectorize_vec_perm_const_ok): Delete.
300 (rs6000_vectorize_vec_perm_const): New function.
301 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
302 an element count and rtx array.
303 (rs6000_expand_extract_even): Update call accordingly.
304 (rs6000_expand_interleave): Likewise.
305 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
306 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
307 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
308 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
309 (rs6000_expand_vec_perm_const): Delete.
310 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
311 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
312 (altivec_expand_vec_perm_const_le): Take each operand individually.
313 Operate on constant selectors rather than rtxes.
314 (altivec_expand_vec_perm_const): Likewise. Update call to
315 altivec_expand_vec_perm_const_le.
316 (rs6000_expand_vec_perm_const): Delete.
317 (rs6000_vectorize_vec_perm_const_ok): Delete.
318 (rs6000_vectorize_vec_perm_const): New function. Remove stray
319 reference to the SPE evmerge intructions.
320 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
321 an element count and rtx array.
322 (rs6000_expand_extract_even): Update call accordingly.
323 (rs6000_expand_interleave): Likewise.
324 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
325 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
327 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
329 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
331 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
332 vector mode and that that mode matches the mode of the data
334 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
335 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
336 directly using expand_vec_perm_1 when forcing selectors into
338 (expand_vec_perm_var): New function, split out from expand_vec_perm.
340 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
342 * optabs-query.h (can_vec_perm_p): Delete.
343 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
344 * optabs-query.c (can_vec_perm_p): Split into...
345 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
346 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
347 particular selector is valid.
348 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
349 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
350 (vect_grouped_load_supported): Likewise.
351 (vect_shift_permute_load_chain): Likewise.
352 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
353 (vect_transform_slp_perm_load): Likewise.
354 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
355 (vectorizable_bswap): Likewise.
356 (vect_gen_perm_mask_checked): Likewise.
357 * fold-const.c (fold_ternary_loc): Likewise. Don't take
358 implementations of variable permutation vectors into account
359 when deciding which selector to use.
360 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
361 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
362 with a false third argument.
363 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
364 to test whether the constant selector is valid and can_vec_perm_var_p
365 to test whether a variable selector is valid.
367 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
369 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
370 * optabs-query.c (can_vec_perm_p): Likewise.
371 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
372 instead of vec_perm_indices.
373 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
374 (vect_gen_perm_mask_checked): Likewise,
375 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
376 (vect_gen_perm_mask_checked): Likewise,
378 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
380 * optabs-query.h (qimode_for_vec_perm): Declare.
381 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
382 (qimode_for_vec_perm): ...this new function.
383 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
385 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
387 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
388 does not have a conditional at the top.
390 2018-01-02 Richard Biener <rguenther@suse.de>
392 * ipa-inline.c (big_speedup_p): Fix expression.
394 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
397 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
400 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
404 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
405 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
406 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
407 cond_taken_branch_cost 3->4.
409 2018-01-01 Jakub Jelinek <jakub@redhat.com>
411 PR tree-optimization/83581
412 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
413 TODO_cleanup_cfg if any changes have been made.
416 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
417 convert_modes if target mode has the right side, but different mode
421 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
422 last argument when extracting from CONCAT. If either from_real or
423 from_imag is NULL, use expansion through memory. If result is not
424 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
425 the parts directly to inner mode, if even that fails, use expansion
429 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
430 check for bswap in mode rather than HImode and use that in expand_unop
433 2017-12-31 Jakub Jelinek <jakub@redhat.com>
436 * config/i386/i386.c (ix86_attribute_table): Remove excess
437 initializer for "shared" attribute.
439 2017-12-30 Tom de Vries <tom@codesourcery.com>
442 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
444 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
445 offload_funcs entries that no longer have a corresponding cgraph_node.
446 Mark the remaining ones as DECL_PRESERVE_P.
447 (output_lto): Call prune_offload_funcs.
449 2017-12-30 Jakub Jelinek <jakub@redhat.com>
451 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
452 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
453 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
454 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
455 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
456 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
457 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
458 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
459 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
460 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
461 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
462 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
463 vaesenc_<mode>, vpclmulqdq_<mode>,
464 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
466 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
468 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
469 _Float<N> and _Float<N>X variants for rounding built-in
471 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
472 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
473 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
474 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
475 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
476 * builtins.c (mathfn_built_in_2): Likewise.
477 * internal-fn.def (CEIL): Likewise.
479 (NEARBYINT): Likewise.
483 * convert.c (convert_to_integer_1): Likewise.
484 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
485 (integer_valued_real_call_p): Likewise.
486 * fold-const-call.c (fold_const_call_ss): Likewise.
487 * gencfn-macros.c (print_case_cfn): Change CFN and operator
488 printers to take a const char * suffix instead of a bool.
489 (print_define_operator_list): Likewise.
490 (fltall_suffixes): New list of suffixes, that include the
491 traditional suffixes as well as all of the _Float<N> and
493 (main): For _Float<N> and _Float<N>X functions, emit both
494 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
495 has the _Float<N> and _Float<N>X case names or operators. The
496 <name>_ALL variant has both the traditional and the
497 _Float<N>/_Float<N>X case names or operators.
498 * match.pd (COPYSIGN optimizations): Provide optimizations for
499 _Float<N> and _Float<N>X types where possible.
500 (MIN/MAX optimizations): Likewise.
501 (sqrt optimizations): Likewise.
502 (rounding optimizations): Likewise.
504 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
506 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
507 gen_int_mode rather than GEN_INT.
508 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
510 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
512 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
513 instead of CONSTANT_P.
514 (gen_vec_series): Likewise.
515 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
517 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
519 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
522 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
524 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
525 (aarch64_ld1x2<VDC:mode>): Likewise.
526 (aarch64_simd_ld1<mode>_x2): Likewise.
527 (aarch64_simd_ld1<mode>_x2): Likewise.
528 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
529 (vld1_s8_x2): Likewise.
530 (vld1_u16_x2): Likewise.
531 (vld1_s16_x2): Likewise.
532 (vld1_u32_x2): Likewise.
533 (vld1_s32_x2): Likewise.
534 (vld1_u64_x2): Likewise.
535 (vld1_s64_x2): Likewise.
536 (vld1_f16_x2): Likewise.
537 (vld1_f32_x2): Likewise.
538 (vld1_f64_x2): Likewise.
539 (vld1_p8_x2): Likewise.
540 (vld1_p16_x2): Likewise.
541 (vld1_p64_x2): Likewise.
542 (vld1q_u8_x2): Likewise.
543 (vld1q_s8_x2): Likewise.
544 (vld1q_u16_x2): Likewise.
545 (vld1q_s16_x2): Likewise.
546 (vld1q_u32_x2): Likewise.
547 (vld1q_s32_x2): Likewise.
548 (vld1q_u64_x2): Likewise.
549 (vld1q_s64_x2): Likewise.
550 (vld1q_f16_x2): Likewise.
551 (vld1q_f32_x2): Likewise.
552 (vld1q_f64_x2): Likewise.
553 (vld1q_p8_x2): Likewise.
554 (vld1q_p16_x2): Likewise.
555 (vld1q_p64_x2): Likewise.
557 2017-12-27 Martin Liska <mliska@suse.cz>
559 PR tree-optimization/83552
560 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
561 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
563 2017-12-27 Tom de Vries <tom@codesourcery.com>
565 * config/nvptx/nvptx.c (nvptx_option_override): Disable
566 -gstatement-frontiers.
568 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
570 PR rtl-optimization/83513
571 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
572 before priority comparison.
574 2017-12-25 Jakub Jelinek <jakub@redhat.com>
577 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
578 ix86_isa_flags2 to ix86_isa_flags.
579 * config/i386/i386-c.c (ix86_target_macros_internal): Test
580 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
581 isa_flags rather than isa_flags2.
582 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
583 and -mavx512bitalg from isa2_opts to isa_opts.
584 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
585 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
586 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
588 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
589 OPTION_MASK_ISA_AVX512F ored with another option similarly to
590 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
591 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
592 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
593 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
594 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
595 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
596 builtins from bdesc_args2 to bdesc_args section.
597 (__builtin_ia32_compressstoreuqi512_mask,
598 __builtin_ia32_compressstoreuhi512_mask,
599 __builtin_ia32_compressstoreuqi256_mask,
600 __builtin_ia32_expandloadqi512_mask,
601 __builtin_ia32_expandloadqi512_maskz,
602 __builtin_ia32_expandloadhi512_mask,
603 __builtin_ia32_expandloadhi512_maskz,
604 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
605 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
606 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
607 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
608 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
609 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
610 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
611 __builtin_ia32_vpshldv_v32hi_maskz,
612 __builtin_ia32_vpopcountb_v64qi_mask,
613 __builtin_ia32_vpopcountw_v32hi_mask,
614 __builtin_ia32_vpshufbitqmb512_mask,
615 __builtin_ia32_vpshufbitqmb256_mask): Add
616 " | OPTION_MASK_ISA_AVX512BW".
617 (__builtin_ia32_expandloadqi256_mask,
618 __builtin_ia32_expandloadqi256_maskz,
619 __builtin_ia32_vpopcountb_v32qi_mask): Add
620 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
621 (__builtin_ia32_expandloadhi256_mask,
622 __builtin_ia32_expandloadhi256_maskz,
623 __builtin_ia32_expandloadqi128_mask,
624 __builtin_ia32_expandloadqi128_maskz,
625 __builtin_ia32_expandloadhi128_mask,
626 __builtin_ia32_expandloadhi128_maskz,
627 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
628 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
629 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
630 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
631 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
632 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
633 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
634 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
635 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
636 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
637 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
638 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
639 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
640 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
641 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
642 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
643 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
644 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
645 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
646 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
647 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
648 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
649 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
650 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
651 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
652 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
653 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
654 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
655 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
656 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
657 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
658 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
659 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
660 __builtin_ia32_vpopcountw_v8hi_mask): Add
661 " | OPTION_MASK_ISA_AVX512VL".
662 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
663 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
664 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
665 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
666 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
667 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
668 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
669 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
670 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
671 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
672 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
673 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
674 _mm_mask_bitshuffle_epi64_mask): Likewise.
675 * common/config/i386/i386-common.c
676 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
677 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
678 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
679 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
680 OPTION_MASK_ISA_AVX512BITALG_UNSET.
681 (OPTION_MASK_ISA2_AVX512F_UNSET,
682 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
683 (ix86_handle_option): For -mno-general-regs-only, clear from
684 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
685 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
686 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
687 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
688 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
689 for moving from ix86_isa_flags2 to ix86_isa_flags.
691 2017-12-23 Jakub Jelinek <jakub@redhat.com>
694 * fold-const.c (struct contains_label_data): New type.
695 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
696 inside of a SWITCH_BODY seen during the walk.
697 (contains_label_p): Use walk_tree instead of
698 walk_tree_without_duplicates, prepare data for contains_label_1 and
701 2017-12-22 Jakub Jelinek <jakub@redhat.com>
704 * tree-iterator.c (alloc_stmt_list): Start with cleared
705 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
709 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
710 enable avx512f explicitly in #pragma GCC target.
711 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
712 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
713 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
714 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
715 __builtin_ia32_vpdpbusds_v8si_mask,
716 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
717 __builtin_ia32_vpdpbusds_v4si_mask,
718 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
719 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
720 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
721 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
722 __builtin_ia32_vpdpwssds_v8si_mask,
723 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
724 __builtin_ia32_vpdpwssds_v4si_mask,
725 __builtin_ia32_vpdpwssds_v4si_maskz): Use
726 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
727 just OPTION_MASK_ISA_AVX512VNNI.
729 2017-12-22 Martin Jambor <mjambor@suse.cz>
732 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
735 2017-12-22 Julia Koval <julia.koval@intel.com>
736 Sebastian Peryt <sebastian.peryt@intel.com>
738 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
739 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
740 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
741 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
742 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
743 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
744 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
745 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
746 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
747 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
748 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
749 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
750 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
751 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
752 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
753 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
754 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
755 _mm256_maskz_popcnt_epi64): New intrinsics.
756 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
757 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
758 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
759 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
760 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
761 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
762 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
763 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
764 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
765 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
766 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
767 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
768 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
769 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
770 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
771 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
772 __builtin_ia32_vpshufbitqmb256_mask,
773 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
774 * config/i386/i386-c.c (__AVX512BITALG__): New.
775 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
776 (ix86_valid_target_attribute_inner_p): Ditto.
777 (ix86_expand_args_builtin): Handle new types.
778 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
779 * config/i386/i386.opt: Add -mavx512bitalg.
780 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
781 avx512bitalgintrin.h.
782 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
783 (vpopcount<mode><mask_name>): Add more types.
784 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
785 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
787 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
789 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
790 Or in OPTION_MASK_ISA_AVX512F_SET.
791 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
792 OPTION_MASK_ISA_AVX512VNNI_UNSET.
793 (ix86_handle_option): Adjust for
794 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
795 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
797 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
798 isa_flag instead of isa_flag2.
799 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
800 isa_opts2 to isa_opts.
801 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
804 2017-12-22 Mike Stump <mikestump@comcast.net>
805 Eric Botcazou <ebotcazou@adacore.com>
807 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
809 2017-12-21 Jakub Jelinek <jakub@redhat.com>
812 * config/i386/i386.c (ix86_function_arg_boundary): Return
813 PARM_BOUNDARY for TYPE_EMPTY_P types.
816 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
817 if navail is >= dir.len.
819 2017-12-21 Steve Ellcey <sellcey@cavium.com>
821 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
824 2017-12-21 Jakub Jelinek <jakub@redhat.com>
826 PR rtl-optimization/80747
827 PR rtl-optimization/83512
828 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
829 succ edge from ENTRY, copy partition from e->dest to the newly
831 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
832 ENTRY, use BB_PARTITION of its successor block as current_partition.
833 Don't copy partition when splitting succ edge from ENTRY.
835 PR tree-optimization/83523
836 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
837 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
838 (convert_mult_to_fma): Likewise.
840 PR tree-optimization/83521
841 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
842 gimple_build_assign without code on result of
843 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
846 2017-12-21 Andrew Pinski <apinski@cavium.com>
847 Steve Ellcey <sellcey@cavium.com>
849 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
850 multi-arch for ilp32.
852 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
855 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
856 constraints to operand 2.
857 (*ashl<mode>3_mask_1): Ditto.
858 (*<shift_insn><mode>3_mask): Ditto.
859 (*<shift_insn><mode>3_mask_1): Ditto.
860 (*<rotate_insn><mode>3_mask): Ditto.
861 (*<rotate_insn><mode>3_mask_1): Ditto.
863 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
865 * reorg.c (make_return_insns): Reemit each insn with its own location.
867 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
869 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
870 generation for cases where splatting a value is not useful.
871 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
872 across a vec_duplicate and a paradoxical subreg forming a vector
873 mode to a vec_concat.
875 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
877 * combine.c (simplify_set): Do not transform subregs to zero_extends
878 if the destination is not a scalar int mode.
880 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
883 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
884 shared zero if the input has overflowed.
886 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
888 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
889 support for Cortex-A55 and Cortex-A75.
891 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
893 * common/config/arm/arm-common.c (compare_opt_names): Add function
894 comment. Use strcmp instead of manual loop.
896 2017-12-21 Martin Liska <mliska@suse.cz>
898 PR gcov-profile/83509
899 * gcov-dump.c (dump_gcov_file): Do not read info about
900 support_unexecuted_blocks for gcda files.
902 2017-12-21 Jakub Jelinek <jakub@redhat.com>
904 PR rtl-optimization/82973
905 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
906 (valid_for_const_vector_p): ... this.
907 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
908 (valid_for_const_vector_p): ... this. Adjust function comment.
909 (gen_vec_duplicate): Adjust caller.
910 * optabs.c (expand_vector_broadcast): Likewise.
911 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
912 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
914 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
915 instead of GET_CODE == CONST_FIXED.
916 (simplify_subreg): Use CONST_FIXED_P macro instead of
917 GET_CODE == CONST_FIXED.
920 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
921 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
922 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
923 (ix86_option_override_internal): Adjust for
924 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
925 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
926 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
927 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
928 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
929 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
930 def_builtin2 for CET builtins.
931 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
932 up formatting in IX86_BUILTIN_RDPID code.
933 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
934 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
935 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
936 ix86_isa_flags2 to ix86_isa_flags.
937 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
939 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
940 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
941 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
943 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
944 Or in OPTION_MASK_ISA_AVX512F_SET.
945 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
946 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
947 (ix86_handle_option): Adjust for
948 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
949 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
951 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
952 Alan Hayward <alan.hayward@arm.com>
953 David Sherwood <david.sherwood@arm.com>
955 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
956 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
957 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
958 Track polynomial offsets.
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-vect-data-refs.c (vect_compute_data_ref_alignment):
965 Treat drb->init as a poly_int. Fail if its misalignment wrt
966 vector_alignment isn't known.
968 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
969 Alan Hayward <alan.hayward@arm.com>
970 David Sherwood <david.sherwood@arm.com>
972 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
974 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
975 (vect_loop_versioning): Take the loop versioning threshold as a
977 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
978 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
979 versioning_threshold.
980 (vect_analyze_loop_2): Compute the loop versioning threshold
981 whenever loop versioning is needed, and store it in the new
982 field rather than combining it with the cost model threshold.
983 (vect_transform_loop): Update call to vect_loop_versioning.
984 Try to combine the loop versioning and cost thresholds here.
986 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
987 Alan Hayward <alan.hayward@arm.com>
988 David Sherwood <david.sherwood@arm.com>
990 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
991 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
992 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
993 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
994 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
995 (group_compare_offset): Update accordingly.
996 (split_small_address_groups_p): Likewise.
997 (record_use): Take addr_offset as a poly_uint64 rather than
998 an unsigned HOST_WIDE_INT.
999 (strip_offset): Return the offset as a poly_uint64 rather than
1000 an unsigned HOST_WIDE_INT.
1001 (record_group_use, split_address_groups): Track polynomial offsets.
1002 (add_iv_candidate_for_use): Likewise.
1003 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1004 than a HOST_WIDE_INT.
1005 (strip_offset_1): Return the offset as a poly_int64 rather than
1008 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1009 Alan Hayward <alan.hayward@arm.com>
1010 David Sherwood <david.sherwood@arm.com>
1012 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1013 rather than a HOST_WIDE_INT.
1014 * tree.c (get_binfo_at_offset): Likewise.
1016 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1017 Alan Hayward <alan.hayward@arm.com>
1018 David Sherwood <david.sherwood@arm.com>
1020 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1021 rather than a HOST_WIDE_INT.
1022 * tree-sra.c (build_ref_for_offset): Likewise.
1024 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1025 Alan Hayward <alan.hayward@arm.com>
1026 David Sherwood <david.sherwood@arm.com>
1028 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1030 * tree.c (mem_ref_offset): Likewise.
1031 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1032 * builtins.c (get_object_alignment_2): Likewise.
1033 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1034 * gimple-fold.c (get_base_constructor): Likewise.
1035 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1036 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1038 * ipa-polymorphic-call.c
1039 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1040 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1041 (get_ancestor_addr_info): Likewise.
1042 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1043 * match.pd: Likewise.
1044 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1045 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1046 * tree-eh.c (tree_could_trap_p): Likewise.
1047 * tree-object-size.c (addr_object_size): Likewise.
1048 * tree-ssa-address.c (copy_ref_info): Likewise.
1049 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1050 (indirect_refs_may_alias_p): Likewise.
1051 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1052 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1053 (non_rewritable_mem_ref_base): Likewise.
1054 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1055 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1056 * varasm.c (decode_addr_const): Likewise.
1058 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1059 Alan Hayward <alan.hayward@arm.com>
1060 David Sherwood <david.sherwood@arm.com>
1062 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1063 offsets for MEM_REFs.
1065 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1066 Alan Hayward <alan.hayward@arm.com>
1067 David Sherwood <david.sherwood@arm.com>
1069 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1070 as a poly_uint64 rather than an unsigned int.
1071 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
1073 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1074 Alan Hayward <alan.hayward@arm.com>
1075 David Sherwood <david.sherwood@arm.com>
1077 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1079 (decode_addr_const): Update accordingly.
1081 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1082 Alan Hayward <alan.hayward@arm.com>
1083 David Sherwood <david.sherwood@arm.com>
1085 * tree.h (bit_field_size, bit_field_offset): New functions.
1086 * hsa-gen.c (gen_hsa_addr): Use them.
1087 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1088 (simplify_vector_constructor): Likewise.
1089 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1090 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1091 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1092 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
1095 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1096 Alan Hayward <alan.hayward@arm.com>
1097 David Sherwood <david.sherwood@arm.com>
1099 * expr.h (emit_group_load, emit_group_load_into_temps)
1100 (emit_group_store): Take the size as a poly_int64 rather than an int.
1101 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1102 (emit_group_load_into_temp, emit_group_store): Likewise.
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 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1109 an unsigned int to a poly_uint64.
1110 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1111 sizes as poly_uint64s rather than unsigned ints.
1112 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1115 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1116 Alan Hayward <alan.hayward@arm.com>
1117 David Sherwood <david.sherwood@arm.com>
1119 * lra-constraints.c (emit_inc): Change inc_amount from an int
1122 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1123 Alan Hayward <alan.hayward@arm.com>
1124 David Sherwood <david.sherwood@arm.com>
1126 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1128 (add_stack_var, stack_var_cmp, partition_stack_vars)
1129 (dump_stack_var_partition): Update accordingly.
1130 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1131 than a HOST_WIDE_INT.
1132 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1133 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1134 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1135 rather than a HOST_WIDE_INT.
1137 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1138 Alan Hayward <alan.hayward@arm.com>
1139 David Sherwood <david.sherwood@arm.com>
1141 * target.def (return_pops_args): Treat both the input and output
1142 sizes as poly_int64s rather than HOST_WIDE_INTS.
1143 * targhooks.h (default_return_pops_args): Update accordingly.
1144 * targhooks.c (default_return_pops_args): Likewise.
1145 * doc/tm.texi: Regenerate.
1146 * emit-rtl.h (incoming_args): Change pops_args, size and
1147 outgoing_args_size from int to poly_int64_pod.
1148 * function.h (expr_status): Change x_pending_stack_adjust and
1149 x_stack_pointer_delta from int to poly_int64.
1150 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1151 (ARGS_SIZE_RTX): Update accordingly.
1152 * calls.c (highest_outgoing_arg_in_use): Change from int to
1154 (stack_usage_watermark, stored_args_watermark): New variables.
1155 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1156 (emit_call_1): Change the stack_size and rounded_stack_size
1157 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1159 (save_fixed_argument_area): Check stack_usage_watermark.
1160 (initialize_argument_information): Change old_pending_adj from
1161 a HOST_WIDE_INT * to a poly_int64_pod *.
1162 (compute_argument_block_size): Return the size as a poly_int64
1164 (finalize_must_preallocate): Track polynomial argument sizes.
1165 (compute_argument_addresses): Likewise.
1166 (internal_arg_pointer_based_exp): Track polynomial offsets.
1167 (mem_overlaps_already_clobbered_arg_p): Rename to...
1168 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1169 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1170 Check stored_args_used_watermark.
1171 (load_register_parameters): Update accordingly.
1172 (check_sibcall_argument_overlap_1): Likewise.
1173 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1174 args size as a poly_int64 rather than an int. Return a bool
1175 indicating whether the optimization was possible and return
1176 the new adjustment by reference.
1177 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1178 Update stored_args_watermark.
1179 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1180 (expand_call): Likewise. Maintain stack_usage_watermark and
1181 stored_args_watermark. Update calls to
1182 combine_pending_stack_adjustment_and_call.
1183 (emit_library_call_value_1): Handle polynomial argument sizes.
1184 Call stack_region_maybe_used_p and mark_stack_region_used.
1185 Maintain stack_usage_watermark.
1186 (store_one_arg): Likewise. Update call to
1187 mem_overlaps_already_clobbered_arg_p.
1188 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1190 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1191 * config/microblaze/microblaze.c (microblaze_function_prologue):
1193 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1194 TARGET_RETURN_POPS_ARGS interface.
1195 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1197 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1198 * config/i386/i386.c (ix86_return_pops_args): Update for new
1199 TARGET_RETURN_POPS_ARGS interface.
1200 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1201 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1202 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1203 TARGET_RETURN_POPS_ARGS interface.
1204 * config/vax/vax.c (vax_return_pops_args): Likewise.
1205 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1206 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1207 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1208 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1209 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1210 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1211 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1212 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1213 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1214 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1215 and x_stack_pointer_delta from int to poly_int64.
1216 * dojump.c (do_pending_stack_adjust): Update accordingly.
1217 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1218 stack_pointer_deltas.
1219 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1220 (pad_to_arg_alignment): Track polynomial offsets.
1221 (assign_parm_find_stack_rtl): Likewise.
1222 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1223 * toplev.c (output_stack_usage): Update reference to
1224 current_function_pushed_stack_size.
1226 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1227 Alan Hayward <alan.hayward@arm.com>
1228 David Sherwood <david.sherwood@arm.com>
1230 * function.c (in_arg_offset, var_offset, dynamic_offset)
1231 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1232 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1233 rather than a HOST_WIDE_INT.
1234 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1235 (instantiate_virtual_regs_in_insn): Likewise.
1237 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1238 Alan Hayward <alan.hayward@arm.com>
1239 David Sherwood <david.sherwood@arm.com>
1241 * rtl.h (get_args_size, add_args_size_note): New functions.
1242 (find_args_size_adjust): Return a poly_int64 rather than a
1244 (fixup_args_size_notes): Likewise. Make the same change to the
1245 end_args_size parameter.
1246 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1247 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1248 * calls.c (emit_call_1): Likewise.
1249 * explow.c (adjust_stack_1): Likewise.
1250 * cfgcleanup.c (old_insns_match_p): Update use of
1251 find_args_size_adjust.
1252 * combine.c (distribute_notes): Track polynomial arg sizes.
1253 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1254 end_true_args_size, beg_delay_args_size and end_delay_args_size
1255 from HOST_WIDE_INT to poly_int64.
1256 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1257 than a HOST_WIDE_INT.
1258 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1259 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1260 polynomial arg sizes.
1261 * emit-rtl.c (try_split): Use get_args_size.
1262 * recog.c (peep2_attempt): Likewise.
1263 * reload1.c (reload_as_needed): Likewise.
1264 * expr.c (find_args_size_adjust): Return the adjustment as a
1265 poly_int64 rather than a HOST_WIDE_INT.
1266 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1267 to a poly_int64 and change the return type in the same way.
1268 (emit_single_push_insn): Track polynomial arg sizes.
1270 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1271 Alan Hayward <alan.hayward@arm.com>
1272 David Sherwood <david.sherwood@arm.com>
1274 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1275 from HOST_WIDE_INT to poly_int64.
1276 * expr.c (push_block, emit_push_insn): Likewise.
1278 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1279 Alan Hayward <alan.hayward@arm.com>
1280 David Sherwood <david.sherwood@arm.com>
1282 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1284 (get_frame_size): Return the size as a poly_int64 rather than a
1286 (frame_offset_overflow): Take the offset as a poly_int64 rather
1287 than a HOST_WIDE_INT.
1288 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1289 (assign_stack_temp): Likewise for the size.
1290 * function.c (get_frame_size): Return a poly_int64 rather than
1292 (frame_offset_overflow): Take the offset as a poly_int64 rather
1293 than a HOST_WIDE_INT.
1294 (try_fit_stack_local): Take the start, length and size as poly_int64s
1295 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1296 rather than a HOST_WIDE_INT.
1297 (add_frame_space): Take the start and end as poly_int64s rather than
1299 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1300 (assign_stack_temp): Likewise for the size.
1301 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1303 (find_temp_slot_from_address): Handle polynomial offsets.
1304 (combine_temp_slots): Likewise.
1305 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1307 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1308 poly_int64 rather than a HOST_WIDE_INT.
1309 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1310 than a HOST_WIDE_INT.
1311 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1312 polynomial frame offsets.
1313 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1314 as a poly_int64 rather than an int.
1315 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1316 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1317 * config/v850/v850.c (compute_frame_size): Likewise.
1318 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1319 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1320 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1321 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1322 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1323 rather than a HOST_WIDE_INT.
1324 * explow.c (get_dynamic_stack_base): Likewise.
1325 * final.c (final_start_function): Use the constant lower bound
1326 of the frame size for -Wframe-larger-than.
1327 * ira.c (do_reload): Adjust for new get_frame_size return type.
1328 * lra.c (lra): Likewise.
1329 * reload1.c (reload): Likewise.
1330 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1331 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1332 * rtlanal.c (get_initial_register_offset): Return the offset as
1333 a poly_int64 rather than a HOST_WIDE_INT.
1335 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1336 Alan Hayward <alan.hayward@arm.com>
1337 David Sherwood <david.sherwood@arm.com>
1339 * reload1.c (elim_table): Change initial_offset, offset and
1340 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1341 (offsets_at): Change the target array's element type from
1342 HOST_WIDE_INT to poly_int64_pod.
1343 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1344 (elimination_costs_in_insn, update_eliminable_offsets)
1345 (verify_initial_elim_offsets, set_offsets_for_label)
1346 (init_eliminable_invariants): Update after above changes.
1348 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1349 Alan Hayward <alan.hayward@arm.com>
1350 David Sherwood <david.sherwood@arm.com>
1352 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1353 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1354 (decomposition): Change start and end from HOST_WIDE_INT
1356 (decompose, immune_p): Update accordingly.
1357 (find_inc_amount): Return a poly_int64 rather than an int.
1358 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1361 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1362 Alan Hayward <alan.hayward@arm.com>
1363 David Sherwood <david.sherwood@arm.com>
1365 * tree.h (get_inner_reference): Return the bitsize and bitpos
1366 as poly_int64_pods rather than HOST_WIDE_INT.
1367 * fold-const.h (ptr_difference_const): Return the pointer difference
1368 as a poly_int64_pod rather than a HOST_WIDE_INT.
1369 * expr.c (get_inner_reference): Return the bitsize and bitpos
1370 as poly_int64_pods rather than HOST_WIDE_INT.
1371 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1373 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1374 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1375 (optimize_bit_field_compare): Update call to get_inner_reference.
1376 (decode_field_reference): Likewise.
1377 (fold_unary_loc): Track polynomial offsets and sizes.
1378 (split_address_to_core_and_offset): Return the bitpos as a
1379 poly_int64_pod rather than a HOST_WIDE_INT.
1380 (ptr_difference_const): Likewise for the pointer difference.
1381 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1382 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1383 * dbxout.c (dbxout_expand_expr): Likewise.
1384 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1385 (loc_list_from_tree_1, fortran_common): Likewise.
1386 * gimple-laddress.c (pass_laddress::execute): Likewise.
1387 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1388 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1389 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1390 * tree-affine.c (tree_to_aff_combination): Likewise.
1391 (get_inner_reference_aff): Likewise.
1392 * tree-data-ref.c (split_constant_offset_1): Likewise.
1393 (dr_analyze_innermost): Likewise.
1394 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1395 * tree-sra.c (ipa_sra_check_caller): Likewise.
1396 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1397 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1398 (instrument_bool_enum_load, instrument_object_size): Likewise.
1399 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1400 to get_inner_reference.
1401 * hsa-gen.c (gen_hsa_addr): Likewise.
1402 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1403 * tsan.c (instrument_expr): Likewise.
1404 * match.pd: Update call to ptr_difference_const.
1406 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1407 Alan Hayward <alan.hayward@arm.com>
1408 David Sherwood <david.sherwood@arm.com>
1410 * fold-const.c (fold_comparison): Track sizes and offsets as
1411 poly_int64s rather than HOST_WIDE_INTs when folding address
1414 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1415 Alan Hayward <alan.hayward@arm.com>
1416 David Sherwood <david.sherwood@arm.com>
1418 * expr.h (get_bit_range): Return the bitstart and bitend as
1419 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1420 as a poly_int64 rather than a HOST_WIDE_INT.
1421 * expr.c (get_bit_range): Likewise.
1422 (expand_assignment): Update call accordingly.
1423 * fold-const.c (optimize_bit_field_compare): Likewise.
1425 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1426 Alan Hayward <alan.hayward@arm.com>
1427 David Sherwood <david.sherwood@arm.com>
1429 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1430 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1431 as an optional final parameter.
1432 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1433 (aarch64_legitimate_address_p): Likewise.
1434 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1435 instead of an rtx code.
1436 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1437 (aarch64_legitimate_address_hook_p): Likewise.
1438 (aarch64_print_ldpstp_address): Likewise.
1439 (aarch64_print_operand_address): Likewise.
1440 (aarch64_address_cost): Likewise.
1441 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1442 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1443 (aarch64_mem_pair_lanes_operand): Likewise.
1445 2017-12-20 Richard Biener <rguenther@suse.de>
1447 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1448 update_stmt_if_modified.
1450 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1452 PR tree-optimization/83491
1453 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1454 before walking uses. Improve coding style and comments.
1456 2017-12-20 Tom de Vries <tom@codesourcery.com>
1458 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1460 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1463 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1465 (pass_ipa_free_fn_summary): Add small_p private data member,
1466 initialize to false in the ctor.
1467 (pass_ipa_free_fn_summary::clone,
1468 pass_ipa_free_fn_summary::set_pass_param,
1469 pass_ipa_free_fn_summary::gate): New methods.
1470 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1471 | TODO_dump_symtab if small_p.
1472 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1473 entry and add another instance of the pass with false parm after
1475 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1476 ipa_free_fn_summary here.
1478 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1480 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1482 2017-12-20 Martin Sebor <msebor@redhat.com>
1485 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1486 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1487 (expand_builtin_strncmp): Same.
1489 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1492 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1493 allowed debug stmts before labels.
1494 (expand_gimple_basic_block): Likewise.
1495 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1496 * gimple-iterator.h (gsi_after_labels): Likewise.
1497 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1498 rename reused variable, and simplify using gsi_move_before.
1499 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1500 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1501 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1502 (gimple_verify_flow_info, gimple_block_label): Likewise.
1503 (make_blocks): Move debug markers after adjacent labels.
1504 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1505 allowed debug insns outside blocks.
1506 * df-scan.c (df_insn_delete): Likewise.
1507 * lra-constraints.c (update_ebb_live_info): Likewise.
1508 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1509 (vt_initialize, delete_vta_debug_insns): Likewise.
1510 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1512 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1513 Alan Hayward <alan.hayward@arm.com>
1514 David Sherwood <david.sherwood@arm.com>
1516 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1517 (round_up_to_byte_boundary): Likewise.
1518 * expr.h (get_bit_range): Add temporary shim.
1519 * gimple-ssa-store-merging.c (store_operand_info): Change the
1520 bitsize, bitpos, bitregion_start and bitregion_end fields from
1521 unsigned HOST_WIDE_INT to poly_uint64.
1522 (merged_store_group): Likewise load_align_base.
1523 (compatible_load_p, compatible_load_p): Update accordingly.
1524 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1525 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1526 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1527 bitregion_start and bitregion_end as poly_uint64s rather than
1528 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1529 (handled_load): Take the bitsize, bitpos,
1530 bitregion_start and bitregion_end as poly_uint64s rather than
1531 unsigned HOST_WIDE_INTs.
1532 (pass_store_merging::process_store): Update call to
1533 mem_valid_for_store_merging.
1535 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1536 Alan Hayward <alan.hayward@arm.com>
1537 David Sherwood <david.sherwood@arm.com>
1539 * builtins.c (get_object_alignment_2): Track polynomial offsets
1540 and sizes. Update the alignment handling.
1542 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1543 Alan Hayward <alan.hayward@arm.com>
1544 David Sherwood <david.sherwood@arm.com>
1546 * tree.h (get_inner_reference): Add a version that returns the
1547 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1548 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1549 the case in which bitpos is not associated with the first byte.
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-affine.h (get_inner_reference_aff): Return the size as a
1557 * tree-affine.c (get_inner_reference_aff): Likewise.
1558 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1559 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1561 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1562 Alan Hayward <alan.hayward@arm.com>
1563 David Sherwood <david.sherwood@arm.com>
1565 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1566 HOST_WIDE_INT rather than a poly_int64.
1568 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1569 Alan Hayward <alan.hayward@arm.com>
1570 David Sherwood <david.sherwood@arm.com>
1572 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1573 HOST_WIDE_INT to poly_int64_pod.
1574 (perform_symbolic_merge): Update accordingly.
1575 (bswap_replace): Likewise.
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 * tree-affine.h (aff_tree::offset): Change from widest_int
1583 (wide_int_ext_for_comb): Delete.
1584 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1585 constants as poly_widest_int rather than widest_int.
1586 (aff_combination_constant_multiple_p): Return the multiplier
1587 as a poly_widest_int.
1588 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1590 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1591 version static and add an overload for poly_widest_int.
1592 (aff_combination_const, aff_combination_add_cst)
1593 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1594 the constants as poly_widest_int rather than widest_int.
1595 (tree_to_aff_combination): Generalize INTEGER_CST case to
1597 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1598 (aff_combination_add_product, aff_combination_mult): Handle
1600 (aff_combination_constant_multiple_p): Return the multiplier
1601 as a poly_widest_int.
1602 * tree-predcom.c (determine_offset): Return the offset as a
1604 (split_data_refs_to_components, suitable_component_p): Update
1606 (valid_initializer_p): Update call to
1607 aff_combination_constant_multiple_p.
1608 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1609 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1610 as a poly_int64 rather than a HOST_WIDE_INT.
1611 (get_address_cost): Handle polynomial offsets.
1612 (iv_elimination_compare_lt): Likewise.
1613 (rewrite_use_nonlinear_expr): Likewise.
1615 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1616 Alan Hayward <alan.hayward@arm.com>
1617 David Sherwood <david.sherwood@arm.com>
1619 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1620 as a poly_int64_pod rather than a HOST_WIDE_INT.
1621 (get_addr_base_and_unit_offset): Likewise.
1622 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1623 (get_addr_base_and_unit_offset): Likewise.
1624 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1625 to poly_int64 in example.
1626 * fold-const.c (fold_binary_loc): Update call to
1627 get_addr_base_and_unit_offset.
1628 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1629 (maybe_canonicalize_mem_ref_addr): Likewise.
1630 (gimple_fold_stmt_to_constant_1): Likewise.
1631 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1633 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1634 * match.pd: Likewise.
1635 * omp-low.c (lower_omp_target): Likewise.
1636 * tree-sra.c (build_ref_for_offset): Likewise.
1637 (build_debug_ref_for_model): Likewise.
1638 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1639 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1640 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1641 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1642 (constant_pointer_difference): Likewise.
1643 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1644 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1645 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1646 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1647 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1648 (set_ssa_val_to): Likewise.
1649 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1650 (maybe_diag_stxncpy_trunc): Likewise.
1651 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1652 * tree.c (build_simple_mem_ref_loc): Likewise.
1653 (array_at_struct_end_p): Likewise.
1655 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1656 Alan Hayward <alan.hayward@arm.com>
1657 David Sherwood <david.sherwood@arm.com>
1659 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1660 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1661 (get_ref_base_and_extent_hwi): Declare.
1662 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1663 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1664 (get_ref_base_and_extent_hwi): New function.
1665 * cfgexpand.c (expand_debug_expr): Update call to
1666 get_ref_base_and_extent.
1667 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1668 * gimple-fold.c (get_base_constructor): Return the offset as a
1669 poly_int64_pod rather than a HOST_WIDE_INT.
1670 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1671 * ipa-polymorphic-call.c
1672 (ipa_polymorphic_call_context::set_by_invariant)
1673 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1674 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1675 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1676 rather than get_ref_base_and_extent.
1677 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1678 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1679 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1681 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1682 call to get_ref_base_and_extent.
1683 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1684 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1685 (stmt_kills_ref_p): Likewise.
1686 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1687 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1689 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1690 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1691 when calling native_encode_expr.
1692 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1693 call to get_ref_base_and_extent.
1694 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1695 get_ref_base_and_extent.
1696 * var-tracking.c (track_expr_p): Likewise.
1698 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1699 Alan Hayward <alan.hayward@arm.com>
1700 David Sherwood <david.sherwood@arm.com>
1702 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1703 HOST_WIDE_INT to poly_int64_pod.
1704 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1705 polynomail parameter offsets.
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 * gengtype.c (main): Handle poly_int64_pod.
1712 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1713 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1714 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1715 from HOST_WIDE_INT to poly_int64_pod.
1716 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1717 (copy_cfa): New function.
1718 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1719 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1720 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1721 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1722 (initial_return_save): Treat offsets as poly_ints.
1723 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1724 offset is nonconstant.
1725 (reg_save): Take the offset as a poly_int64. Fall back to
1726 DW_CFA_expression for nonconstant offsets.
1727 (queue_reg_save): Take the offset as a poly_int64.
1728 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1730 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1731 Alan Hayward <alan.hayward@arm.com>
1732 David Sherwood <david.sherwood@arm.com>
1734 * rtl.h (operand_subword, operand_subword_force): Take the offset
1735 as a poly_uint64 an unsigned int.
1736 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1738 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1739 Alan Hayward <alan.hayward@arm.com>
1740 David Sherwood <david.sherwood@arm.com>
1742 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1743 'p' format code. Use INT_LIST rather than SUBREG as the example of
1744 a code with an XINT and an XEXP. Remove the implication that
1745 accessing an rtx field using XINT is expected to work.
1746 * rtl.def (SUBREG): Change format from "ei" to "ep".
1747 * rtl.h (rtunion::rt_subreg): New field.
1748 (XCSUBREG): New macro.
1749 (SUBREG_BYTE): Use it.
1750 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1751 Update constructor accordingly.
1752 (subreg_shape::operator ==): Update accordingly.
1753 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1754 than an unsigned int.
1755 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1756 a poly_uint64 rather than an unsigned int.
1757 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1758 than an unsigned int.
1759 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1760 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1761 an unsigned int. Take the sizes as poly_uint64s.
1762 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1763 an unsigned int. Take the shift as a poly_uint64 rather than
1765 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1766 as a poly_uint64 rather than an unsigned int.
1767 (simplify_subreg_regno): Likewise.
1768 (byte_lowpart_offset): Return the memory offset as a poly_int64
1770 (subreg_memory_offset): Likewise. Take the subreg offset as a
1771 poly_uint64 rather than an unsigned int.
1772 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1773 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1774 poly_uint64 rather than an unsigned int.
1775 * rtl.c (rtx_format): Describe 'p' in comment.
1776 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1777 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1778 offset as a poly_uint64 rather than an unsigned int.
1779 (byte_lowpart_offset): Return the memory offset as a poly_int64
1781 (subreg_memory_offset): Likewise. Take the subreg offset as a
1782 poly_uint64 rather than an unsigned int.
1783 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1784 mode sizes as poly_uint64s rather than unsigned ints. Return a
1785 poly_uint64 rather than an unsigned int.
1786 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1787 (copy_insn_1): Handle 'p'.
1788 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1789 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
1790 an unsigned int. Return the shift in the same way.
1791 (subreg_lsb): Return the shift as a poly_uint64 rather than an
1793 (subreg_size_offset_from_lsb): Take the sizes and shift as
1794 poly_uint64s rather than unsigned ints. Return the offset as
1796 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
1797 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
1799 * rtlhash.c (add_rtx): Handle 'p'.
1800 * genemit.c (gen_exp): Likewise.
1801 * gengenrtl.c (type_from_format, gendef): Likewise.
1802 * gensupport.c (subst_pattern_match, get_alternatives_number)
1803 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
1804 (subst_dup): Likewise.
1805 * gengtype.c (adjust_field_rtx_def): Likewise.
1806 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
1807 (match_pattern_2): Likewise.
1808 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
1809 (rtx_test::subreg_field): New function.
1810 (operator ==, safe_to_hoist_p, transition_parameter_type)
1811 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1812 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1813 * genpeep.c (match_rtx): Likewise.
1814 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1815 (rtx_writer::print_rtx_operand): Handle 'p'.
1816 (print_value): Handle SUBREG.
1817 * read-rtl.c (apply_int_iterator): Likewise.
1818 (rtx_reader::read_rtx_operand): Handle 'p'.
1819 * alias.c (rtx_equal_for_memref_p): Likewise.
1820 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1821 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1823 * calls.c (expand_call): Likewise.
1824 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1825 (make_extraction, gen_lowpart_for_combine): Likewise.
1826 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1828 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1829 rather than an unsigned int. Treat subreg offsets as poly_ints.
1830 (exp_equiv_p): Handle 'p'.
1831 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1832 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1833 * dse.c (find_shift_sequence): Likewise.
1834 * dwarf2out.c (rtl_for_decl_location): Likewise.
1835 * expmed.c (extract_low_bits): Likewise.
1836 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1837 (expand_expr_real_2): Likewise.
1838 * final.c (alter_subreg): Likewise.
1839 (leaf_renumber_regs_insn): Handle 'p'.
1840 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1841 Treat subreg offsets as poly_ints.
1842 * fwprop.c (forward_propagate_and_simplify): Likewise.
1843 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
1844 * ira.c (get_subreg_tracking_sizes): Likewise.
1845 * ira-conflicts.c (go_through_subreg): Likewise.
1846 * ira-lives.c (process_single_reg_class_operands): Likewise.
1847 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1848 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1849 as a poly_uint64 rather than an unsigned int.
1850 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1851 subreg offsets as poly_ints.
1852 * lra-constraints.c (operands_match_p): Handle 'p'.
1853 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1854 * lra-spills.c (assign_mem_slot): Likewise.
1855 * postreload.c (move2add_valid_value_p): Likewise.
1856 * recog.c (general_operand, indirect_operand): Likewise.
1857 * regcprop.c (copy_value, maybe_mode_change): Likewise.
1858 (copyprop_hardreg_forward_1): Likewise.
1859 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
1860 (record_subregs_of_mode): Likewise.
1861 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
1862 * reload.c (operands_match_p): Handle 'p'.
1863 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
1864 * reload1.c (alter_reg, choose_reload_regs): Likewise.
1865 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1866 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1867 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1868 (simplify_const_poly_int_tests<N>::run): Likewise.
1869 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1870 a poly_uint64 rather than an unsigned int.
1871 * valtrack.c (debug_lowpart_subreg): Likewise.
1872 * var-tracking.c (var_lowpart): Likewise.
1873 (loc_cmp): Handle 'p'.
1875 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1876 Alan Hayward <alan.hayward@arm.com>
1877 David Sherwood <david.sherwood@arm.com>
1879 * ira.c (get_subreg_tracking_sizes): New function.
1880 (init_live_subregs): Take an integer size rather than a register.
1881 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1882 to init_live_subregs.
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 * expr.c (store_constructor_field): Change bitsize from a
1889 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1890 HOST_WIDE_INT to a poly_int64.
1891 (store_constructor): Change size from a HOST_WIDE_INT to
1893 (store_field): Likewise bitsize and bitpos.
1895 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1896 Alan Hayward <alan.hayward@arm.com>
1897 David Sherwood <david.sherwood@arm.com>
1899 * expmed.h (store_bit_field): Change bitregion_start and
1900 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1901 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1902 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1903 (store_fixed_bit_field, store_split_bit_field): Likewise.
1904 * expr.c (store_constructor_field, store_field): Likewise.
1905 (optimize_bitfield_assignment_op): Likewise. Make the same change
1906 to bitsize and bitpos.
1907 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1908 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1909 same change in the constructor arguments.
1910 (get_best_mode): Change bitregion_start and bitregion_end from
1911 unsigned HOST_WIDE_INT to poly_uint64.
1912 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1913 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1915 (bit_field_mode_iterator::next_mode): Update for new types
1916 of m_bitregion_start and m_bitregion_end.
1917 (get_best_mode): Change bitregion_start and bitregion_end from
1918 unsigned HOST_WIDE_INT to poly_uint64.
1920 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1921 Alan Hayward <alan.hayward@arm.com>
1922 David Sherwood <david.sherwood@arm.com>
1924 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1925 accepts poly_uint64 offsets.
1926 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1927 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1928 * expmed.c (lowpart_bit_field_p): Likewise.
1929 (extract_bit_field_as_subreg): New function, split out from...
1930 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1931 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1932 extractions, check that BITSIZE matches the size of the extracted
1933 value and that BITNUM is an exact multiple of that size.
1934 If all else fails, try forcing the value into memory if
1935 BITNUM is variable, and adjusting the address so that the
1936 offset is constant. Split the part that can only handle constant
1937 bitsize and bitnum out into...
1938 (extract_integral_bit_field): ...this new function.
1939 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1940 rather than unsigned HOST_WIDE_INTs.
1942 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1943 Alan Hayward <alan.hayward@arm.com>
1944 David Sherwood <david.sherwood@arm.com>
1946 * expmed.h (store_bit_field): Take bitsize and bitnum as
1947 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1948 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1949 that returns the byte size.
1950 (store_bit_field_1): Take bitsize and bitnum as
1951 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1952 to simple_mem_bitfield_p. Split the part that can only handle
1953 constant bitsize and bitnum out into...
1954 (store_integral_bit_field): ...this new function.
1955 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1956 than unsigned HOST_WIDE_INTs.
1957 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1959 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1960 Alan Hayward <alan.hayward@arm.com>
1961 David Sherwood <david.sherwood@arm.com>
1963 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1964 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1966 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1967 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1968 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1969 offset as a poly_int64 rather than an int.
1970 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1971 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1972 * lra-constraints.c (equiv_address_substitution): Track offsets
1974 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1975 (curr_insn_transform): Handle the new form of sp_offset.
1976 * lra-eliminations.c (lra_elim_table): Change previous_offset
1977 and offset from HOST_WIDE_INT to poly_int64.
1978 (print_elim_table, update_reg_eliminate): Update accordingly.
1979 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1980 (get_elimination): Update accordingly.
1981 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1982 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1983 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1984 poly_int64 offsets generally.
1985 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1986 (mark_not_eliminable, init_elimination): Update accordingly.
1987 (remove_reg_equal_offset_note): Return a bool and pass the new
1988 offset back by pointer as a poly_int64.
1989 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1990 rather than a HOST_WIDE_INT.
1991 (do_remat): Track offsets poly_int64s.
1992 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1994 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1995 Alan Hayward <alan.hayward@arm.com>
1996 David Sherwood <david.sherwood@arm.com>
1998 * rtl.h (mem_attrs): Add a default constructor. Change size and
1999 offset from HOST_WIDE_INT to poly_int64.
2000 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2001 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2002 (widen_memory_access): Take the sizes and offsets as poly_int64s
2003 rather than HOST_WIDE_INTs.
2004 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2005 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2007 (adjust_offset_for_component_ref): Change the offset from a
2008 HOST_WIDE_INT to a poly_int64.
2009 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2010 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2011 * dce.c (find_call_stack_args): Likewise.
2012 * dse.c (record_store): Likewise.
2013 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2014 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2015 * read-rtl-function.c (test_loading_mem): Likewise.
2016 * rtlanal.c (may_trap_p_1): Likewise.
2017 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2018 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2019 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2020 (mem_attrs::mem_attrs): New function.
2021 (set_mem_attributes_minus_bitpos): Change bitpos from a
2022 HOST_WIDE_INT to poly_int64.
2023 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2024 (clear_mem_offset, clear_mem_size, change_address)
2025 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2026 initialize mem_attrs.
2027 (set_mem_offset, set_mem_size, adjust_address_1)
2028 (adjust_automodify_address_1, offset_address, widen_memory_access):
2029 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2031 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2032 Alan Hayward <alan.hayward@arm.com>
2033 David Sherwood <david.sherwood@arm.com>
2035 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2036 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2037 rather than 0 to represent an unknown size. Assert that the size
2038 is known when the mode isn't BLKmode.
2039 (may_trap_p_1): Use -1 for unknown sizes.
2040 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2042 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2043 Alan Hayward <alan.hayward@arm.com>
2044 David Sherwood <david.sherwood@arm.com>
2046 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2047 to poly_int64. Update commentary for positions_needed.large.
2048 (read_info_type): Change offset and width from HOST_WIDE_INT
2050 (set_usage_bits): Likewise.
2051 (canon_address): Return the offset as a poly_int64 rather than
2052 a HOST_WIDE_INT. Use strip_offset_and_add.
2053 (set_all_positions_unneeded, any_positions_needed_p): Use
2054 positions_needed.large to track stores with non-constant widths.
2055 (all_positions_needed_p): Likewise. Take the offset and width
2056 as poly_int64s rather than ints. Assert that rhs is nonnull.
2057 (record_store): Cope with non-constant offsets and widths.
2058 Nullify the rhs of an earlier store if we can't tell which bytes
2060 (find_shift_sequence): Take the access_size and shift as poly_int64s
2062 (get_stored_val): Take the read_offset and read_width as poly_int64s
2063 rather than HOST_WIDE_INTs.
2064 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2065 non-constant offsets and widths.
2067 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2068 Alan Hayward <alan.hayward@arm.com>
2069 David Sherwood <david.sherwood@arm.com>
2071 * inchash.h (inchash::hash::add_poly_int): New function.
2072 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2073 Use poly_int64 rather than HOST_WIDE_INT.
2074 (ao_ref::max_size_known_p): New function.
2075 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2076 rather than HOST_WIDE_INT.
2077 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2078 to temporaries until its interface is adjusted to match.
2079 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2080 (aliasing_component_refs_p, decl_refs_may_alias_p)
2081 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2082 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2083 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2085 * alias.c (ao_ref_from_mem): Likewise.
2086 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2087 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2088 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2089 (maybe_trim_complex_store, maybe_trim_constructor_store)
2090 (live_bytes_read, dse_classify_store): Likewise.
2091 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2092 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2093 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2094 (vn_reference_lookup_3): Likewise.
2095 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
2097 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2098 Alan Hayward <alan.hayward@arm.com>
2099 David Sherwood <david.sherwood@arm.com>
2101 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2102 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2103 instead of ranges_overlap_p.
2105 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2106 Alan Hayward <alan.hayward@arm.com>
2107 David Sherwood <david.sherwood@arm.com>
2109 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2110 sizes as poly_int64s rather than HOST_WIDE_INTs.
2112 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2113 Alan Hayward <alan.hayward@arm.com>
2114 David Sherwood <david.sherwood@arm.com>
2116 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2117 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2118 * gimple-fold.c (fold_ctor_reference): Likewise.
2120 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2121 Alan Hayward <alan.hayward@arm.com>
2122 David Sherwood <david.sherwood@arm.com>
2124 * target.def (dwarf_poly_indeterminate_value): New hook.
2125 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2126 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2127 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2128 * doc/tm.texi: Regenerate.
2129 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2130 offset as a poly_int64.
2131 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2132 offset as a poly_int64.
2133 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2134 Take the offset as a poly_int64.
2135 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2136 (frame_pointer_fb_offset): Change to a poly_int64.
2137 (int_loc_descriptor): Take the offset as a poly_int64. Use
2138 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2139 (based_loc_descr): Take the offset as a poly_int64.
2140 Use strip_offset_and_add to handle (plus X (const)).
2141 Use new_reg_loc_descr instead of an open-coded version of the
2142 previous implementation.
2143 (mem_loc_descriptor): Handle CONST_POLY_INT.
2144 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2145 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2147 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2148 Alan Hayward <alan.hayward@arm.com>
2149 David Sherwood <david.sherwood@arm.com>
2151 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2152 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2153 * inchash.h (inchash::hash::add_poly_hwi): New function.
2154 * gengtype.c (main): Register poly_int64.
2155 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2156 offset as a poly_int.
2157 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2158 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2159 offset as a poly_int64.
2160 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2161 * print-rtl.c (print_poly_int): New function.
2162 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2164 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2166 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2167 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2168 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2169 Enforce track_offset_p for parts of a PARALLEL.
2170 (vt_add_function_parameter): Use const_offset for the final
2171 offset to track. Use get_tracked_reg_offset for the parts
2174 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2175 Alan Hayward <alan.hayward@arm.com>
2176 David Sherwood <david.sherwood@arm.com>
2178 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2179 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2180 * doc/tm.texi: Regenerate.
2181 * hooks.h (hook_bool_uint_uint_true): Delete.
2182 (hook_bool_puint64_puint64_true): Declare.
2183 * hooks.c (hook_bool_uint_uint_true): Delete.
2184 (hook_bool_puint64_puint64_true): New function.
2185 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2186 instead of unsigned ints.
2187 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2188 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2190 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2191 Alan Hayward <alan.hayward@arm.com>
2192 David Sherwood <david.sherwood@arm.com>
2194 * optabs.h (expand_operand): Add an int_value field.
2195 (create_expand_operand): Add an int_value parameter and use it
2196 to initialize the new expand_operand field.
2197 (create_integer_operand): Replace with a declaration of a function
2198 that accepts poly_int64s. Move the implementation to...
2199 * optabs.c (create_integer_operand): ...here.
2200 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2201 the mode preserves the value of int_value, instead of calling
2202 const_int_operand on the rtx. Use gen_int_mode to generate
2205 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2206 Alan Hayward <alan.hayward@arm.com>
2207 David Sherwood <david.sherwood@arm.com>
2209 * dumpfile.h (dump_dec): Declare.
2210 * dumpfile.c (dump_dec): New function.
2211 * pretty-print.h (pp_wide_integer): Turn into a function and
2212 declare a poly_int version.
2213 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2215 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2216 Alan Hayward <alan.hayward@arm.com>
2217 David Sherwood <david.sherwood@arm.com>
2219 * doc/generic.texi (POLY_INT_CST): Document.
2220 * tree.def (POLY_INT_CST): New tree code.
2221 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2222 * tree-core.h (tree_poly_int_cst): New struct.
2223 (tree_node): Add a poly_int_cst field.
2224 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2225 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2226 instead of a wide_int_ref.
2227 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2229 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2230 instead of an unsigned HOST_WIDE_INT.
2231 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2232 (ptrdiff_tree_p): Declare.
2233 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2234 extern inline implementations if the target doesn't use POLY_INT_CST.
2235 (poly_int_tree_p): New function.
2236 (wi::unextended_tree): New class.
2237 (wi::int_traits <unextended_tree>): New override.
2238 (wi::extended_tree): Add a default constructor.
2239 (wi::extended_tree::get_tree): New function.
2240 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2241 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2242 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2243 (wi::tree_to_poly_wide_ref): New typedefs.
2244 (wi::ints_for): Provide overloads for extended_tree and
2246 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2247 (wi::to_wide): New functions.
2248 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2249 * tree.c (poly_int_cst_hasher): New struct.
2250 (poly_int_cst_hash_table): New variable.
2251 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2252 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2254 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2255 (init_ttree): Initialize poly_int_cst_hash_table.
2256 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2257 a poly_int64 instead of a HOST_WIDE_INT.
2258 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2259 instead of an unsigned HOST_WIDE_INT.
2260 (wide_int_to_tree): Rename to...
2261 (wide_int_to_tree_1): ...this.
2262 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2263 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2264 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2265 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2266 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2267 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2269 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2270 (streamer_read_tree_body): Likewise.
2271 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2272 (streamer_write_tree_body): Likewise.
2273 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2274 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2275 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2276 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2277 * gimple-expr.h (is_gimple_constant): Likewise.
2278 * gimplify.c (maybe_with_size_expr): Likewise.
2279 * print-tree.c (print_node): Likewise.
2280 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2281 * tree-pretty-print.c (dump_generic_node): Likewise.
2282 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2283 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2284 * tree-vrp.c (compare_values_warnv): Likewise.
2285 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2286 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2287 (force_expr_to_var_cost): Likewise.
2288 * tree-ssa-loop.c (for_each_index): Likewise.
2289 * fold-const.h (build_invariant_address, size_int_kind): Take a
2290 poly_int64 instead of a HOST_WIDE_INT.
2291 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2292 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2294 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2295 (int_const_binop_2): New function, split out from...
2296 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2297 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2298 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2299 * gimple-ssa-strength-reduction.c (slsr_process_add)
2300 (slsr_process_mul): Check for INTEGER_CSTs before using them
2302 * stor-layout.c (bits_from_bytes): New function.
2303 (bit_from_pos): Use it.
2304 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2305 by BITS_PER_UNIT to get the TYPE_SIZE.
2306 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2307 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2309 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2310 Alan Hayward <alan.hayward@arm.com>
2311 David Sherwood <david.sherwood@arm.com>
2313 * doc/rtl.texi (const_poly_int): Document. Also document the
2314 rtl sharing behavior.
2315 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2316 * rtl.h (const_poly_int_def): New struct.
2317 (rtx_def::u): Add a cpi field.
2318 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2319 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2320 (wi::rtx_to_poly_wide_ref): New typedef
2321 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2322 (poly_int_rtx_p): New functions.
2323 (trunc_int_for_mode): Declare a poly_int64 version.
2324 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2325 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2327 (strip_offset): Declare.
2328 (strip_offset_and_add): New function.
2329 * rtl.def (CONST_POLY_INT): New rtx code.
2330 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2331 (shared_const_p): Use poly_int_rtx_p.
2332 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2334 (gen_int_shift_amount): Likewise.
2335 * emit-rtl.c (const_poly_int_hasher): New class.
2336 (const_poly_int_htab): New variable.
2337 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2338 (const_poly_int_hasher::hash): New function.
2339 (const_poly_int_hasher::equal): Likewise.
2340 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2341 (immed_wide_int_const): Rename to...
2342 (immed_wide_int_const_1): ...this and make static.
2343 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2344 instead of a wide_int_ref.
2345 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2346 (gen_lowpart_common): Handle CONST_POLY_INT.
2347 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2348 * cselib.c (cselib_hash_rtx): Likewise.
2349 * dwarf2out.c (const_ok_for_output_1): Likewise.
2350 * expr.c (convert_modes): Likewise.
2351 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2352 * rtlhash.c (add_rtx): Likewise.
2353 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2354 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2355 Handle existing CONST_POLY_INT rtxes.
2356 * expmed.h (expand_shift): Take a poly_int64 instead of a
2358 * expmed.c (expand_shift): Likewise.
2359 * rtlanal.c (strip_offset): New function.
2360 (commutative_operand_precedence): Give CONST_POLY_INT the same
2361 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2363 * rtl-tests.c (const_poly_int_tests): New struct.
2364 (rtl_tests_c_tests): Use it.
2365 * simplify-rtx.c (simplify_const_unary_operation): Handle
2367 (simplify_const_binary_operation): Likewise.
2368 (simplify_binary_operation_1): Fold additions of symbolic constants
2369 and CONST_POLY_INTs.
2370 (simplify_subreg): Handle extensions and truncations of
2372 (simplify_const_poly_int_tests): New struct.
2373 (simplify_rtx_c_tests): Use it.
2374 * wide-int.h (storage_ref): Add default constructor.
2375 (wide_int_ref_storage): Likewise.
2376 (trailing_wide_ints): Use GTY((user)).
2377 (trailing_wide_ints::operator[]): Add a const version.
2378 (trailing_wide_ints::get_precision): New function.
2379 (trailing_wide_ints::extra_size): Likewise.
2381 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2382 Alan Hayward <alan.hayward@arm.com>
2383 David Sherwood <david.sherwood@arm.com>
2385 * emit-rtl.h (gen_int_shift_amount): Declare.
2386 * emit-rtl.c (gen_int_shift_amount): New function.
2387 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2389 * calls.c (shift_return_value): Likewise.
2390 * cse.c (fold_rtx): Likewise.
2391 * dse.c (find_shift_sequence): Likewise.
2392 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2393 (expand_shift, expand_smod_pow2): Likewise.
2394 * lower-subreg.c (shift_cost): Likewise.
2395 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2396 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2397 (expand_vec_perm_var): Likewise.
2398 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2399 (simplify_binary_operation_1): Likewise.
2400 * combine.c (try_combine, find_split_point, force_int_to_mode)
2401 (simplify_shift_const_1, simplify_shift_const): Likewise.
2402 (change_zero_ext): Likewise. Use simplify_gen_binary.
2404 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2406 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2408 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2410 * doc/invoke.texi (ARM Options): Document accepted extension options
2411 for -march=armv8.3-a.
2413 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2416 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2417 or --with-float=softfp, set the default CPU to arm10e.
2419 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2421 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2422 * config/visium/predicates.md (const_shift_operand): Likewise.
2423 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2424 (visium_legitimize_reload_address): Likewise.
2426 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2429 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2431 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2434 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2436 2017-12-20 Julia Koval <julia.koval@intel.com>
2438 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2439 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2440 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2441 * config.gcc: Include vpclmulqdqintrin.h.
2442 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2443 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2445 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2446 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2447 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2448 * config/i386/i386.c (isa2_opts): Add -mcx16.
2449 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2450 (ix86_option_override_internal): Move mcx16 to flags2.
2451 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2452 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2453 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2454 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2455 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2456 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2457 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2458 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2459 * doc/invoke.texi: Add -mvpclmulqdq.
2461 2017-12-20 Tom de Vries <tom@codesourcery.com>
2464 * config/i386/i386.c (ix86_static_chain): Move
2465 DECL_STATIC_CHAIN test ...
2466 * calls.c (rtx_for_static_chain): ... here. New function.
2467 * calls.h (rtx_for_static_chain): Declare.
2468 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2469 instead of targetm.calls.static_chain.
2470 * df-scan.c (df_get_entry_block_def_set): Same.
2472 2017-12-19 Tom de Vries <tom@codesourcery.com>
2474 PR tree-optimization/83493
2475 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2478 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2480 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2481 inform with hardcoded english plural handling.
2483 2017-12-18 Jeff Law <law@redhat.com>
2485 PR tree-optimization/83477
2486 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2487 a non-virtual PHI, always push a new range.
2489 2017-12-19 Martin Sebor <msebor@redhat.com>
2492 * builtins.c (compute_objsize): Handle non-constant offsets.
2494 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2496 PR tree-optimization/83444
2497 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2498 character load case, if get_stridx on MEM_REF's operand doesn't
2499 look usable, retry with get_addr_stridx.
2501 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2504 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2505 (variable_tracking_main_1): Keep markers even when VTA fails.
2508 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2509 even if there are markers before it.
2510 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2512 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2514 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2515 typical order conditions.
2516 * sel-sched.c (extract_new_fences_from): Likewise.
2517 * config/visium/constraints.md (J, K, L): Likewise.
2518 * config/visium/predicates.md (const_shift_operand): Likewise.
2519 * config/visium/visium.c (visium_legitimize_address,
2520 visium_legitimize_reload_address): Likewise.
2521 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2522 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2523 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2524 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2525 SET_DUMP_DETAIL): Likewise.
2526 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2527 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2528 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2529 avr_set_core_architecture, avr_set_current_function,
2530 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2531 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2532 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2533 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2534 avr_map_decompose, avr_fold_builtin): Likewise.
2535 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2536 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2537 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2538 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2539 * config/m32c/m32c.c (m32c_conditional_register_usage,
2540 m32c_address_cost): Likewise.
2541 * config/m32c/predicates.md (shiftcount_operand,
2542 longshiftcount_operand): Likewise.
2543 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2544 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2545 can_use_cdx_ldstw): Likewise.
2546 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2547 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2549 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2550 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2551 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2552 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2553 * config/vax/vax.c (adjacent_operands_p): Likewise.
2554 * config/ft32/constraints.md (L, b, KA): Likewise.
2555 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2557 * cfgexpand.c (expand_stack_alignment): Likewise.
2558 * gcse.c (insert_expr_in_table): Likewise.
2559 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2560 * cgraphunit.c (cgraph_node::expand): Likewise.
2561 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2562 * emit-rtl.c (add_insn): Likewise.
2563 * input.c (dump_location_info): Likewise.
2564 * passes.c (NEXT_PASS): Likewise.
2565 * read-rtl-function.c (parse_note_insn_name,
2566 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2568 * sched-rgn.c (sched_rgn_init): Likewise.
2569 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2570 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2571 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2573 * explow.c (eliminate_constant_term): Likewise.
2574 * final.c (leaf_renumber_regs_insn): Likewise.
2575 * cfgrtl.c (print_rtl_with_bb): Likewise.
2576 * genhooks.c (emit_init_macros): Likewise.
2577 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2578 * tree-data-ref.c (conflict_fn): Likewise.
2579 * selftest.c (assert_streq): Likewise.
2580 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2581 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2582 fold_binary_loc, multiple_of_p): Likewise.
2583 * reload.c (push_reload, find_equiv_reg): Likewise.
2584 * et-forest.c (et_nca, et_below): Likewise.
2585 * dbxout.c (dbxout_symbol_location): Likewise.
2586 * reorg.c (relax_delay_slots): Likewise.
2587 * dojump.c (do_compare_rtx_and_jump): Likewise.
2588 * gengtype-parse.c (type): Likewise.
2589 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2590 simplify_const_relational_operation): Likewise.
2591 * reload1.c (do_output_reload): Likewise.
2592 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2593 * gengtype.c (type_for_name): Likewise.
2594 * gimple-ssa-sprintf.c (format_directive): Likewise.
2596 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2599 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2600 accessing it. Adjust comment.
2602 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2605 * predict.c (zero_one_minusone): New function.
2606 (apply_return_prediction): Avoid return prediction for functions
2607 returning only -1, 0 and 1 values, unless they only return -1 and 0
2610 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2612 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2613 fall-through warning.
2614 (arc_compute_frame_size): Remove unused variables.
2615 (arc_print_operand): Fix fprintif format.
2616 (arc_can_follow_jump): Clean fall-through warning.
2618 2017-12-19 Marek Polacek <polacek@redhat.com>
2621 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2624 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2626 * config/arc/arc.c (overriderregs): New variable.
2627 (arc_override_options): Track fixed/call saved/call options.
2628 (arc_conditional_register_usage): Check against overrideregs
2629 variable whenever we change register properties.
2631 2017-12-19 Nathan Sidwell <nathan@acm.org>
2633 * opts.c (finish_options): Don't prefix dump_base_name if it
2634 already contains directories.
2636 2017-12-19 Martin Liska <mliska@suse.cz>
2638 PR rtl-optimization/82675
2639 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2640 more element in sbitmap.
2642 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2644 * gimplify.c (gimplify_expr): Use error_operand_p.
2646 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2649 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2652 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2654 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2655 the jump is to a label.
2657 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2659 PR tree-optimization/83444
2660 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2665 * ipa-inline.c (flatten_remove_node_hook): New function.
2666 (ipa_inline): Keep only nodes with flatten attribute at the end of
2667 the array in the order from ipa_reverse_postorder, only walk that
2668 portion of array for flattening, if there is more than one such
2669 node, temporarily register a removal hook and ignore removed nodes.
2671 PR tree-optimization/80631
2672 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2673 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2674 IFN_REDUC_MAX or IFN_REDUC_MIN.
2676 2017-12-18 Jeff Law <law@redhat.com>
2678 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2679 record symbolic equivalences from backedges in the CFG.
2682 2017-11-19 Jeff Law <law@redhat.com>
2684 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2685 of degenerates resulting from ignoring an edge.
2687 2017-12-18 Martin Sebor <msebor@redhat.com>
2690 PR tree-optimization/78450
2691 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2692 (handle_builtin_strlen): Call it.
2694 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2696 PR rtl-optimization/83424
2697 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2699 2017-12-18 Marek Polacek <polacek@redhat.com>
2702 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2703 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2706 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2708 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2709 (umaddsidi4, umaddsidi_split): Likewise.
2711 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2713 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2716 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2719 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2721 2017-12-18 Richard Biener <rguenther@suse.de>
2723 PR tree-optimization/81877
2724 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2725 (outermost_indep_loop): Adjust.
2726 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2727 (can_sm_ref_p): Adjust.
2729 2017-12-18 Richard Biener <rguenther@suse.de>
2732 * tree.c (array_at_struct_end_p): Return true if the underlying
2733 object has space for at least one element in excess of what
2734 the array domain specifies.
2736 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2738 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2739 example to avoid overfull hbox.
2740 * doc/invoke.texi (Option Summary): Add missing @gol.
2741 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2744 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2745 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2747 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2748 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2749 (AArch64 Options): Document -mverbose-cost-dump.
2750 (ARM Options): Likewise, plus -mflip-thumb.
2752 2017-12-17 Martin Sebor <msebor@redhat.com>
2755 * gimple-ssa-warn-restrict.c
2756 (builtin_memref::offset_out_of_bounds): Correct the handling of
2759 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2761 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2763 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2765 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2767 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2769 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2772 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2774 * doc/invoke.texi: Fix some typos.
2776 2017-12-16 Martin Sebor <msebor@redhat.com>
2778 PR tree-optimization/78918
2779 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2780 * builtins.c (check_sizes): Rename...
2781 (check_access): ...to this. Rename function arguments for clarity.
2782 (check_memop_sizes): Adjust names.
2783 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2784 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2785 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2786 (check_strncat_sizes, expand_builtin_strncat): Same.
2787 (expand_builtin_strncpy, expand_builtin_memset): Same.
2788 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2789 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
2790 (maybe_emit_sprintf_chk_warning): Same.
2791 (expand_builtin_strcpy): Adjust.
2792 (expand_builtin_stpcpy): Same.
2793 (expand_builtin_with_bounds): Detect out-of-bounds accesses
2794 in pointer-checking forms of memcpy, memmove, and mempcpy.
2795 (gcall_to_tree_minimal, max_object_size): Define new functions.
2796 * builtins.h (max_object_size): Declare.
2797 * calls.c (alloc_max_size): Call max_object_size instead of
2798 hardcoding ssizetype limit.
2799 (get_size_range): Handle new argument.
2800 * calls.h (get_size_range): Add a new argument.
2801 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
2802 * doc/invoke.texi (-Wrestrict): Adjust, add example.
2803 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
2805 (gimple_fold_builtin_memory_chk): Same.
2806 (gimple_fold_builtin_stxcpy_chk): New function.
2807 * gimple-ssa-warn-restrict.c: New source.
2808 * gimple-ssa-warn-restrict.h: New header.
2809 * gimple.c (gimple_build_call_from_tree): Propagate location.
2810 * passes.def (pass_warn_restrict): Add new pass.
2811 * tree-pass.h (make_pass_warn_restrict): Declare.
2812 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2814 (handle_builtin_strcat): Same.
2815 (strlen_optimize_stmt): Rename...
2816 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2817 stpncpy, strncpy, and their checking forms.
2819 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2821 PR rtl-optimization/82849
2822 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2823 and get_max_loop_iterations_int.
2825 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2826 Alan Hayward <alan.hayward@arm.com>
2827 David Sherwood <david.sherwood@arm.com>
2829 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2830 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2831 size as a poly_uint64.
2832 (mode_for_vector, mode_for_int_vector): Take the number of vector
2833 elements as a poly_uint64.
2834 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2835 size as a poly_uint64.
2836 (mode_for_vector, mode_for_int_vector): Take the number of vector
2837 elements as a poly_uint64.
2839 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2840 Alan Hayward <alan.hayward@arm.com>
2841 David Sherwood <david.sherwood@arm.com>
2843 * machmode.h (MACRO_MODE): New macro.
2844 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2845 * targhooks.c (default_libcall_value, default_secondary_reload)
2846 (default_memory_move_cost, default_register_move_cost)
2847 (default_class_max_nregs): Likewise.
2849 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2850 Alan Hayward <alan.hayward@arm.com>
2851 David Sherwood <david.sherwood@arm.com>
2853 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2854 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
2856 * genautomata.c (main): Likewise.
2857 * genconditions.c (write_header): Likewise.
2858 * genemit.c (main): Likewise.
2859 * genextract.c (print_header): Likewise.
2860 * genopinit.c (main): Likewise.
2861 * genoutput.c (output_prologue): Likewise.
2862 * genpeep.c (main): Likewise.
2863 * genpreds.c (write_insn_preds_c): Likewise.
2864 * genrecog.c (writer_header): Likewise.
2865 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
2866 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
2867 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2868 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2869 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2870 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2871 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2872 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2873 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2874 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2875 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2876 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2877 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2878 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2879 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2880 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2881 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2882 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2883 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2884 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2885 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2886 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2887 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2888 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2889 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2890 * config/darwin.c (IN_TARGET_CODE): Likewise.
2891 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2892 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2893 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2894 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2895 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2896 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2897 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2898 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2899 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2900 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2901 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2902 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2903 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2904 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2905 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2906 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2907 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2908 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2909 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2910 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2911 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2912 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2913 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2914 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2915 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2916 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2917 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2918 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2919 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2920 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2921 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2922 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2923 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2924 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2925 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2926 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2927 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2928 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2929 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2930 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2931 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2932 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2933 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2934 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2935 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2936 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2937 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2938 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2939 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2940 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2941 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2942 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2943 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2944 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2945 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2946 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2947 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2948 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2949 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2950 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2951 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2952 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2953 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2954 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2955 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2956 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2957 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2958 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2959 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2960 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2961 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2962 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2963 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2964 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2965 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2966 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2967 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2968 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2969 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2970 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2971 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2972 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2973 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2974 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2975 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2976 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2977 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2978 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2979 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2980 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2981 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2982 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2983 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2984 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2985 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2986 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2987 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2988 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2989 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2990 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2991 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2992 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2993 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2994 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2995 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2996 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2998 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2999 Alan Hayward <alan.hayward@arm.com>
3000 David Sherwood <david.sherwood@arm.com>
3002 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3005 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3006 Alan Hayward <alan.hayward@arm.com>
3007 David Sherwood <david.sherwood@arm.com>
3009 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3010 * doc/md.texi (vec_series@var{m}): Document.
3011 * tree.def (VEC_SERIES_EXPR): New tree code.
3012 * tree.h (build_vec_series): Declare.
3013 * tree.c (build_vec_series): New function.
3014 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3015 * tree-pretty-print.c (dump_generic_node): Likewise.
3016 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3017 * tree-inline.c (estimate_operator_cost): Likewise.
3018 * expr.c (expand_expr_real_2): Likewise.
3019 * optabs-tree.c (optab_for_tree_code): Likewise.
3020 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3021 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3022 * expmed.c (make_tree): Handle VEC_SERIES.
3023 * optabs.def (vec_series_optab): New optab.
3024 * optabs.h (expand_vec_series_expr): Declare.
3025 * optabs.c (expand_vec_series_expr): New function.
3026 * tree-vect-generic.c (expand_vector_operations_1): Check that
3027 the operands also have vector type.
3029 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3030 Alan Hayward <alan.hawyard@arm.com>
3031 David Sherwood <david.sherwood@arm.com>
3033 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3034 (VEC_COND_EXPR): Add missing @tindex.
3035 * doc/md.texi (vec_duplicate@var{m}): Document.
3036 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3037 * tree.c (build_vector_from_val): Add stubbed-out handling of
3038 variable-length vectors, using VEC_DUPLICATE_EXPR.
3039 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3040 * cfgexpand.c (expand_debug_expr): Likewise.
3041 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3042 * tree-inline.c (estimate_operator_cost): Likewise.
3043 * tree-pretty-print.c (dump_generic_node): Likewise.
3044 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3045 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3046 (test_vec_duplicate_folding): New function.
3047 (fold_const_c_tests): Call it.
3048 * optabs.def (vec_duplicate_optab): New optab.
3049 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3050 * optabs.h (expand_vector_broadcast): Declare.
3051 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3052 vec_duplicate_optab.
3053 * expr.c (store_constructor): Try using vec_duplicate_optab for
3055 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3057 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3060 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3061 div/mod latencies a bit.
3063 2017-12-15 Jeff Law <law@redhat.com>
3065 PR tree-optimization/36550
3066 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3067 (mark_threaded_blocks): Rewrite code to avoid block copying when
3068 optimizing for size. Don't pessimize blocks which will be
3069 copied, but all the statements will be dead.
3071 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3073 PR tree-optimization/81165
3074 * tree-ssa-threadupdate.c (uses_in_bb): New.
3075 (estimate_threading_killed_stmts): New.
3076 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3077 * tree-ssa-threadedge.c
3078 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3081 2017-12-15 Jeff Law <law@redhat.com>
3083 PR tree-optimization/83410
3084 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3085 threads when parallelizing loops.
3087 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3089 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3091 * config/alpha/alpha.c (vms_attribute_table): Swap
3092 affects_type_identity and handler fields, adjust comments.
3093 * config/mips/mips.c (mips_attribute_table): Likewise.
3094 * config/visium/visium.c (visium_attribute_table): Likewise.
3095 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3096 * config/microblaze/microblaze.c (microblaze_attribute_table):
3098 * config/spu/spu.c (spu_attribute_table): Likewise.
3099 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3100 * config/arc/arc.c (arc_attribute_table): Likewise.
3101 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3102 * config/v850/v850.c (v850_handle_interrupt_attribute,
3103 v850_handle_data_area_attribute): Formatting fixes.
3104 (v850_attribute_table): Swap affects_type_identity and handler
3105 fields, adjust comments.
3106 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3107 * config/arm/arm.c (arm_attribute_table): Likewise.
3108 * config/avr/avr.c (avr_attribute_table): Likewise.
3109 * config/s390/s390.c (s390_attribute_table): Likewise.
3110 * config/sh/sh.c (sh_attribute_table): Likewise.
3111 * config/i386/i386.c (ix86_handle_cconv_attribute,
3112 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3113 (ix86_attribute_table): Swap affects_type_identity and handler
3114 fields, adjust comments.
3115 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3116 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3117 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3118 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3119 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3120 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3121 * config/rx/rx.c (rx_attribute_table): Likewise.
3122 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3123 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3124 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3125 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3126 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3127 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3128 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3129 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3130 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3131 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3132 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3134 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3136 (register_scoped_attributes, decl_attributes): Formatting fixes.
3138 PR tree-optimization/83269
3139 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3140 subtraction in arg0's type if type is signed and arg0 is unsigned.
3144 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3145 convert? on @0 instead of convert. Check type of @1, not @0.
3146 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3147 convert? on @0 instead of convert. Check type of @1, not @0.
3148 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3149 only optimize if either both @1 and @2 types are narrower
3150 precision, or both are wider or equal precision, and in the former
3151 case only if both have undefined overflow.
3153 2017-12-15 Richard Biener <rguenther@suse.de>
3156 * internal-fn.def (IFN_NOP): Add.
3157 * internal-fn.c (expand_NOP): Do nothing.
3158 * lto-streamer-in.c (input_function): Instead of removing
3159 sanitizer calls replace them with IFN_NOP calls.
3161 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3162 Alan Hayward <alan.hayward@arm.com>
3163 David Sherwood <david.sherwood@arm.com>
3165 * dse.c (store_info, read_info_type): Replace begin and end with
3167 (print_range): New function.
3168 (set_all_positions_unneeded, any_positions_needed_p)
3169 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3171 (record_store): Likewise. Optimize the case in which all positions
3173 (get_stored_val): Replace read_begin and read_end with read_offset
3175 (replace_read): Update call accordingly.
3177 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3179 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3180 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3181 (loop_cand::loop_cand): Initialize above members.
3182 (loop_cand::supported_operations): Delete.
3183 (loop_cand::can_interchange_p): Inline above function.
3184 (loop_cand::classify_simple_reduction): Record number of constant
3185 initialized simple reductions.
3186 (should_interchange_loops): New parameters. Check stmt cost of loops
3188 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3189 Update call to should_interchange_loops.
3190 (should_interchange_loop_nest): Update call to
3191 should_interchange_loops.
3193 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3196 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3197 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3198 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3200 2017-12-15 Julia Koval <julia.koval@intel.com>
3202 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3203 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3205 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3206 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3207 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3209 2017-12-15 Julia Koval <julia.koval@intel.com>
3211 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3212 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3213 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3214 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3215 _mm_aesenc_epi128): New intrinsics.
3217 2017-12-15 Julia Koval <julia.koval@intel.com>
3219 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3220 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3222 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3223 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3224 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3226 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3228 * gimple-ssa-strength-reduction.c (analyze_increments):
3229 Distinguish replacement costs for constant strides from those for
3232 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3234 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3236 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3238 * doc/invoke.texi: Document -Wcast-function-type.
3239 * recog.h (stored_funcptr): Change signature.
3240 * tree-dump.c (dump_node): Avoid warning.
3241 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3243 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3246 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3249 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3251 PR tree-optimization/83312
3252 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3253 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3254 "val" param, and to cope with arbitrary basic blocks.
3255 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3256 handle NULL_TREE for "val", dropping "bb" param.
3257 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3258 drop "bb" param. Handle NULL_TREE for "val".
3259 (find_case_label_for_value): Make "switch_stmt" param const.
3260 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3262 (vrp_prop::check_all_array_refs): Reimplement as...
3263 (check_array_bounds_dom_walker::before_dom_children): ...this new
3264 vfunc. Replace linear search through BB block list, excluding
3265 those with non-executable in-edges via dominator walk.
3267 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3269 * config/arm/arm.opt (mverbose-cost-dump): New option.
3270 * config/arm/arm.c (arm_rtx_costs): Use it.
3272 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3275 * reload1.c (emit_input_reload_insns): Skip debug markers.
3277 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3279 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3283 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3284 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3285 * config/sh/sh.c: Skip debug insns besides notes.
3286 * config/sh/sh.md: Likewise.
3287 * config/sh/sh_treg_combine.cc: Likewise.
3288 * config/sh/sync.md: Likewise.
3290 2017-12-14 Tom de Vries <tom@codesourcery.com>
3292 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3293 Add item for weak_undefined.
3295 2017-12-14 Richard Biener <rguenther@suse.de>
3297 PR tree-optimization/67842
3298 * sese.h (bb_in_region): Remove #if 0'ed code.
3300 2017-12-14 Richard Biener <rguenther@suse.de>
3302 PR tree-optimization/83326
3303 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3304 may_be_zero parameter and handle it by not marking the first
3305 peeled copy as not exiting the loop.
3306 (try_peel_loop): Likewise.
3307 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3308 to handle the case of constant or zero iterations and perform
3309 loop header copying on-the-fly.
3311 2017-12-14 Richard Biener <rguenther@suse.de>
3313 PR tree-optimization/83418
3314 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3315 Instead of asserting we don't get unfolded comparisons deal with them.
3317 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3320 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3321 basic blocks. Assert debug bind insns don't appear outside of bbs,
3322 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3325 PR tree-optimization/83198
3326 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3327 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3328 value if arg is a REAL_CST with incompatible type.
3330 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3331 Bin Cheng <bin.cheng@arm.com>
3334 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3336 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3339 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3341 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3342 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3344 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3345 Alan Hayward <alan.hayward@arm.com>
3346 David Sherwood <david.sherwood@arm.com>
3348 * poly-int.h: New file.
3349 * poly-int-types.h: Likewise.
3350 * coretypes.h: Include them.
3351 (POLY_INT_CONVERSION): Define.
3352 * target.def (estimated_poly_value): New hook.
3353 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3354 * doc/tm.texi: Regenerate.
3355 * doc/poly-int.texi: New file.
3356 * doc/gccint.texi: Include it.
3357 * doc/rtl.texi: Describe restrictions on subreg modes.
3358 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3359 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3360 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3361 * targhooks.h (default_estimated_poly_value): Declare.
3362 * targhooks.c (default_estimated_poly_value): New function.
3363 * target.h (estimated_poly_value): Likewise.
3364 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3365 (wi::unary_traits): Delete.
3366 (wi::binary_traits::signed_shift_result_type): Define for
3367 offset_int << HOST_WIDE_INT, etc.
3368 (generic_wide_int::operator <<=): Define for all types and use
3369 wi::lshift instead of <<.
3370 (wi::hwi_with_prec): Add a default constructor.
3371 (wi::ints_for): New class.
3372 (operator <<): Define for all wide-int types.
3373 (operator /): New function.
3374 (operator %): Likewise.
3375 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3376 (ASSERT_MAYBE_NE_AT): New macros.
3378 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3379 Dominik Vogt <vogt@linux.vnet.ibm.com>
3382 * emit-rtl.c (init_emit): Remove ??? comment.
3383 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3384 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3385 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3386 alignment of 3 virtual registers to BITS_PER_WORD.
3388 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3390 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3392 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3393 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3395 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3398 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3400 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3402 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3403 with the exception of debug begin stmt markers appear before
3407 * final.c (rest_of_handle_final): Call variable_tracking_main only
3408 if !flag_var_tracking.
3410 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3411 Jakub Jelinek <jakub@redhat.com>
3415 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3416 labels debug stmts that can only appear after labels.
3418 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3420 PR rtl-optimization/82398
3421 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3422 EXPR_USEFULNESS in priority comparison.
3424 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3426 PR rtl-optimization/83393
3427 * combine.c (move_deaths): If reg_stat points to a too new insn in
3428 last_death, do not use it: find the proper insn instead.
3430 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3433 * config/rs6000/rs6000-string.c (expand_block_compare,
3434 expand_strn_compare): Fix set_mem_size() calls.
3436 2017-12-12 Jeff Law <law@redhat.com>
3438 PR tree-optimization/83298
3439 PR tree-optimization/83362
3440 PR tree-optimization/83383
3441 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3442 push_value_range a public interface. Add new argument to
3443 record_ranges_from_stmt.
3444 * gimple-ssa-evrp-analyze.c
3445 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3446 Update comments. Handle recording temporary equivalences.
3447 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3448 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3449 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3450 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3451 gimple-ssa-evrp-analyze.h.
3452 (record_temporary_equivalences_from_phis): Add new argument. When
3453 the PHI arg is an SSA_NAME, set the result's range to the range
3455 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3456 from statements too.
3457 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3458 Pass it down to children as needed.
3459 (thread_outgoing_edges): Likewise.
3460 (thread_across_edge): Likewise. Push/pop range state as needed.
3461 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3463 2017-12-12 Julia Koval <julia.koval@intel.com>
3465 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3466 (PTA_CANNONLAKE): Remove PTA_CLWB.
3468 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3471 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3472 (vcondv2div2df): Likewise.
3473 (vconduv2dfv2di): Likewise.
3475 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3477 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3478 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3479 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3480 ATTR_MATHFN_FPROUNDING_ERRNO.
3482 2017-12-12 Richard Biener <rguenther@suse.de>
3484 PR tree-optimization/83385
3485 * graphite-scop-detection.c (get_order, order): Remove.
3486 (bb_to_rpo): New global.
3488 (build_scops): Sort pbbs in RPO order.
3490 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3492 * combine.c (simplify_set): Do not transform subregs to zero_extends
3493 if the destination mode is a vector mode.
3495 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3497 PR tree-optimization/83359
3498 * tree-cfg.h (fold_loop_internal_call): Declare.
3499 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3500 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3501 (find_loop_dist_alias): New function.
3502 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3503 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3506 PR tree-optimization/80631
3507 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3508 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3509 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3510 hardcoding zero as the value if COND_EXPR is never true. For
3511 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3512 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3513 hardcoding MAX_EXPR as the reduction operation.
3514 (is_nonwrapping_integer_induction): Allow negative step.
3515 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3516 vect_create_epilog_for_reduction, if no value is suitable, don't
3517 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3519 2017-12-12 Richard Biener <rguenther@suse.de>
3521 PR tree-optimization/81889
3522 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3523 range info from the non-wrapping IV instead of just the range
3526 2017-12-12 Julia Koval <julia.koval@intel.com>
3528 * config.gcc: Add vaesintrin.h.
3529 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3531 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3532 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3534 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3535 * config/i386/immintrin.h: Include vaesintrin.h.
3536 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3537 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3538 _mm_aesdec_epi128): New intrinsics.
3540 2017-12-12 Julia Koval <julia.koval@intel.com>
3542 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3543 OPTION_MASK_ISA_VAES_UNSET): New.
3544 (ix86_handle_option): Handle -mvaes.
3545 * config/i386/cpuid.h: Define bit_VAES.
3546 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3547 * config/i386/i386-c.c (__VAES__): New.
3548 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3549 (ix86_valid_target_attribute_inner_p): Ditto.
3550 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3551 * config/i386/i386.opt: Add -mvaes.
3552 * doc/invoke.texi: Ditto.
3554 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3556 * debug.h (gcc_debug_hooks): Add inline_entry.
3557 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3558 * debug.c (do_nothing_debug_hooks): Likewise.
3559 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3560 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3561 (dwarf2_lineno_debug_hooks): Likewise.
3563 * common.opt (gstatement-frontiers): New, setting
3564 debug_nonbind_markers_p.
3565 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3566 * toplev.c (process_options): Autodetect value for debug statement
3568 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3569 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3571 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3572 markers. Integrate source bind into debug stmt expand loop.
3573 (pass_expand::execute): Check debug marker limit. Avoid deep
3574 TER and expand debug locations for debug bind insns only.
3575 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3577 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3578 * final.c (reemit_insn_block_notes): Take current block from
3579 nonbind markers. Declare note where it's first set.
3580 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3581 begin stmt markers if enabled.
3582 (notice_source_line): Handle nonbind markers. Fail if their
3583 location is unknown or that of builtins.
3584 (rest_of_handle_final): Convert begin stmt markers to notes if
3585 var-tracking didn't run.
3586 (rest_of_clean_state): Skip begin stmt markers.
3587 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3589 * function.c (allocate_struct_function): Set begin_stmt_markers.
3590 * function.h (struct function): Add debug_marker_count counter
3591 and debug_nonbind_markers flag.
3592 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3593 * gimple-low.c (lower_function_body): Adjust
3594 debug_nonbind_markers.
3595 (lower_stmt): Drop or skip gimple debug stmts.
3596 (lower_try_catch): Skip debug stmts.
3597 * gimple.c (gimple_build_debug_begin_stmt): New.
3598 (gimple_copy): Increment debug_marker_count if copying one.
3599 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3600 * gimplify.c (rexpr_location): New.
3601 (rexpr_has_location): New.
3602 (warn_switch_unreachable_r): Handle gimple debug stmts.
3603 (shortcut_cond_r): Call expr_location.
3605 (find_goto_label): New.
3606 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3608 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3610 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3611 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3612 (LANG_HOOKS_INITIALIZER): ... this.
3613 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3614 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3616 (update_ebb_live_info): Skip debug insn markers.
3617 * lra.c (debug_insn_static_data): Rename to...
3618 (debug_bind_static_data): ... this.
3619 (debug_marker_static_data): New.
3620 (lra_set_insn_recog_data): Select one of the above depending
3622 (lra_update_isn_regno_info): Don't assume debug insns have
3624 (push_insns): Skip debug insns.
3625 * lto-streamer-in.c (input_function): Drop debug stmts
3626 depending on active options. Adjust debug_nonbind_markers.
3627 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3628 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3629 begin stmt marker notes.
3630 (print_insn): Likewise.
3631 * recog.c (extract_insn): Recognize rtl for debug markers.
3632 * rtl.def (DEBUG_MARKER): New.
3633 * tree-inline.c: Include params.h.
3634 (remap_gimple_stmt): Handle nonbind markers.
3635 (maybe_move_debug_stmts_to_successors): Likewise.
3636 (copy_debug_stmt): Likewise.
3637 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3638 markers regardless of no side effects.
3639 (tsi_link_before): Don't update container's side effects when adding
3640 a begin stmt marker.
3641 (tsi_link_after): Likewise.
3642 (expr_first): Skip begin stmt markers.
3643 (expr_last): Likewise.
3644 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3645 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3646 Disregard nonbind markers.
3647 * tree.c (make_node_stat): Don't set side effects for begin stmt
3649 (build1_stat): Likewise.
3650 * tree.def (DEBUG_BEGIN_STMT): New.
3651 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3652 * var-tracking.c (delete_debug_insns): Renamed to...
3653 (delete_vta_debug_insns): ... this.
3654 (reemit_marker_as_note): New.
3655 (vt_initialize): Reemit markers.
3656 (delete_vta_debug_insns): Likewise.
3657 (vt_debug_insns_local): Reemit or delete markers.
3658 (variable_tracking_main_1): Likewise.
3659 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3660 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3661 (gimple_debug_nonbind_marker_p): New.
3662 (gimple_build_debug_bind): Adjust.
3663 (gimple_build_debug_begin_stmt): New.
3664 * doc/invoke.texi (max-debug-marker-count): New param.
3665 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3666 (debug_parameter_ref, debug_marker): New.
3667 (NOTE_INSN_BEGIN_STMT): New.
3668 (DEBUG_INSN): Describe begin stmt markers.
3670 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3671 without debug insns, we wouldn't, but clean up debug insns
3672 after a control flow insn nevertheless.
3674 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3675 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3676 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3677 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3678 (rtl_tidy_fallthru_edge): Likewise.
3679 (rtl_verify_fallthru): Likewise.
3680 (rtl_verify_bb_layout): Likewise.
3681 (skip_insns_after_block): Likewise.
3682 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3683 * dwarf2out.c: Include print-rtl.h.
3684 (dwarf2out_next_real_insn): New.
3685 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3686 Dump debug binds in asm comments.
3687 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3688 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3689 callers to use gsi_start_nondebug_bb instead.
3690 (gsi_after_labels): Skip gimple debug stmts.
3691 (gsi_start_nondebug): New.
3692 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3693 (proper_loop_form_for_interchange): Adjust.
3694 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3695 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3696 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3697 (collect_fallthrough_labels): Likewise.
3698 (should_warn_for_implicit_fallthrough): Likewise.
3699 (warn_implicit_fallthrough_r): Likewise.
3700 (expand_FALLTHROUGH_r): Likewise.
3701 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3702 (graphite_copy_stmts_from_block): Skip nonbind markers.
3703 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3704 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3705 * jump.c (clean_barriers): Skip debug insns.
3706 * omp-expand.c (expand_parallel_call): Skip debug insns.
3707 (expand_task_call): Likewise.
3708 (remove_exit_barrier): Likewise.
3709 (expand_omp_taskreg): Likewise.
3710 (expand_omp_for_init_counts): Likewise.
3711 (expand_omp_for_generic): Likewise.
3712 (expand_omp_for_static_nochunk): Likewise.
3713 (expand_omp_for_static_chunk): Likewise.
3714 (expand_omp_simd): Likewise.
3715 (expand_omp_taskloop_for_outer): Likewise.
3716 (expand_omp_taskloop_for_inner): Likewise.
3717 (expand_oacc_for): Likewise.
3718 (expand_omp_sections): Likewise.
3719 (expand_omp_single): Likewise.
3720 (expand_omp_synch): Likewise.
3721 (expand_omp_atomic_load): Likewise.
3722 (expand_omp_atomic_store): Likewise.
3723 (expand_omp_atomic_fetch_op): Likewise.
3724 (expand_omp_atomic_pipeline): Likewise.
3725 (expand_omp_atomic_mutex): Likewise.
3726 (expand_omp_target): Likewise.
3727 (grid_expand_omp_for_loop): Likewise.
3728 (grid_expand_target_grid_body): Likewise.
3729 (build_omp_regions_1): Likewise.
3730 * omp-low.c (check_combined_parallel): Skip debug stmts.
3731 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3732 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3733 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3735 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3736 (make_edges): Likewise.
3737 (cleanup_dead_labels): Likewise.
3738 (gimple_can_merge_blocks_p): Likewise.
3739 (stmt_starts_bb_p): Likewise.
3740 (gimple_block_label): Likewise.
3741 (gimple_redirect_edge_and_branch): Likewise.
3742 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3744 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3746 * tree-pretty-print (print_declaration): Omit initializer in slim
3748 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3750 (eliminate_unnecessary_stmts): Stabilize block removal order.
3751 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3752 * var-tracking.c (get_first_insn): New.
3753 (vt_emit_notes): Call it.
3754 (vt_initialize): Walk any insns before the first BB.
3755 (delete_debug_insns): Likewise.
3757 * gimple.h (enum gimple_debug_subcode): Add
3758 GIMPLE_DEBUG_BEGIN_STMT.
3759 (gimple_debug_begin_stmt_p): New.
3760 (gimple_debug_nonbind_marker_p): New.
3761 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3762 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3763 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3764 * insn-notes.def (BEGIN_STMT): New.
3765 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3766 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3767 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3768 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3769 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3770 (INSN_DEBUG_MARKER_KIND): New.
3771 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3772 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3773 (INSN_VAR_LOCATION_PTR): New.
3774 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3776 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3777 deps only if debug bind insns are enabled.
3778 (pass_expand::execute): Avoid deep TER and expand
3779 debug locations for debug bind insns only.
3780 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3781 debug stmts special handling down to debug bind stmts.
3782 * combine.c (try_combine): Narrow debug insns special handling
3783 down to debug bind insns.
3784 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3785 Narrow debug insns preexisting special handling down to debug
3787 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3788 handling down to debug bind insns.
3789 * function.c (instantiate_virtual_regs): Skip debug markers,
3790 adjust handling of debug binds.
3791 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
3792 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
3793 * haifa-sched.c (schedule_insn): Narrow special handling of debug
3794 insns to debug bind insns.
3795 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
3796 special handling of debug stmts to debug bind stmts.
3797 * ipa-split.c (split_function): Likewise.
3798 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
3799 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
3801 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
3802 * regrename.c (build_def_use): Likewise.
3803 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
3804 (pass_cprop_hardreg): Narrow special casing of debug insns to
3806 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
3807 * reload1.c (reload): Likewise.
3808 * sese.c (sese_insert_phis_for_liveouts): Narrow special
3809 casing of debug stmts to debug bind stmts.
3810 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
3811 * ssa-iterators.h (num_imm_uses): Likewise.
3812 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
3813 debug stmts to debug bind stmts.
3814 * tree-inline.c (tree_function_versioning): Narrow special casing
3815 of debug stmts to debug bind stmts.
3816 * tree-loop-distribution.c (generate_loops_for_partition):
3817 Narrow special casing of debug stmts to debug bind stmts.
3818 * tree-sra.c (analyze_access_subtree): Narrow special casing
3819 of debug stmts to debug bind stmts.
3820 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3821 stmts to debug bind stmts.
3822 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3823 casing of debug stmts to debug bind stmts.
3824 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3825 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3826 casing of debug stmts to debug bind stmts.
3827 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3829 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3830 debug stmts to debug bind stmts.
3831 (gimple_replace_ssa_lhs): Likewise.
3832 (insert_debug_temp_for_var_def): Likewise.
3833 (insert_debug_temps_for_defs): Likewise.
3834 (reset_debug_uses): Likewise.
3835 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3836 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3837 (adjust_debug_stmts): Likewise.
3838 (adjust_phi_and_debug_stmts): Likewise.
3839 (vect_do_peeling): Likewise.
3840 * tree-vect-loop.c (vect_transform_loop): Likewise.
3841 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3842 * var-tracking.c (adjust_mems): Narrow special casing of debug
3843 insns to debug bind insns.
3844 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
3845 (compute_bb_dataflow, vt_find_locations): Likewise.
3846 (vt_expand_loc, emit_notes_for_changes): Likewise.
3847 (vt_init_cfa_base): Likewise.
3848 (vt_emit_notes): Likewise.
3849 (vt_initialize): Likewise.
3850 (vt_finalize): Likewise.
3852 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3853 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3854 (next_nonnote_nondebug_insn_bb): New.
3855 (prev_nonnote_nondebug_insn_bb): New.
3856 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
3857 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
3858 (prev_nonnote_nondebug_insn_bb): Declare.
3859 (next_nonnote_nondebug_insn_bb): Declare.
3860 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
3861 * cfgrtl.c (get_last_bb_insn): Likewise.
3862 * lra.c (push_insns): Likewise.
3864 2017-12-11 David Malcolm <dmalcolm@redhat.com>
3867 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
3869 * selftest.c (selftest::test_runner::test_runner): New ctor.
3870 (selftest::test_runner::~test_runner): New dtor.
3871 * selftest.h (class selftest::test_runner): New class.
3873 2017-12-11 Carl Love <cel@us.ibm.com>
3875 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3876 vec_extract_fp32_from_shortl]): Add #defines.
3877 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3878 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3879 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3880 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3881 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3882 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3883 * doc/extend.texi: Add documentation for the added builtins.
3885 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3887 PR rtl-optimization/80693
3888 PR rtl-optimization/81019
3889 PR rtl-optimization/81020
3890 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3891 are not mentioned in i3. Place the REG_UNUSED note on i2,
3892 possibly modified to REG_DEAD, if it did not originate in i3.
3894 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3896 * recog.c (store_data_bypass_p_1): New function.
3897 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3898 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3900 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3902 PR rtl-optimization/83361
3903 * ifcvt.c (if_convert): Call fixup_partitions.
3905 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3907 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3908 early folding of splat_u{8,16,32}.
3910 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3912 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3913 output_operand_lossage first argument with capital letter.
3914 (aarch64_override_options): Don't start error and sorry first argument
3915 with capital letter.
3917 2017-12-11 Andi Kleen <ak@linux.intel.com>
3919 PR gcov-profile/83355
3920 * auto-profile.c (string_table::get_index_by_decl): Don't
3921 recurse when abstract origin points to itself.
3923 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3925 PR tree-optimization/83320
3926 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3927 (prune_datarefs_not_in_loop): Ditto.
3929 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3931 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3933 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3935 PR tree-optimization/83337
3936 * gimple-loop-interchange.cc (compute_access_stride): Handle
3937 bitfield DRs properly.
3939 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3941 PR tree-optimization/83338
3942 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3943 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3944 vector of unsigned integers to vector of signed integers.
3946 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3948 PR rtl-optimization/83317
3949 * lra-constraints.c (process_address_1): Add insn code check.
3951 2017-12-08 Michael Matz <matz@suse.de>
3953 Fix PR tree-optimization/83323
3954 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3955 head-controlled loops and loop BBs.
3956 * common.opt (funroll-and-jam): Remove, instead ...
3957 (floop-unroll-and-jam): ... reuse this option.
3958 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3959 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3960 (-floop-unroll-and-jam): ... this option.
3962 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3964 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3965 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3966 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3968 2017-12-08 Richard Biener <rguenther@suse.de>
3971 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3972 handle accesses outside of zero-sized vars.
3974 2017-12-08 Martin Jambor <mjambor@suse.cz>
3976 PR tree-optimization/83141
3977 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3978 test for MEM_REFs implicitely changing types with padding. Remove
3980 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3982 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3984 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3986 * config/avr/avr.c (avr_attribute_table): Likewise.
3987 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3988 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3989 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3990 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3991 of struct attribute_spec.
3992 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3994 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3997 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3999 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4002 * profile-count.c (profile_count::from_gcov_type): Move from
4003 profile-count.h; handle overflow.
4004 * profile-count.h (profile_count::from_gcov_type): Move offline.
4006 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4008 PR rtl-optimization/83304
4009 * combine.c (move_deaths): If we do not know where a register died,
4012 2017-12-08 Richard Biener <rguenther@suse.de>
4014 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4015 Provide -fopt-info-loop feedback when we interchange in a nest.
4017 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4019 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4020 for armv6 ARM CPU IDs.
4022 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4024 * common/config/arm/arm-common.c: Include <algorithm>.
4025 (INCLUDE_VECTOR): Define.
4026 (compare_opt_names): New function.
4027 (arm_rewrite_selected_arch): Only strip out extensions that can be
4028 expressed through -mfpu. Sort the remaining extensions
4031 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4033 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4034 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4035 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4036 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4038 2017-06-08 Tristan Gingold <gindold@adacore.com>
4041 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4042 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4044 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4046 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4047 of struct attribute_spec.
4049 2017-12-08 Julia Koval <julia.koval@intel.com>
4051 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4052 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4053 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4054 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4055 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4056 _mm_maskz_dpwssds_epi32): Ditto.
4058 2017-12-08 Richard Biener <rguenther@suse.de>
4060 PR tree-optimization/81303
4061 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4062 conditions try to create a comparison vector type matching
4063 the data vector type.
4064 (vectorizable_condition): Adjust.
4065 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4066 Leave invariant conditions alone in case we can vectorize those.
4068 2017-12-08 Julia Koval <julia.koval@intel.com>
4070 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4071 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4072 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4073 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4074 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4076 2017-12-08 Julia Koval <julia.koval@intel.com>
4078 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4079 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4080 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4081 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4082 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4083 _mm_maskz_dpbusds_epi32): New intrinsics.
4085 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4087 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4089 (ld<bh>io_signed): Likewise.
4090 (st<bhw>io): Likewise.
4091 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4092 SMALL_INT12 constant integer operand.
4094 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4096 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4098 (nios2_large_constant_allowed): ...this. Adjust uses.
4099 (nios2_plus_symbolic_constant_p): Rename to...
4100 (nios2_plus_large_constant_p): ...this. Adjust uses.
4101 (nios2_legitimate_address_p): Correct CONST_INT handling.
4102 (nios2_symbolic_memory_operand_p): Rename to...
4103 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4104 (nios2_large_constant_p): Check for large constant integers too.
4105 (nios2_split_large_constant): Handle constant integers.
4106 (nios2_split_symbolic_memory_operand): Rename to...
4107 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4108 (nios2_legitimize_constant_address): Handle constant integers.
4109 (r0rel_constant_p): Handle small constant integers.
4110 (nios2_print_operand_address): Handle r0-relative integer addresses.
4111 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4112 * config/nios2/nios2.md: Adjust for renamed functions.
4114 2017-12-07 Andrew Waterman <andrew@sifive.com>
4116 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4117 (TARGET_HAVE_SRODATA_SECTION): New define.
4118 (riscv_select_section): New function.
4120 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4121 Alexander Monakov <amonakov@ispras.ru>
4122 Jakub Jelinek <jakub@redhat.com>
4125 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4127 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4129 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4130 a bool success value. Don't call output_operand_lossage here.
4131 (aarch64_print_ldpstp_address): Return a bool success value.
4132 (aarch64_print_operand_address): Call output_addr_const if
4133 aarch64_print_address_internal fails.
4134 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4135 'y'; call output_operand_lossage instead. Call output_operand_lossage
4136 if aarch64_print_ldpstp_address fails.
4138 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4140 * tree-vector-builder.h
4141 (tree_vector_builder::binary_encoded_nelts): Declare.
4142 * tree-vector-builder.c
4143 (tree_vector_builder::binary_encoded_nelts): New function.
4144 * fold-const.c (negate_expr_p): Likewise.
4145 (operand_equal_p, fold_checksum_tree): Likewise.
4146 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4147 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4148 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4149 (uniform_vector_p): Likewise.
4150 * varasm.c (const_hash_1, compare_constant): Likewise.
4151 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4152 (valid_lattice_transition): Operate directly on the VECTOR_CST
4154 * ipa-icf.c: Include tree-vector-builder.h.
4155 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4156 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4158 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4160 * tree.c (build_vector): Delete.
4161 * tree.h (build_vector): Make static and move into the self-testing
4164 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4166 * vector-builder.h (vector_builder::derived): New const overload.
4167 (vector_builder::elt): New function.
4168 * tree-vector-builder.h (tree_vector_builder::type): New function.
4169 (tree_vector_builder::apply_step): Declare.
4170 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4172 * gimple-fold.h (tree_vector_builder): Declare.
4173 (gimple_build_vector): Take a tree_vector_builder instead of a
4174 type and vector of elements.
4175 * gimple-fold.c (gimple_build_vector): Likewise.
4176 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4178 (get_initial_defs_for_reduction): Likewise.
4179 (vectorizable_induction): Likewise.
4181 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4183 * tree-vector-builder.h
4184 (tree_vector_builder::new_binary_operation): Declare.
4185 * tree-vector-builder.c
4186 (tree_vector_builder::new_binary_operation): New function.
4187 * fold-const.c (fold_relational_const): Use it.
4188 (const_binop): Likewise. Check that both input vectors have
4189 the same number of elements, thus excluding things like WIDEN_SUM.
4190 Check whether it is possible to operate directly on the encodings
4193 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4195 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4196 new_unary_operation, operating only on the encoded elements.
4197 (const_unop): Likewise.
4198 (exact_inverse): Likewise.
4199 (distributes_over_addition_p): New function.
4200 (const_binop): Use tree_vector_builder and new_unary_operation
4201 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4202 on the encoded elements unless the encoding is strided and the
4203 operation does not distribute over addition.
4204 (fold_convert_const): Use tree_vector_builder and
4205 new_unary_operation. Operate only on the encoded elements
4206 for truncating integer conversions, or for non-stepped encodings.
4208 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4210 * config/sparc/sparc.c: Include tree-vector-builder.h.
4211 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4212 * expmed.c: Include tree-vector-builder.h.
4213 (make_tree): Use tree_vector_builder instead of build_vector.
4214 * fold-const.c: Include tree-vector-builder.h.
4215 (const_binop): Use tree_vector_builder instead of build_vector.
4216 (const_unop): Likewise.
4217 (native_interpret_vector): Likewise.
4218 (fold_vec_perm): Likewise.
4219 (fold_ternary_loc): Likewise.
4220 * gimple-fold.c: Include tree-vector-builder.h.
4221 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4223 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4224 (simplify_vector_constructor): Use tree_vector_builder instead
4226 * tree-vect-generic.c: Include tree-vector-builder.h.
4227 (add_rshift): Use tree_vector_builder instead of build_vector.
4228 (expand_vector_divmod): Likewise.
4229 (optimize_vector_constructor): Likewise.
4230 * tree-vect-loop.c: Include tree-vector-builder.h.
4231 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4232 of build_vector. Explicitly use a stepped encoding for
4234 * tree-vect-slp.c: Include tree-vector-builder.h.
4235 (vect_get_constant_vectors): Use tree_vector_builder instead
4237 (vect_transform_slp_perm_load): Likewise.
4238 (vect_schedule_slp_instance): Likewise.
4239 * tree-vect-stmts.c: Include tree-vector-builder.h.
4240 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4241 (vect_gen_perm_mask_any): Likewise.
4242 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4243 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4245 (build_vector_from_val): Likewise. Explicitly use a duplicate
4248 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4250 * doc/generic.texi (VECTOR_CST): Describe new representation of
4252 * vector-builder.h: New file.
4253 * tree-vector-builder.h: Likewise.
4254 * tree-vector-builder.c: Likewise.
4255 * Makefile.in (OBJS): Add tree-vector-builder.o.
4256 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4257 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4258 (tree_vector): Change the number of elements to
4259 vector_cst_encoded_nelts.
4260 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4261 (VECTOR_CST_ELTS): Delete.
4262 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4263 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4264 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4265 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4266 (VECTOR_CST_ENCODED_ELT): Likewise.
4267 (vector_cst_encoded_nelts): New function.
4268 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4269 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4270 (vector_cst_int_elt, vector_cst_elt): Declare.
4271 * tree.c: Include tree-vector-builder.h.
4272 (tree_code_size): Abort if passed VECTOR_CST.
4273 (tree_size): Update for new VECTOR_CST layout.
4274 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4275 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4276 (build_vector): Use tree_vector_builder.
4277 (vector_cst_int_elt, vector_cst_elt): New functions.
4278 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4279 encoded elements and then create the vector in the canonical form.
4280 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4281 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4282 (tree_c_tests): Call test_vector_cst_patterns.
4283 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4285 (hash_tree): Likewise.
4286 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4287 (streamer_write_tree_header): Likewise.
4288 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4289 (streamer_alloc_tree): Likewise. Update call to make_vector.
4290 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4292 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4294 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4295 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4296 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4298 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4299 Richard Biener <rguenther@suse.de>
4301 PR tree-optimization/81303
4302 * Makefile.in (gimple-loop-interchange.o): New object file.
4303 * common.opt (floop-interchange): Reuse the option from graphite.
4304 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4305 -floop-interchange and mention it for -O3.
4306 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4307 * gimple-loop-interchange.cc: New file.
4308 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4309 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4310 * passes.def (pass_linterchange): New pass.
4311 * timevar.def (TV_LINTERCHANGE): New time var.
4312 * tree-pass.h (make_pass_linterchange): New declaration.
4313 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4314 interchange. Record IV before/after increment in new parameters.
4315 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4316 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4318 (check_reduction_path): ...New function here.
4319 * tree-vectorizer.h (check_reduction_path): New declaration.
4321 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4324 PR rtl-optimization/80818
4325 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4326 always early clobbered.
4327 * lra-lives.c (process_bb_lives): Check input hard regs for early
4328 clobbered non-operand hard reg.
4330 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4333 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4334 types_compatible_p, just that TYPE_MODE is the same.
4336 2017-12-07 Martin Sebor <msebor@redhat.com>
4339 * attribs.c (empty_attribute_table): Initialize new member of
4340 struct attribute_spec.
4341 (decl_attributes): Add argument. Handle mutually exclusive
4342 combinations of attributes.
4343 (selftests::test_attribute_exclusions): New function.
4344 (selftests::attribute_c_tests): Ditto.
4345 * attribs.h (decl_attributes): Add default argument.
4346 * selftest.h (attribute_c_tests): Declare.
4347 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4348 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4350 * doc/extend.texi (Common Function Attributes): Update const and pure.
4351 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4352 of struct attribute_spec.
4353 * config/arc/arc.c (arc_attribute_table): Same.
4354 * config/arm/arm.c (arm_attribute_table): Same.
4355 * config/avr/avr.c ( avr_attribute_table): Same.
4356 * config/bfin/bfin.c (bfin_attribute_table): Same.
4357 * config/cr16/cr16.c (cr16_attribute_table): Same.
4358 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4359 * config/h8300/h8300.c (h8300_attribute_table): Same.
4360 * config/i386/i386.c (ix86_attribute_table): Same.
4361 * config/ia64/ia64.c (ia64_attribute_table): Same.
4362 * config/m32c/m32c.c (m32c_attribute_table): Same.
4363 * config/m32r/m32r.c (m32r_attribute_table): Same.
4364 * config/m68k/m68k.c (m68k_attribute_table): Same.
4365 * config/mcore/mcore.c (mcore_attribute_table): Same.
4366 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4367 * config/mips/mips.c (mips_attribute_table): Same.
4368 * config/msp430/msp430.c (msp430_attribute_table): Same.
4369 * config/nds32/nds32.c (nds32_attribute_table): Same.
4370 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4371 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4372 * config/rl78/rl78.c (rl78__attribute_table): Same.
4373 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4374 * onfig/rx/rx.c (rx_attribute_table): Same.
4375 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4376 * config/sh/sh.c (sh_attribute_table): Same.
4377 * config/sparc/sparc.c (sparc_attribute_table): Same.
4378 * config/spu/spu.c (spu_attribute_table): Same.
4379 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4380 * config/v850/v850.c (v850_attribute_table): Same.
4381 * config/visium/visium.c (visium_attribute_table): Same.
4383 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4386 * config/arm/arm.c (INCLUDE_STRING): Define.
4387 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4388 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4391 2017-12-07 Michael Matz <matz@suse.de>
4393 Add unroll and jam pass
4395 * gimple-loop-jam.c: New file.
4396 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4397 * common.opt (funroll-and-jam): New option.
4398 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4399 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4400 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4401 * passes.def: Add pass_loop_jam.
4402 * timevar.def (TV_LOOP_JAM): Add.
4403 * tree-pass.h (make_pass_loop_jam): Declare.
4404 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4405 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4406 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4407 to flow_loop_tree_node_add.
4408 (duplicate_subloops, copy_loops_to): Append to sibling list.
4409 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4410 * doc/invoke.texi (-funroll-and-jam): Document new option.
4411 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4413 2017-12-07 Richard Biener <rguenther@suse.de>
4415 PR tree-optimization/83296
4416 PR tree-optimization/67769
4417 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4418 flow sensitive info in an unrelated BB.
4419 (value_replacement): Use reset_flow_sensitive_info.
4420 (minmax_replacement): Reset flow sensitive info on the def
4421 we move. Do not reset flow sensitive info in the whole BB
4422 we move the stmt to.
4423 (abs_replacement): Likewise.
4425 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4428 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4429 rs6000_cpu to the given -mcpu=, or to the default processor.
4431 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4433 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4434 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4435 instead of rs6000_cpu_attr.
4436 (rs6000_adjust_cost): Ditto.
4437 (is_microcoded_insn): Ditto.
4438 (rs6000_adjust_priority): Ditto.
4439 (rs6000_issue_rate): Ditto.
4440 (rs6000_use_sched_lookahead): Ditto.
4441 (rs6000_use_sched_lookahead_guard): Ditto.
4442 (rs6000_sched_reorder): Ditto.
4443 (force_new_group): Ditto.
4444 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4445 (group_ending_nop): Ditto.
4447 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4449 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4450 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4451 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4452 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4454 (rs6000_reassociation_width): Ditto.
4455 (rs6000_emit_epilogue): Ditto.
4456 (rs6000_adjust_cost): Ditto.
4457 (is_microcoded_insn): Ditto.
4458 (is_cracked_insn): Ditto.
4459 (rs6000_adjust_priority): Ditto.
4460 (rs6000_sched_reorder): Ditto.
4461 (rs6000_sched_reorder2): Ditto.
4462 (insn_must_be_first_in_group): Ditto.
4463 (insn_must_be_last_in_group): Ditto.
4464 (rs6000_register_move_cost): Ditto.
4465 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4468 2017-12-07 Julia Koval <julia.koval@intel.com>
4470 * config.gcc: Add vaesintrin.h.
4471 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4473 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4474 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4476 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4477 * config/i386/immintrin.h: Include vaesintrin.h.
4478 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4479 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4480 _mm_aesdec_epi128): New intrinsics.
4482 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4484 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4485 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4486 c-family/c-spellcheck.cc.
4487 (best_macro_match::best_macro_match): Likewise.
4489 (struct edit_distance_traits<cpp_hashnode *>): Move to
4490 c-family/c-spellcheck.h.
4491 (class best_macro_match): Likewise.
4493 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4495 PR tree-optimization/83293
4496 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4497 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4498 might insert into empty bb.
4501 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4502 simplify for plus with :c added, and pointer_plus without that.
4503 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4504 with undefined overflow and the conversion is not widening,
4505 perform negation in utype and only convert to type afterwards.
4506 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4507 simplify for plus with :c added, and pointer_plus without that.
4508 If type is integral with undefined overflow and the conversion is
4509 not widening, perform minus in utype and only convert to type
4510 afterwards. Move the last pointer_diff_expr simplify into the
4513 2017-12-06 Martin Sebor <msebor@redhat.com>
4515 PR tree-optimization/82646
4516 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4517 strncpy, not maxlen.
4519 2017-12-06 Martin Sebor <msebor@redhat.com>
4521 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4524 PR tree-optimization/83075
4525 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4526 strncat/strncpy don't change length of source string.
4528 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4531 2017-11-29 Martin Aberg <maberg@gaisler.com>
4533 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4534 to prevent b2bst errata sequence.
4535 (sqrtdf2_fix): Likewise.
4537 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4539 PR tree-optimization/81945
4540 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4541 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4542 to dest_cfun has orig_loop_num set, either remap it to the new
4543 loop number if the loop got moved too, or clear it.
4545 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4547 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4549 (thunderx2t99_nothing): New insn reservation.
4550 (thunderx2t99_mrs): New insn reservation.
4551 (thunderx2t99_multiple): New insn reservation.
4552 (thunderx2t99_alu_basi): Add bfx to reservation.
4553 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4555 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4558 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4560 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4562 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4563 tree-ssa-pre.c::remove_dead_inserted_code.
4564 * tree-ssa-dce.h: New file.
4565 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4566 (remove_dead_inserted_code): Move and rename to function
4567 tree-ssa-dce.c::simple_dce_from_worklist.
4568 (pass_pre::execute): Update use.
4570 2017-12-05 Richard Biener <rguenther@suse.de>
4572 PR tree-optimization/83277
4573 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4574 to code-gen liveout vars.
4576 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4578 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4579 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4580 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4583 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4585 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4587 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4588 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4591 2017-12-05 Richard Biener <rguenther@suse.de>
4593 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4594 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4595 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4596 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4598 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4600 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4602 2017-12-05 Martin Liska <mliska@suse.cz>
4603 Jakub Jelinek <jakub@redhat.com>
4605 * doc/invoke.texi: Document the options.
4606 * flag-types.h (enum sanitize_code): Add
4607 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4608 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4609 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4610 * opts.c: Define new sanitizer options.
4611 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4612 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4614 2017-12-05 Julia Koval <julia.koval@intel.com>
4616 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4617 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4618 (ix86_handle_option): Handle -mavx512vnni.
4619 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4620 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4621 * config/i386/i386-c (__AVX512VNNI__): New.
4622 * config/i386/i386.c (ix86_target_string): Handle new option.
4623 (ix86_valid_target_attribute_inner_p): Handle new option.
4624 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4625 * config/i386/i386.opt (mavx512vnni): New option.
4627 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4630 * athlon.md: Disable for generic.
4631 * haswell.md: Enable for generic.
4632 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4633 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4635 (ix86_adjust_cost): Move generic to haswell path.
4637 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4639 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4640 instead of MEM_P in a couple more places. Fix formatting issues.
4642 2017-12-04 Jim Wilson <jimw@sifive.com>
4644 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4645 instead of GP_REG_LAST-1.
4646 (riscv_adjust_libcall_cfi_prologue): Likewise.
4647 (riscv_adjust_libcall_cri_epilogue): Likewise.
4648 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4651 2017-12-04 Luis Machado <luis.machado@linaro.org>
4653 * ipa-pure-const.c (check_decl): Add missing newline.
4654 (state_from_flags): Likewise.
4656 2017-12-04 Jeff Law <law@redhat.com>
4658 PR tree-optimizatin/78496
4659 * gimple-ssa-evrp-analyze.h
4660 (evrp_range_analyzer::get_vr_values): Simplify.
4661 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4662 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4663 and gimple-ssa-evrp-analyze.h.
4664 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4665 (simplify_stmt_for_jump_threading): Copy a blob of code from
4666 tree-vrp.c to use ranges to simplify statements.
4667 (dom_opt_dom_walker::before_dom_children): Call
4668 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4669 (dom_opt_dom_walker::after_dom_children): Similarly for
4670 evrp_range_analyzer::leave.
4671 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4674 * gimple-ssa-evrp-analyze.c
4675 (evrp_range_analyzer::extract_range_from_stmt): Always use
4676 vr_values::update_value_range so preexisting range info is
4677 medged with new range info, even if the new range is VR_VARYING.
4679 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4681 * combine.c: Adjust comment.
4682 (use_crosses_set_p): Delete.
4683 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4684 (try_combine): Ditto.
4686 2017-12-04 Richard Biener <rguenther@suse.de>
4688 PR tree-optimization/83255
4689 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4690 Re-add zero-iteration check.
4692 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4694 PR rtl-optimization/83245
4695 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4696 hard registers as earlyclobber, also if not in an asm.
4698 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4702 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4705 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4706 rs6000_cpu based on cpu_index, not tune_index.
4708 2017-12-04 Richard Biener <rguenther@suse.de>
4710 PR tree-optimization/83238
4711 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4712 code match comment, rejecting invalid SESE regions.
4714 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4716 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4717 require base operand is a REG_POINTER prior to reload on targets
4718 with non-equivalent space registers.
4720 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4722 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4723 (update_specialized_profile): Fix updating of counts.
4724 (perhaps_add_new_callers): Likewise.
4726 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4729 * x86-tune.def: Remove obsolette FIXMEs.
4730 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4731 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4732 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4734 (X86_TUNE_PAD_RETURNS): Disable for generic.
4736 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4738 PR tree-optimization/83170
4739 PR tree-optimization/83241
4740 * gimple-ssa-store-merging.c
4741 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4742 gimple_vuse (ins_stmt) in case it has changed.
4743 (imm_store_chain_info::output_merged_store): Likewise.
4745 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4749 * tree-cfg.c (pass_warn_function_return::execute): Handle
4750 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4751 with BUILTINS_LOCATION.
4755 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4756 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4757 instead of DECL_MODE.
4759 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4760 last argument to rtx pointer.
4761 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4762 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4763 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4764 depending on the chosen ISAs.
4765 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4766 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4767 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4769 * config/i386/sse.md (mov<mode>_internal): Likewise.
4770 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4772 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4774 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4776 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4778 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4779 parameters from prototype.
4780 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4781 parameters. Don't print a comment.
4782 (emit_fusion_gpr_load): Adjust.
4783 (emit_fusion_load_store): Adjust.
4784 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4785 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4786 comment on the second line.
4788 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4791 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4792 rs6000_cpu based on cpu_index, not tune_index.
4794 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4796 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
4797 which_alternative instead of which_alternative + 1.
4798 (output_asm_insn): Print an extra tab if the template is short.
4800 2017-12-01 Jim Wilson <jimw@sifive.com>
4802 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
4804 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
4805 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
4807 * doc/tm.texi: Regenerate.
4809 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4812 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
4813 whether we can allocate pseudos before trying to fix an address.
4814 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
4815 memory address is indexed or indirect.
4816 (floatuns_<mode>si2_hw2): Likewise.
4818 2017-12-01 Jason Merrill <jason@redhat.com>
4820 * Makefile.in (TAGS): Add c-family/*.cc.
4822 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4824 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4825 (call_value_insn): Likewise.
4826 (sibcall_insn): Likewise.
4827 (sibcall_value_insn): Likewise.
4828 (movsi_aarch64): Likewise.
4829 (movdi_aarch64): Likewise.
4830 (add_losym_): Likewise.
4831 (ldr_got_small_): Likewise.
4832 (ldr_got_small_sidi): Likewise.
4833 (ldr_got_small_28k_): Likewise.
4834 (ldr_got_small_28k_sidi): Likewise.
4835 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4836 Move output_addr_const to symbolic case. Add error check.
4838 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4840 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4841 * config/s390/s390.md: Disable MVC merging peephole if it would
4842 disable operand forwarding.
4843 (new peephole2): Split MVCs if it would turn them into up to 2
4846 2017-12-01 Richard Biener <rguenther@suse.de>
4848 PR tree-optimization/83232
4849 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
4850 detection of same access. Instead of breaking the group here
4851 do not consider the duplicate. Add comment explaining real fix.
4853 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4855 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
4857 2017-12-01 Sudakshina Das <sudi.das@arm.com>
4859 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
4861 2017-12-01 Jakub Jelinek <jakub@redhat.com>
4863 * function.h (struct function): Remove cilk_frame_decl,
4864 is_cilk_function and calls_cilk_spawn fields.
4865 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
4867 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
4868 cilk_elemental field.
4869 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
4870 * target.def: Adjust comment.
4871 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
4872 Don't test cilk_elemental.
4874 PR tree-optimization/83233
4875 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4876 bswap_stat name for the struct.
4879 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4880 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4881 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4882 SWITCH_BREAK_LABEL_P set on the label.
4883 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4884 added for default case if it was missing and not all cases covered.
4885 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4886 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4888 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4889 to avoid -Wimplicit-fallthrough warning.
4890 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4891 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4893 PR tree-optimization/83221
4894 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4896 (init_reassoc): Formatting fix.
4899 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4900 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4904 * tree-cfg.c: Include asan.h.
4905 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4907 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4909 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4911 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4913 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4914 vpcmpeqd instruction.
4916 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4918 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4919 registers type for 128bit mode.
4921 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4923 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4925 * gengtype-state.c (read_state_pair): Likewise.
4926 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4927 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4928 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4929 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4930 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4931 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4932 * builtins.c (fold_builtin_3): Likewise.
4933 * graphite-scop-detection.c
4934 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4935 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4938 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4941 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4942 parameter. Return true if ignore_decl_rtl_set_p is true and other
4943 conditions are satisfied.
4944 * asan.h (asan_protect_global): Add new parameter.
4945 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4946 to asan_protect_global calls.
4948 2017-11-30 Jim Wilson <jimw@sifive.com>
4950 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4951 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4952 -mstrict-align, add info on default value. Delete redundant lines for
4953 -mabi. Add missing -mexplicit-relocs docs.
4955 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4957 * config/arc/arc.md (trap): New pattern.
4959 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4961 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4962 instruction to end into a delay slot.
4963 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4964 can be placed into a delay slot against reg_note.
4966 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4968 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4969 labels number of usages.
4971 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4973 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4975 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4977 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4980 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4981 multiplication by power of 2 constant into two shifts + comparison.
4983 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4986 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4988 2017-11-30 Richard Biener <rguenther@suse.de>
4990 PR tree-optimization/83202
4991 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4992 allow_peel argument and guard peeling.
4993 (canonicalize_loop_induction_variables): Likewise.
4994 (canonicalize_induction_variables): Pass false.
4995 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4996 peeling from cunrolli.
4998 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5000 * combine.c (try_combine): Print a message to dump file whenever
5001 I0, I1, or I2 cannot be combined into I3.
5003 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5005 PR rtl-optimization/83156
5006 PR rtl-optimization/82621
5007 * combine.c (try_combine): Don't split an I2 if one of the dests is
5008 set again before I3. Allow unused dests.
5010 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5012 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5014 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5016 PR rtl-optimization/80818
5017 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5018 recursively. Use insn code for clobber.
5019 (lra_set_insn_recog_data): Pass the new arg to
5020 collect_non_operand_hard_regs.
5021 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5023 (lra_update_insn_regno_info): Pass insn to
5024 add_regs_to_insn_regno_info.
5026 2017-11-29 Jim Wilson <jimw@sifive.com>
5027 Andrew Waterman <andrew@sifive.com>
5029 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5030 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5031 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5032 use SINGLE_SHIFT_COST.
5033 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5034 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5036 2017-11-29 Julia Koval <julia.koval@intel.com>
5038 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5039 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5040 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5041 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5042 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5043 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5044 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5045 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5046 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5047 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5048 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5049 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5050 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5051 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5052 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5053 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5054 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5055 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5056 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5057 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5058 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5059 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5060 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5061 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5062 __builtin_ia32_vpshldv_v2di_mask,
5063 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5064 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5065 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5067 2017-11-29 Julia Koval <julia.koval@intel.com>
5069 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5070 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5071 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5072 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5073 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5074 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5075 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5076 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5077 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5078 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5079 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5080 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5081 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5082 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5083 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5084 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5085 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5086 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5087 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5088 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5089 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5091 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5093 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5094 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5097 2017-11-29 Martin Aberg <maberg@gaisler.com>
5099 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5100 to prevent b2bst errata sequence.
5101 (sqrtdf2_fix): Likewise.
5103 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5105 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5106 (div_sqrt_insn_p): New function.
5107 (sparc_do_work_around_errata): Insert NOP instructions to
5108 prevent sequences that could trigger the TN-0013 errata for
5109 certain LEON3 processors.
5110 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5111 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5112 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5113 (in_branch_delay): Prevent div and sqrt in delay slot if
5115 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5117 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5119 * config/sparc/sparc.c (atomic_insn_p): New function.
5120 (sparc_do_work_around_errata): Insert NOP instructions to
5121 prevent sequences that could trigger the TN-0010 errata for
5123 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5124 instruction referable in atomic_insns_p.
5126 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5128 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5129 (atomic_compare_and_swap_leon3_1): Likewise.
5132 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5134 * config/sparc/sparc.c (fpop_insn_p): New function.
5135 (sparc_do_work_around_errata): Insert NOP instructions to
5136 prevent sequences that could trigger the TN-0012 errata for
5138 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5139 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5140 (in_branch_annul_delay): Prevent floating-point instructions
5141 in delay slot of annulled integer branch.
5143 2017-11-29 Richard Biener <rguenther@suse.de>
5145 PR tree-optimization/83202
5146 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5148 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5149 nodes vectorized to the same stmts multiple times.
5150 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5151 (vect_analyze_slp_instance): Adjust.
5152 (scalar_stmts_to_slp_tree_map_t): New typedef.
5153 (vect_schedule_slp_instance): Add a map recording the SLP node
5154 representing the vectorized stmts for a set of scalar stmts.
5155 Avoid code-generating redundancies.
5156 (vect_schedule_slp): Allocate map and pass it down.
5158 2017-11-29 Nathan Sidwell <nathan@acm.org>
5161 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5164 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5166 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5167 cases for printing LDP/STP memory addresses.
5168 (aarch64_print_address_internal): Renamed from
5169 aarch64_print_operand_address, added parameter, add Pmode check.
5170 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5171 (aarch64_print_operand_address): Indirect to
5172 aarch64_print_address_internal.
5173 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5174 'y' operand output specifier.
5176 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5179 * tree.c (build_simple_mem_ref_loc): Handle
5180 get_addr_base_and_unit_offset returning a MEM_REF.
5183 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5186 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5189 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5191 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5193 (layout::move_to_column): Likewise.
5195 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5198 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5199 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5201 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5202 Martin Jambor <mjambor@suse.cz>
5205 * tree.h (expr_type_first_operand_type_p): Declare
5206 * tree.c (expr_type_first_operand_type_p): New function.
5207 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5208 (ipa_value_from_jfunc): Adjust declaration.
5209 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5210 Use it as result type for arithmetics, unless it is NULL in which case
5211 be more conservative.
5212 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5213 ipa_get_jf_pass_through_result.
5214 (propagate_vals_across_pass_through): Likewise.
5215 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5216 is to propagate_vals_across_pass_through.
5217 (propagate_constants_across_call): Pass PARM_TYPE to
5218 propagate_scalar_across_jump_function.
5219 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5220 ipa_value_from_jfunc.
5221 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5222 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5223 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5224 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5225 target_type, pass it to ipa_value_from_jfunc.
5226 (update_indirect_edges_after_inlining): Pass parameter type to
5227 try_make_edge_direct_simple_call.
5229 2017-11-28 Jeff Law <law@redhat.com>
5231 * gimple-ssa-evrp-analyze.c
5232 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5233 refine ranges if scev_initialized_p returns true.
5234 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5236 2017-11-28 Julia Koval <julia.koval@intel.com>
5238 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5239 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5240 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5241 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5242 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5243 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5244 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5245 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5246 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5247 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5248 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5249 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5250 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5251 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5252 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5253 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5254 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5255 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5256 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5257 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5258 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5259 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5261 2017-11-28 Julia Koval <julia.koval@intel.com>
5263 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5264 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5265 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5266 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5267 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5268 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5269 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5270 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5271 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5272 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5273 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5274 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5275 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5276 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5277 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5278 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5279 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5280 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5281 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5282 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5283 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5284 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5285 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5286 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5287 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5288 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5289 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5290 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5291 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5292 __builtin_ia32_vpshld_v2di_mask): New builtins.
5293 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5294 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5296 2017-11-28 Richard Biener <rguenther@suse.de>
5298 PR tree-optimization/80776
5299 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5301 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5303 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5304 If the incoming edge is an effective fallthru because the other
5305 edge only reaches a __builtin_unreachable () then record ranges
5306 derived from the controlling condition in SSA info.
5307 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5308 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5310 2017-11-28 Olivier Hainque <hainque@adacore.com>
5312 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5313 on mingw build hosts.
5315 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5318 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5320 2017-11-28 Richard Biener <rguenther@suse.de>
5323 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5324 copies generated from memcpy use a character array as reference
5327 2017-11-28 Julia Koval <julia.koval@intel.com>
5328 Sebastian Peryt <sebastian.peryt@intel.com>
5330 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5331 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5332 cilk-common.o, cilk.h, cilk-common.c): Remove.
5334 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5335 * builtins.c (is_builtin_name): Remove cilkplus condition.
5336 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5337 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5338 cilk-builtins.def, cilkplus.def): Remove.
5339 * cif-code.def (CILK_SPAWN): Remove.
5340 * cilk-builtins.def: Delete.
5341 * cilk-common.c: Ditto.
5343 * cilkplus.def: Ditto.
5344 * config/darwin.h (fcilkplus): Delete.
5345 * cppbuiltin.c: Ditto.
5346 * doc/extend.texi: Remove cilkplus doc.
5347 * doc/generic.texi: Ditto.
5348 * doc/invoke.texi: Ditto.
5349 * doc/passes.texi: Ditto.
5350 * gcc.c (fcilkplus): Remove.
5351 * gengtype.c (cilk.h): Remove.
5352 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5354 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5356 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5358 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5359 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5360 cilkplus conditions.
5361 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5362 inline_read_section): Ditto.
5363 * ipa-inline-analysis.c (cilk.h): Remove.
5364 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5365 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5366 append_linker_options): Remove condition for fcilkplus.
5367 * lto/lto-lang.c (cilk.h): Remove.
5368 (lto_init): Remove condition for fcilkplus.
5369 * omp-expand.c (expand_cilk_for_call): Delete.
5370 (expand_omp_taskreg, expand_omp_for_static_chunk,
5371 expand_omp_for): Remove cilkplus
5373 (expand_cilk_for): Delete.
5374 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5375 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5376 execute_lower_omp, diagnose_sb_0): Ditto.
5377 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5378 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5379 * tree-nested.c: Ditto.
5380 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5381 (dump_generic_node): Ditto.
5382 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5383 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5385 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5387 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5389 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5390 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5391 for vector mode and !TARGET_SIMD.
5393 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5395 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5397 * tree.h (SWITCH_LABELS): Remove.
5398 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5399 assert SWITCH_BODY is non-NULL.
5400 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5402 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5404 PR tree-optimization/80788
5405 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5406 has TREE_OVERFLOW set, call drop_tree_overflow.
5408 2017-11-28 Richard Biener <rguenther@suse.de>
5410 PR tree-optimization/83158
5411 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5413 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5416 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5417 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5419 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5421 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5422 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5424 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5427 * varasm.c (compare_constant): Take the mode of the constants into
5428 account when comparing floating point constants.
5430 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5432 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5433 from explicit instantiation of debug_helper.
5434 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5436 2017-11-27 Richard Biener <rguenther@suse.de>
5438 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5441 2017-11-27 Richard Biener <rguenther@suse.de>
5443 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5444 (type_hash_canon): Also clear int_cst_hash_table entry for
5446 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5448 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5450 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5451 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5452 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5454 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5456 * hash-map.h (gt_cleare_cache): Avoid UB.
5458 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5460 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5461 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5462 (decide_unroll_runtime_iterations): Likewise.
5463 (decide_unroll_stupid): Likewise.
5465 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5468 * config/i386/i386.md: Add a loop with incssp.
5470 2017-11-27 Martin Jambor <mjambor@suse.cz>
5472 PR tree-optimization/81248
5473 * tree-sra.c (splice_param_accesses): Remove size check.
5474 (decide_one_param_reduction): Fix size check.
5475 * gimple-pretty-print.c (dump_profile): Silence warning.
5476 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5478 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5481 * dbxout.c (lastlineno): New variable.
5482 (dbx_debug_hooks): Use dbxout_switch_text_section as
5483 switch_text_section debug hook.
5484 (dbxout_function_end): Switch to current_function_section
5485 rather than function_section. If crtl->has_bb_partition,
5486 output just one N_FUN, depending on in_cold_section_p.
5487 (dbxout_source_line): Remember last lineno in lastlineno.
5488 (dbxout_switch_text_section): New function.
5489 (dbxout_function_decl): Adjust dbxout_block caller.
5490 (dbx_block_with_cold_children): New function.
5491 (dbxout_block): Return true if any LBRAC/RBRAC have been
5492 emitted. Use dbx_block_with_cold_children at depth == 0
5493 in second partition. Add PARENT_BLOCKNUM argument, pass
5494 it optionally adjusted to children. Output LBRAC/RBRAC
5495 around recursive call only if the block is in the current
5496 partition, if not and anything was output, emit empty
5498 * final.c (final_scan_insn): Compute cold_function_name
5499 before calling switch_text_section debug hook. Call
5500 that hook even if dwarf2out_do_frame if not emitting
5504 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5505 TREE_READONLY decls.
5507 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5509 PR rtl-optimization/82488
5510 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5512 2017-11-26 Julia Koval <julia.koval@intel.com>
5514 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5516 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5519 2017-11-26 Julia Koval <julia.koval@intel.com>
5521 * config/i386/driver-i386.c (host_detect_local_cpu):
5522 Detect skylake-avx512.
5524 2017-11-26 Julia Koval <julia.koval@intel.com>
5526 * config.gcc: Add -march=cannonlake.
5527 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5528 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5529 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5530 (PTA_CANNONLAKE): New.
5531 (processor_target_table): Add cannonlake.
5532 (ix86_option_override_internal): Ditto.
5533 (fold_builtin_cpu): Ditto.
5534 (get_builtin_code_for_version): Handle cannonlake.
5535 (M_INTEL_COREI7_CANNONLAKE): New.
5536 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5537 * doc/invoke.texi: Add -march=cannonlake.
5539 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5541 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5542 (try_init_one_plugin): Alternative implementation for MinGW.
5543 * Makefile.in (plugin_implib): New.
5544 (gengtype-lex.c): Fix broken AIX workaround.
5545 * configure: Regenerate.
5546 * doc/plugins.texi: Document support for MinGW.
5548 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5550 PR rtl-optimization/81553
5551 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5552 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5553 is a shift where C1 has different mode than the whole shift, use C1's
5554 mode for MULT rather than the shift's mode.
5557 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5558 builtins not enabled in the currently selected ISA.
5560 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5562 PR tree-optimization/71026
5563 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5564 (insert_reciprocals): Change to insert reciprocals before a division
5565 by a square and to insert the square of a reciprocal.
5566 (execute_cse_reciprocals_1): Change to consider division by a square.
5567 (register_division_in): Add importance parameter.
5569 2017-11-24 Richard Biener <rguenther@suse.de>
5571 PR tree-optimization/82402
5572 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5573 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5575 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5577 * match.pd (0-ptr): New transformation.
5579 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5582 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5583 when skipping checking.
5585 2017-11-24 Richard Biener <rguenther@suse.de>
5587 PR tree-optimization/83128
5588 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5589 (vn_reference_lookup_3): Likewise.
5591 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5594 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5595 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5598 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5599 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5600 check for the latter with gimple_call_builtin_p. Do not handle
5601 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5603 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5605 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5608 2017-11-23 Julia Koval <julia.koval@intel.com>
5610 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5611 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5612 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5613 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5614 _mm512_maskz_expandloadu_epi16): New intrinsics.
5615 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5616 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5617 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5618 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5619 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5620 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5621 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5622 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5623 _mm256_maskz_expandloadu_epi8): New intrinsics.
5624 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5625 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5626 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5627 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5628 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5629 * config/i386/sse.md (VI248_VLBW): New iterator.
5630 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5632 2017-11-23 Julia Koval <julia.koval@intel.com>
5634 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5635 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5636 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5637 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5638 _mm512_mask_compressstoreu_epi16): New.
5639 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5640 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5641 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5642 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5643 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5644 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5645 _mm256_mask_compressstoreu_epi8): New.
5646 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5647 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5648 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5649 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5650 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5651 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5652 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5653 __builtin_ia32_compresshi128_mask,
5654 __builtin_ia32_compressstoreuqi512_mask,
5655 __builtin_ia32_compressstoreuhi512_mask,
5656 __builtin_ia32_compressstoreuqi256_mask,
5657 __builtin_ia32_compressstoreuqi128_mask,
5658 __builtin_ia32_compressstoreuhi256_mask,
5659 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5660 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5662 (ix86_expand_special_args_builtin): Handle new types.
5663 (s4fma_expand): Handle new builtin array.
5664 * config/i386/immintrin.h: Include new headers.
5665 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5666 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5668 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5670 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5671 fixes. Declare temp and g variables at the top in order to avoid
5672 {} in most of the cases.
5674 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5676 * match.pd (ptr-0): New transformation.
5678 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5680 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5681 (struct addr_mode_cost_table): New.
5682 (struct tune_params): Add field addr_mode_costs.
5683 * config/arm/arm.c (generic_addr_mode_costs): New.
5684 (arm_slowmul_tune): Initialise addr_mode_costs field.
5685 (arm_fastmul_tune): Likewise.
5686 (arm_strongarm_tune): Likewise.
5687 (arm_xscale_tune): Likewise.
5688 (arm_9e_tune): Likewise.
5689 (arm_marvell_pj4_tune): Likewise.
5690 (arm_v6t2_tune): Likewise.
5691 (arm_cortex_tune): Likewise.
5692 (arm_cortex_a8_tune): Likewise.
5693 (arm_cortex_a7_tune): Likewise.
5694 (arm_cortex_a15_tune): Likewise.
5695 (arm_cortex_a35_tune): Likewise.
5696 (arm_cortex_a53_tune): Likewise.
5697 (arm_cortex_a57_tune): Likewise.
5698 (arm_exynosm1_tune): Likewise.
5699 (arm_xgene1_tune): Likewise.
5700 (arm_cortex_a5_tune): Likewise.
5701 (arm_cortex_a9_tune): Likewise.
5702 (arm_cortex_a12_tune): Likewise.
5703 (arm_cortex_a73_tune): Likewise.
5704 (arm_v7m_tune): Likewise.
5705 (arm_cortex_m7_tune): Likewise.
5706 (arm_v6m_tune): Likewise.
5707 (arm_fa726te_tune): Likewise.
5708 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5710 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5712 * config/arm/arm.c (arm_mem_costs): New function.
5713 (arm_rtx_costs_internal): Use arm_mem_costs.
5715 2017-11-23 Mark Wielaard <mark@klomp.org>
5717 * dwarf2out.c (init_sections_and_labels): Use generation to create
5718 unique ranges_section_label and ranges_base_label. Return generation.
5719 (output_rnglists): Add generation argument. Use generation to create
5720 unique ranges labels.
5721 (dwarf2out_finish): Get generation from init_sections_and_labels
5722 and pass generation to output_rnglists.
5724 2017-11-23 Mike Stump <mikestump@comcast.net>
5725 Eric Botcazou <ebotcazou@adacore.com>
5727 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5728 * cfgloop.h (struct loop): Add unroll field.
5729 * function.h (struct function): Add has_unroll bitfield.
5730 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5731 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5732 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5733 (pass_rtl_unroll_loops::gate): Likewise.
5734 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5735 for which loop->unroll==1.
5736 (decide_unroll_constant_iterations): Use note for consistency and
5737 take loop->unroll into account. Return early if loop->unroll is set.
5738 Fix thinko in existing test.
5739 (decide_unroll_runtime_iterations): Use note for consistency and
5740 take loop->unroll into account.
5741 (decide_unroll_stupid): Likewise.
5742 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5743 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5744 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5746 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5747 (print_loop): Print loop->unroll if set.
5748 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5749 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5750 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5752 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5753 loop->unroll is set and smaller than the trip count. Otherwise bypass
5754 entirely the heuristics if loop->unroll is set. Remove dead note.
5755 Fix off-by-one bug in other note.
5756 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5757 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5759 (tree_unroll_loops_completely): Make static.
5760 (pass_complete_unroll::execute): Use correct type for variable.
5761 (pass_complete_unrolli::execute): Fix formatting.
5762 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5764 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5766 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5767 enable when TARGET_PREFER_AVX128 is set.
5769 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5771 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5773 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5774 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5776 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5778 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5779 frequencies. Fix estimation of aggregate parameters.
5781 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5783 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5786 2017-11-23 Tom de Vries <tom@codesourcery.com>
5788 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
5790 2017-11-23 Tom de Vries <tom@codesourcery.com>
5792 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
5795 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
5798 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
5799 sibcall_value_pcrel_fdpic): Use local variable instead of
5801 (calli_tbr_rel): Add missing operand 2.
5802 (call_valuei_tbr_rel): Add missing operand 3.
5804 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5807 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
5808 bitpos/bitsize covering the whole destination, use store_expr only if
5809 the complex mode is the same. Otherwise, use expand_normal and if
5810 it returns CONCAT, subreg each part separately instead of trying to
5811 subreg the whole result.
5813 2017-11-23 Richard Biener <rguenther@suse.de>
5815 PR tree-optimization/23094
5816 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
5817 come here from walking over backedges in the first iteration.
5818 (vn_reference_lookup_3): Skip clobbers that store the same value.
5820 2017-11-23 Richard Biener <rguenther@suse.de>
5822 PR tree-optimization/81403
5823 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5824 a block we need a leader relative to.
5825 (phi_translate_1): For nary processing require a leader from
5826 get_representative_for given we run expression simplification
5827 using match-and-simplify. Remove previous fix.
5829 2017-11-22 Jeff Law <law@redhat.com>
5831 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5832 Use new method allocate_value_range rather than accessing the
5833 vrp_value_range_pool data member directly.
5834 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5835 to use extract_range_from_stmt method to avoid need for
5836 extract_range_from_assignment method.
5837 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5838 method rather than setting values_propgated data member directly.
5839 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5840 and values propagated data members and extract_range_from_assignment
5841 method. Reorder private data members to conform to standards.
5842 Add new methods set_lattice_propagation_complete and
5843 allocate_value_range.
5845 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
5847 PR rtl-optimization/83030
5848 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
5849 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
5850 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5851 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5852 CROSSING_JUMP_P flag is already set before setting it.
5853 * cfgrtl.c (fixup_partition_crossing): Likewise.
5854 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
5857 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5859 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
5860 Handle the case where both arguments are using gen_const_vec_series.
5862 2017-11-22 David Malcolm <dmalcolm@redhat.com>
5865 * pretty-print.c (pp_format): Move quoting implementation to
5866 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
5867 to pass address of "quote" local.
5868 (pp_begin_quote): New function.
5869 (pp_end_quote): New function.
5870 * pretty-print.h (printer_fn): Convert penultimate param from bool
5872 (pp_begin_quote): New decl.
5873 (pp_end_quote): New decl.
5874 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5875 param from bool to bool *.
5876 * tree-diagnostic.h (default_tree_printer): Likewise.
5878 2017-11-22 Jeff Law <law@redhat.com>
5880 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5881 Initialize vr_values.
5882 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5883 vr_values via vr_values class instance rather than delegators.
5884 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5885 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5886 (evrp_range_analyzer::push_value_range): Likewise.
5887 (evrp_range_analyzer::pop_value_range): Likewise.
5888 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5889 most delegators. Those remaining are exposed as public interfaces.
5890 Make vr_values a pointer and private.
5891 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5893 (evrp_range_analyzer::get_vr_value): New method.
5894 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5895 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5896 (class evrp_dom_walker): Attach evrp_folder class, initialize
5897 it in the ctor. Remove temporary delegators.
5898 (evrp_dom_walker::before_dom_children): Call methods in attached
5899 evrp_range_analyzer class via class instance pointer. Use
5900 free value_range_constant_singleton to remove need for
5901 op_with_constant_singleton_value delegator method. Do not
5902 create a vrp_prop class instance for every call! Narrow
5903 scope of a couple variables.
5904 (evrp_dom_walker::cleanup): Call methods in attached
5905 evrp_range_analyzer class via class instance pointer.
5906 * vr-values.h (class vr_values): Privatize many methods and
5909 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5911 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5912 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5913 * expr.c (expand_expr_real_2): Likewise.
5914 * fold-const.c (const_unop): Likewise.
5915 * optabs-tree.c (optab_for_tree_code): Likewise.
5916 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5917 * tree-inline.c (estimate_operator_cost): Likewise.
5918 * tree-pretty-print.c (dump_generic_node): Likewise.
5919 (op_code_prio): Likewise.
5920 (op_symbol_code): Likewise.
5921 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5922 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5923 * internal-fn.c (direct_internal_fn_optab): New function.
5924 (direct_internal_fn_array, direct_internal_fn_supported_p
5925 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5926 * fold-const-call.c (fold_const_reduction): New function.
5927 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5929 * tree-vect-loop.c: Include internal-fn.h.
5930 (reduction_code_for_scalar_code): Rename to...
5931 (reduction_fn_for_scalar_code): ...this and return an internal
5933 (vect_model_reduction_cost): Take an internal_fn rather than
5935 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5937 (vectorizable_reduction): Use internal functions rather than tree
5938 codes for the reduction operation. Update calls to the functions
5940 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5941 Use calls to internal functions rather than REDUC tree codes.
5942 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5944 2017-11-22 Olivier Hainque <hainque@adacore.com>
5946 * config/vxworks.c (vxworks_override_options): Pick default
5947 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5948 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5949 DWARF_GNAT_ENCODINGS_DEFAULT.
5950 * config/vxworksae.h: Likewise.
5952 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5954 PR tree-optimization/83104
5955 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5958 2017-11-22 Marek Polacek <polacek@redhat.com>
5959 H.J. Lu <hongjiu.lu@intel.com>
5960 Jason Merrill <jason@redhat.com>
5965 * calls.c (initialize_argument_information): Call
5966 warn_parameter_passing_abi target hook.
5967 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5968 argument onto stack.
5969 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5970 * common.opt: Update -fabi-version description.
5971 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5972 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5974 (ix86_is_empty_record): New function.
5975 (ix86_warn_parameter_passing_abi): New function.
5976 (TARGET_EMPTY_RECORD_P): Redefine.
5977 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5978 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5979 * doc/tm.texi: Regenerated.
5980 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5981 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5982 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5983 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5984 instead of int_size_in_bytes.
5985 * expr.c (copy_blkmode_to_reg): Likewise.
5986 * function.c (aggregate_value_p): Return 0 for empty types.
5987 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5989 (locate_and_pad_parm): Call arg size_in_bytes instead
5991 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5992 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5993 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5995 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5996 (std_gimplify_va_arg_expr): Skip empty records. Call
5997 arg_size_in_bytes instead size_in_bytes.
5998 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5999 * tree-core.h (tree_type_common): Add empty_flag.
6000 (tree_decl_common): Update comments.
6001 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6003 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6004 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6006 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6007 * tree.c (default_is_empty_type): New function.
6008 (default_is_empty_record): New function.
6009 (arg_int_size_in_bytes): New function.
6010 (arg_size_in_bytes): New function.
6011 (get_ultimate_context): New function.
6012 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6013 TRANSLATION_UNIT_WARN_EMPTY_P.
6014 (default_is_empty_record, arg_int_size_in_bytes,
6015 arg_size_in_bytes, get_ultimate_context): Declare.
6017 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6019 * config/arm/arm.c (cmse_clear_registers): New function.
6020 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6021 code by call to cmse_clear_registers.
6022 (cmse_nonsecure_entry_clear_before_return): Likewise.
6024 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6026 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6027 (vdot_s32, vdotq_s32): New.
6028 (vdot_lane_u32, vdotq_lane_u32): New.
6029 (vdot_lane_s32, vdotq_lane_s32): New.
6032 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6035 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6036 for unsigned values with fewer HWIs than the precision.
6037 (test_overflow): New function.
6038 (wide_int_cc_tests): Call it.
6040 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6042 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6043 has the same mode class as Pmode.
6045 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6047 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6048 padding_bits_to_clear_ptr.
6049 (cmse_nonsecure_entry_clear_before_return): Likewise.
6051 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6053 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6054 auto_sbitap instead of integer bitfield to control register needing
6057 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6059 PR tree-optimization/83044
6060 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6061 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6062 that need the upper bound. Subtract offset from
6063 get_addr_base_and_unit_offset only if positive and subtract it
6064 before division by eltsize rather than after it.
6067 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6068 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6069 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6073 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6074 expand_binop with *mul_widen_optab, make sure at least one of the
6075 operands doesn't have VOIDmode.
6078 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6080 PR rtl-optimization/82044
6081 PR tree-optimization/82042
6082 * dse.c (record_store): Check for overflow.
6083 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6084 add_wild_read instead of clear_rhs_from_active_local_stores on
6085 overflow and log it into dump_file.
6087 2017-11-22 Richard Biener <rguenther@suse.de>
6089 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6090 fake edges to exit when looking for a place to insert.
6091 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6093 (insert_into_preds_of_block): Commit edge insertion immediately,
6094 assert that doesn't require new BBs.
6095 (fini_pre): Release expressions.
6096 (pass_pre::execute): Shuffle things around a bit, if the fn
6097 is too large do not compute AVAIL either as this is really the
6100 2017-11-22 Richard Biener <rguenther@suse.de>
6102 PR tree-optimization/83089
6103 * tree-if-conv.c (pass_if_conversion::execute): If anything
6104 changed reset SCEV and free the number of iteration estimates.
6106 2017-11-21 Martin Sebor <msebor@redhat.com>
6108 PR tree-optimization/82945
6109 * calls.h (warn_nonstring_bound): Remove unused function.
6111 2017-11-21 Martin Sebor <msebor@redhat.com>
6113 PR tree-optimization/82945
6114 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6115 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6116 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6118 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6119 * calls.h (get_attr_nonstring_decl): Declare new function.
6120 * doc/extend.texi (attribute nonstring): Update.
6121 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6122 get_attr_nonstring_decl and handle it.
6123 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6124 detection of nul-termination.
6125 (strlen_to_stridx): Change to a pointer.
6126 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6127 (pass_strlen::execute): Same.
6129 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6131 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6132 for the new option -mprefer-vector-width=[none|128|256|512].
6133 * config/i386/i386.c (ix86_target_string): remove old style options
6134 -mprefer-avx256 and make -mprefer-avx128 as alias.
6135 (ix86_option_override_internal): Apply defaults for the
6136 -mprefer-vector-width=[128|256] option.
6137 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6138 Implement macros to work with -mprefer-vector-width=.
6139 * config/i386/i386.opt: Implemented option
6140 -mprefer-vector-width=[none|128|256|512].
6141 * doc/invoke.texi: Documentation for
6142 -mprefer-vector-width=[none|128|256|512].
6144 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6146 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6148 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6150 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6152 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6154 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6156 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6157 * expr.c (expand_expr_real_2): Likewise.
6158 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6159 fold_binary_loc): Likewise.
6160 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6161 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6162 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6163 MINUS_EXPR transformations.
6164 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6165 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6166 * tree-inline.c (estimate_operator_cost): Likewise.
6167 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6168 op_symbol_code): Likewise.
6169 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6170 * vr-values.c (extract_range_from_binary_expr): Likewise.
6171 * varasm.c (initializer_constant_valid_p_1): Likewise.
6172 * tree.def: New tree code POINTER_DIFF_EXPR.
6174 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6176 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6177 integer suffix to movbe mnemonic.
6178 (*bswaphi2_movbe): Ditto.
6179 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6181 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6184 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6185 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6186 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6188 2017-11-21 Martin Liska <mliska@suse.cz>
6190 * tree-inline.c (expand_call_inline): Remove not needed
6193 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6194 Jakub Jelinek <jakub@redhat.com>
6197 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6198 Remove static keyword from f variable.
6200 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6202 PR tree-optimization/83086
6203 * gimple-ssa-store-merging.c
6204 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6205 rather than n.base_addr.
6207 2017-11-21 Martin Liska <mliska@suse.cz>
6209 PR rtl-optimization/82044
6210 PR tree-optimization/82042
6211 * dse.c (check_mem_read_rtx): Check for overflow.
6213 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6215 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6218 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6220 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6221 and its corresponding call arg location note.
6223 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6224 Andrew Burgess <andrew.burgess@embecosm.com>
6226 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6228 (arc_return_slot_offset): Likewise.
6229 (arc_eh_return_address_location): New declaration.
6230 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6231 (MUST_SAVE_REGISTER): Add exception handler case.
6232 (MUST_SAVE_RETURN_ADDR): Likewise.
6233 (arc_frame_pointer_required): Likewise.
6234 (arc_frame_pointer_needed): New function.
6235 (arc_compute_frame_size): Changed.
6236 (arc_expand_prologue): Likewise.
6237 (arc_expand_epilogue): Likewise.
6238 (arc_initial_elimination_offset): Likewise.
6239 (arc_return_slot_offset): Delete.
6240 (arc_eh_return_address_location): New function.
6241 (arc_builtin_setjmp_frame_value): Likewise.
6242 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6243 (EH_RETURN_STACKADJ_RTX): Define.
6244 (EH_RETURN_HANDLER_RTX): Likewise.
6245 * config/arc/arc.md (eh_return): Delete.
6247 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6249 * print-tree.h (debug_vec_tree): Remove prototype.
6250 * gdbinit.in (pvt): Remove macro.
6252 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6254 PR tree-optimization/83047
6255 * gimple-ssa-store-merging.c
6256 (imm_store_chain_info::output_merged_store): If the loads with the
6257 same vuse are in different basic blocks, for load_gsi pick a load
6258 location that is dominated by the other loads.
6261 * config/i386/i386.c (ix86_memmodel_check): Start
6262 -Winvalid-memory-model diagnostics with lowercase letter.
6265 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6266 set in_cold_section_p to the partition containing loc_list->first.
6267 When seeing loc_list->last_before_switch node, update secname and
6268 perform range_across_switch second partition handling only after that.
6271 * run-rtl-passes.c: Include debug.h.
6272 (run_rtl_passes): Call debug_hooks->assembly_start.
6273 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6277 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6278 OPTAB_DIRECT in calls to expand_simple_binop.
6280 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6283 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6285 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6288 * spellcheck.h (best_match::blithely_get_best_candidate): New
6291 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6293 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6294 with lower case letter.
6296 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6298 * config/i386/i386.md (bswaphi2): New expander.
6299 (*bswaphi2_movbe): New insn pattern.
6300 (bswaphi -> rorhi pepehole2): New peephole pattern.
6302 2017-11-20 Jeff Law <law@redhat.com>
6304 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6305 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6306 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6307 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6308 gimple-ssa-evrp-analyze.h.
6310 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6311 set BB_VISITED here.
6312 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6314 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6316 * tree-predcom.c: Add general comment on Store-Store chains.
6317 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6319 (get_chain_last_ref_at): Rename into...
6320 (get_chain_last_write_at): ...this.
6321 (get_chain_last_write_before_load): New function.
6322 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6323 CT_STORE_STORE when write reference is added.
6324 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6325 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6326 (initialize_root_vars_store_elim_1): Ditto.
6327 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6328 definition is created.
6329 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6330 chain by replacing it with dominant stored value.
6332 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6334 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6336 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6338 * vr-values.c (extract_range_from_binary_expr): Use a full range
6341 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6343 * config/arm/arm.md (R4_REGNUM): Define constant.
6344 (nonsecure_call_internal): Remove r4 clobber.
6345 (nonsecure_call_value_internal): Likewise.
6346 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6347 clobber and resequence match_operands.
6348 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6349 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6350 (nonsecure_call_value_reg_thumb2): Likewise.
6352 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6354 PR tree-optimization/78821
6355 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6356 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6357 to the address of the base rather than the base itself.
6358 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6359 (find_bswap_or_nop_finalize): New function.
6360 (find_bswap_or_nop): Use it.
6361 (bswap_replace): Return a tree rather than bool, change first
6362 argument from gimple * to gimple_stmt_iterator, allow inserting
6363 into an empty sequence, allow ins_stmt to be NULL - then emit
6364 all stmts into gsi. Fix up MEM_REF address gimplification.
6365 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6366 (struct store_immediate_info): Add N and INS_STMT non-static
6368 (store_immediate_info::store_immediate_info): Initialize them
6369 from newly added ctor args.
6370 (merged_store_group::apply_stores): Formatting fixes. Sort by
6372 (stmts_may_clobber_ref_p): For stores call also
6373 refs_anti_dependent_p.
6374 (gather_bswap_load_refs): New function.
6375 (imm_store_chain_info::try_coalesce_bswap): New method.
6376 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6377 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6378 (imm_store_chain_info::output_merged_store): Fail if number of
6379 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6380 and NOP_EXPR rhs_code.
6381 (pass_store_merging::process_store): Compute n and ins_stmt, if
6382 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6383 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6385 (pass_store_merging::execute): Calculate dominators.
6387 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6388 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6389 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6390 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6391 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6392 class pass_optimize_bswap, bswap_replace,
6393 pass_optimize_bswap::execute): Moved to ...
6394 * gimple-ssa-store-merging.c: ... this file.
6395 Include optabs-tree.h.
6396 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6397 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6398 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6399 anonymous namespace, remove static keywords.
6400 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6401 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6403 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6406 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6408 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6410 * vec.h (debug_helper): New function.
6411 (DEFINE_DEBUG_VEC): New macro.
6412 * hash-set.h (debug_helper): New function.
6413 (DEFINE_DEBUG_HASH_SET): New macro.
6414 * cfg.c (debug_slim (edge)): New function.
6415 Call DEFINE_DEBUG_VEC for edges.
6416 Call DEFINE_DEBUG_HASH_SET for edges.
6417 * cfghooks.c (debug_slim (basic_block)): New function.
6418 Call DEFINE_DEBUG_VEC for basic blocks.
6419 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6420 * print-tree.c (debug_slim): New function to handle trees.
6421 Call DEFINE_DEBUG_VEC for trees.
6422 Call DEFINE_DEBUG_HASH_SET for trees.
6423 (debug (vec<tree, va_gc>) &): Remove.
6424 (debug (<vec<tree, va_gc>) *): Remove.
6425 * print-rtl.c (debug_slim): New function to handle const_rtx.
6426 Call DEFINE_DEBUG_VEC for rtx_def.
6427 Call DEFINE_DEBUG_VEC for rtx_insn.
6428 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6429 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6430 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6431 (debug (vec<rtx_insn *> *ptr): Remove.
6432 (debug_insn_vector): Remove.
6433 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6435 2017-11-20 Tom de Vries <tom@codesourcery.com>
6437 PR rtl-optimization/82020
6438 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6439 IF_THEN_ELSE condition.
6441 2017-11-19 Jeff Law <law@redhat.com>
6443 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6444 of degenerates resulting from ignoring an edge.
6446 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6449 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6451 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6454 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6455 for uninitialied counts.
6457 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6460 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6462 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6465 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6467 2017-11-19 Tom de Vries <tom@codesourcery.com>
6469 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6472 2017-11-19 Tom de Vries <tom@codesourcery.com>
6474 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6477 2017-11-19 Tom de Vries <tom@codesourcery.com>
6479 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6482 2017-11-19 Tom de Vries <tom@codesourcery.com>
6484 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6486 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6488 2017-11-19 Tom de Vries <tom@codesourcery.com>
6490 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6492 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6493 semicolon after MCORE_EXPORT_NAME call.
6495 2017-11-19 Tom de Vries <tom@codesourcery.com>
6498 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6499 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6501 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6503 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6504 (cgraph_edge::clone): Cleanup updating of profile.
6505 * ipa-cp.c (update_profiling_info): Likewise.
6506 * ipa-inline-transform.c (inline_transform): Likewise.
6507 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6508 * ipa-split.c (execute_split_functions): Do not split when function
6510 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6511 * profile-count.c (profile_count::dump): Add global0.
6512 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6514 (profile_count::to_sreal_scale): Likewise.
6515 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6516 (profile_count::combine_with_ipa_count): New function.
6517 * profile-count.h (profile_guessed_global0adjusted): New.
6518 (profile_count::adjusted_zero): New.
6519 (profile_count::global0adjusted): New.
6520 (profile_count::combine_with_ipa_count): New.
6521 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6522 correct profile of return block of split functions.
6523 (copy_cfg_body): Remove unused profile_count.
6524 (copy_body): Likewise.
6525 (expand_call_inline): Update.
6526 (tree_function_versioning): Update.
6528 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6530 * hash-set.h (hash_set::empty): New.
6531 * tree-ssa-threadbackward.h: Delete.
6532 * tree-ssa-threadbackward.c (class thread_jumps): New.
6533 Move max_threaded_paths into class.
6534 (fsm_find_thread_path): Remove arguments that are now in class.
6535 (profitable_jump_thread_path): Rename to...
6536 (thread_jumps::profitable_jump_thread_path): ...this.
6537 (convert_and_register_jump_thread_path): Rename to...
6538 (thread_jumps::convert_and_register_current_path): ...this.
6539 (check_subpath_and_update_thread_path): Rename to...
6540 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6541 (register_jump_thread_path_if_profitable): Rename to...
6542 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6543 (handle_phi): Rename to...
6544 (thread_jumps::handle_phi): ...this.
6545 (handle_assignment): Rename to...
6546 (thread_jumps::handle_assignment): ...this.
6547 (fsm_find_control_statement_thread_paths): Rename to...
6548 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6549 (find_jump_threads_backwards): Rename to...
6550 (thread_jumps::find_jump_threads_backwards): ...this.
6551 Initialize path local data.
6552 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6553 from within thread_jumps class.
6554 (pass_early_thread_jumps::execute): Same.
6556 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6558 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6560 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6562 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6563 to canonical location.
6565 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6568 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6570 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6572 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6574 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6575 rcrt1.o%s/grcrt1.o%s for -static-pie.
6577 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6579 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6580 ix86_shift_rotate_cost): Break out from ...
6581 (ix86_rtx_costs): ... here.
6582 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6585 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6587 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6588 when entry block was promoted unlikely.
6589 (estimate_bb_frequencies): Increase frequency scale.
6590 * profile-count.h (profile_count): Export precision info.
6592 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6594 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6595 disturbin profile of entry block.
6597 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6599 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6602 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6604 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6605 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6607 2017-11-17 Jeff Law <law@redhat.com>
6609 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6610 from evrp_dom_walker class. Various methods moved into new class.
6611 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6612 (evrp_range_analyzer::enter): New method.
6613 (evrp_range_analyzer::leave): New method.
6614 (evrp_dom_walker): Remove delegators no longer needed by this class.
6615 Replace vr_values data member with evrp_range_analyzer
6617 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6618 method extracted from evrp_dom_walker::before_dom_children.
6619 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6620 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6622 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6623 Add private copy constructor and move assignment operators.
6624 Privatize methods and class data where trivially possible.
6625 (evrp_dom_walker::cleanup): New function, extracted from
6626 execute_early_vrp. Simplify access to class data.
6628 * vr-values.h (get_output_for_vrp): Prototype.
6629 * vr-values.c (get_output_for_vrp): New function extracted from
6630 vrp_visit_assignment_or_call and extract_range_from_stmt.
6631 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6633 2017-11-17 Luis Machado <luis.machado@linaro.org>
6635 * config/aarch64/aarch64.c
6636 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6637 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6638 tune_params::AUTOPREFETCHER_WEAK.
6640 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6643 * config/arm/arm.c (arm_valid_target_attribute_rec):
6644 Parse "arch=" and "+<ext>".
6645 (arm_valid_target_attribute_tree): Re-init global options.
6646 (arm_option_override): Make non-static.
6647 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6648 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6649 (__ARM_FEATURE_CRC32): Support undef.
6650 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6651 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6653 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6655 * gdbinit.in (break-on-diagnostic): New command.
6657 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6659 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6660 return to indirect jump for EH return if control-flow
6661 protection is enabled. Change explicit 'false' argument in
6662 pro_epilogue_adjust_stack with a value of
6664 * config/i386/i386.md (simple_return_indirect_internal):
6665 Remove SImode restriction to support 64-bit.
6667 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6669 * combine.c (added_notes_insn): New.
6670 (try_combine): Handle added_notes_insn like added_links_insn.
6671 Rewrite return value code.
6672 (distribute_notes): Set added_notes_insn to the earliest insn we added
6675 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6677 PR rtl-optimization/82621
6678 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6679 dest of one of those SETs is unused.
6681 2017-11-17 Richard Biener <rguenther@suse.de>
6684 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6685 * tree-pretty-print.c (dump_generic_node): Handle
6686 annot_expr_parallel_kind.
6687 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6688 * gimplify.c (gimple_boolify): Likewise.
6690 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6692 * config.gcc (extra_headers): Add cet.h for x86 targets.
6693 * config/i386/cet.h: New file.
6694 * doc/install.texi: Add --enable-cet/--disable-cet.
6696 2017-11-17 Richard Biener <rguenther@suse.de>
6698 PR tree-optimization/83017
6699 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6700 (gen_parallel_loop): Properly count iterations.
6701 (parallelize_loops): Handle loop->can_be_parallel independent
6702 of flag_loop_parallelize_all. Make static profitability test match
6704 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6705 * invoke.texi (parloops-min-per-thread): Document.
6707 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6709 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6710 upstreaming review comments.
6712 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6714 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6715 with fast unaligned access.
6716 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6718 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6720 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6721 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6723 2017-11-17 Richard Biener <rguenther@suse.de>
6725 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6726 folding of references.
6728 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6731 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6732 of replacing call to strncmp with corresponding call to strcmp when
6735 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6737 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6738 option prefer-avx256 for skylake-avx512 configuration.
6739 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6740 (get_builtin_code_for_version): Ditto.
6742 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6743 Monk Chiang <sh.chiang04@gmail.com>
6745 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6746 (FIXED_REGISTERS): Reserve more register numbers.
6747 (CALL_USED_REGISTERS): Likewise.
6748 (REG_ALLOC_ORDER): Likewise.
6749 (REG_CLASS_CONTENTS): Likewise.
6750 (REGISTER_NAMES): Likewise.
6752 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6753 Kito Cheng <kito.cheng@gmail.com>
6755 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6757 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6759 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6761 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6762 parameter list for vec_splats.
6764 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6766 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6768 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6770 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6773 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6774 with the required version of automake 1.11.6.
6776 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6778 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6780 (DU_C2_3_power9): Correct reservation combinations.
6781 (FP_DIV_power9, VEC_DIV_power9): New.
6782 (power9-alu): Split out rotate/shift...
6783 (power9-rot): ...to here, correct dispatch resource.
6784 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6786 (power9-fp): Correct latency.
6787 (power9-sdiv): Add div/sqrt resource.
6788 (power9-ddiv): Correct latency, add div/sqrt resource.
6789 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
6790 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
6792 (power9-qpdiv, power9-qpmul): Adjust resource usage.
6794 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
6796 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
6797 switch statement mapping KF built-ins to TF built-ins if we don't
6798 have the proper ISA 3.0 assembler support.
6800 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6802 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
6803 (lower_emutls_function_body): Do not compute it.
6805 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6807 * ipa-split.c (split_bb_info): Turn time to sreal.
6808 (split_point): Likewise.
6809 (dump_split_point): Likewise.
6810 (fine_split_points): Likewise.
6811 (execute_split_functions): Only zero split_bbs; turn time to sreals.
6813 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6815 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6818 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6820 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6822 (expensive_function_p): Remove useless assert.
6823 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6825 2017-11-16 Martin Sebor <msebor@redhat.com>
6827 PR tree-optimization/82588
6828 PR tree-optimization/82583
6829 * tree-vrp.c (check_array_ref): Handle flexible array members,
6830 string literals, and inner indices.
6831 (search_for_addr_array): Add detail to diagnostics.
6833 2017-11-16 Nathan Sidwell <nathan@acm.org>
6837 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6838 (SET_DECL_ASSEMBLER_NAME): Forward to
6839 overwrite_decl_assembler_name.
6840 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6841 (overwrite_decl_assembler_name): Declare.
6842 * tree.c (overwrite_decl_assembler_name): New.
6843 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
6844 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
6845 (LANG_HOOKS_INITIALIZER): Add it.
6846 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
6847 * langhooks.c (lhd_set_decl_assembler_name): Use
6848 SET_DECL_ASSEMBLER_NAME.
6849 (lhd_overwrite_decl_assembler_name): Default implementation.
6851 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6852 Jackson Woodruff <jackson.woodruff@arm.com>
6854 PR tree-optimization/71026
6855 * match.pd: Canonicalize constant multiplies in division.
6857 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6859 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
6860 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
6861 Remove OPT_fomit_frame_pointer entry.
6862 * common/config/arc/arc-common.c: Likewise.
6863 * common/config/arm/arm-common.c: Likewise.
6864 * common/config/avr/avr-common.c: Likewise.
6865 * common/config/c6x/c6x-common.c: Likewise.
6866 * common/config/cr16/cr16-common.c: Likewise.
6867 * common/config/cris/cris-common.c: Likewise.
6868 * common/config/epiphany/epiphany-common.c: Likewise.
6869 * common/config/fr30/fr30-common.c: Likewise.
6870 * common/config/frv/frv-common.c: Likewise.
6871 * common/config/ia64/ia64-common.c: Likewise.
6872 * common/config/iq2000/iq2000-common.c: Likewise.
6873 * common/config/lm32/lm32-common.c: Likewise.
6874 * common/config/m32r/m32r-common.c: Likewise.
6875 * common/config/mcore/mcore-common.c: Likewise.
6876 * common/config/microblaze/microblaze-common.c: Likewise.
6877 * common/config/mips/mips-common.c: Likewise.
6878 * common/config/mmix/mmix-common.c: Likewise.
6879 * common/config/mn10300/mn10300-common.c: Likewise.
6880 * common/config/nios2/nios2-common.c: Likewise.
6881 * common/config/pa/pa-common.c: Likewise.
6882 * common/config/pdp11/pdp11-common.c: Likewise.
6883 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6884 * common/config/riscv/riscv-common.c: Likewise.
6885 * common/config/rs6000/rs6000-common.c: Likewise.
6886 * common/config/rx/rx-common.c: Likewise.
6887 * common/config/s390/s390-common.c: Likewise.
6888 * common/config/sh/sh-common.c: Likewise.
6889 * common/config/sparc/sparc-common.c: Likewise.
6890 * common/config/tilegx/tilegx-common.c: Likewise.
6891 * common/config/tilepro/tilepro-common.c: Likewise.
6892 * common/config/v850/v850-common.c: Likewise.
6893 * common/config/visium/visium-common.c: Likewise.
6894 * common/config/xstormy16/xstormy16-common.c: Likewise.
6895 * common/config/xtensa/xtensa-common.c: Likewise.
6896 * invoke.texi (-fomit-frame-pointer): Update documentation.
6898 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6900 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6902 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6904 * cfg.c (scale_bbs_frequencies_int,
6905 cale_bbs_frequencies_gcov_type): Remove.
6906 * cfg.h (scale_bbs_frequencies_int,
6907 cale_bbs_frequencies_gcov_type): Remove.
6909 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6911 * tree-ssa-loop-manip.c
6912 (scale_dominated_blocks_in_loop): Update to profile counts.
6913 (tree_transform_and_unroll_loop): Likewise.
6915 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6917 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6918 scale_bbs_frequencies_int.
6920 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6922 * final.c (compute_alignments): Use counts rather than frequencies.
6924 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6926 * cfgloopanal.c: Include sreal.h
6927 (average_num_loop_insns): Use counts and sreal for accounting.
6929 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6931 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6934 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6935 Kito Cheng <kito.cheng@gmail.com>
6937 * config/nds32/constraints.md: Provide more constraints.
6938 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6939 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6940 support constraints usage.
6942 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6944 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6946 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6947 Kito Cheng <kito.cheng@gmail.com>
6949 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6950 * config/nds32/nds32.opt: Refine the layout.
6951 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6952 TARGET_EXT_STRING): Support new options.
6953 * config/nds32/nds32.h: Likewise.
6954 * config/nds32/nds32.md: Likewise.
6955 * config/nds32/nds32-predicates.c: Likewise.
6956 * config/nds32/constraints.md: Likewise.
6957 * common/config/nds32/nds32-common.c: Likewise.
6959 2017-11-16 Julia Koval <julia.koval@intel.com>
6962 * config/i386/gfniintrin.h: Add sse check.
6963 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6965 2017-11-16 Julia Koval <julia.koval@intel.com>
6967 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6968 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6969 (ix86_handle_option): Handle -mavx512vbmi2.
6970 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6971 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6972 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6973 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6974 (ix86_valid_target_attribute_inner_p): Ditto.
6975 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6976 * config/i386/i386.opt (mavx512vbmi2): New option.
6977 * doc/invoke.texi: Add new option.
6979 2017-11-16 Julia Koval <julia.koval@intel.com>
6981 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6982 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6983 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6984 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6985 _mm512_gf2p8mul_epi8): New intrinsics.
6986 * config/i386/i386-builtin-types.def
6987 (V64QI_FTYPE_V64QI_V64QI): New type.
6988 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6989 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6990 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6991 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6992 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6993 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6995 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6997 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6998 explicitly as a stream of bytes.
7000 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7002 * config/rs6000/altivec.h (vec_xst_be): New #define.
7003 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7004 and externalize from *altivec_vperm_<mode>_internal.
7005 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7007 (XL_BE_V8HI): Likewise.
7008 (XL_BE_V4SI): Likewise.
7009 (XL_BE_V4SI): Likewise.
7010 (XL_BE_V2DI): Likewise.
7011 (XL_BE_V4SF): Likewise.
7012 (XL_BE_V2DF): Likewise.
7013 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7014 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7015 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7016 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7017 VSX_BUILTIN_VEC_XST_BE.
7018 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7019 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7021 (altivec_init_builtins): Replace conditional calls to def_builtin
7022 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7023 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7024 with unconditional calls. Remove calls to def_builtin for
7025 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7026 __builtin_vec_xst_be.
7027 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7028 to define_expand, and add alternate RTL generation for P8.
7029 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7030 vsx_ld_elemrev_v8hi.
7031 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7032 add alternate RTL generation for P8.
7033 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7034 vsx_ld_elemrev_v16qi.
7035 (vsx_st_elemrev_v8hi): Convert define_insn
7036 to define_expand, and add alternate RTL generation for P8.
7037 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7038 vsx_st_elemrev_v8hi.
7039 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7040 add alternate RTL generation for P8.
7041 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7042 vsx_st_elemrev_v16qi.
7044 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7047 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7048 TARGET_AVX512ER check.
7049 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7050 neither -mzeroupper nor -mno-zeroupper is used and
7051 TARGET_EMIT_VZEROUPPER is set.
7052 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7053 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7055 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7057 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7058 folding of vector compares.
7059 (fold_build_vec_cmp): New helper function.
7060 (fold_compare_helper): New helper function.
7061 (builtin_function_type): Add compare builtins to the list of functions
7062 having unsigned arguments. Cosmetic updates to comment indentation.
7063 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7064 the not+eq combination.
7066 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7068 PR tree-optimization/82726
7069 PR tree-optimization/70754
7070 * tree-predcom.c (order_drefs_by_pos): New function.
7071 (combine_chains): Move code setting has_max_use_after to...
7072 (try_combine_chains): ...here. New parameter. Sort combined chains
7073 according to position information.
7074 (tree_predictive_commoning_loop): Update call to above function.
7075 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7077 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7079 PR tree-optimization/82726
7081 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7083 PR tree-optimization/70754
7084 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7085 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7086 combined stmt before it if not NULL.
7087 (combine_chains): Process refs reversely and compute dominance point
7091 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7093 PR tree-optimization/79663
7094 * tree-predcom.c (combine_chains): Process refs in reverse order
7095 only for ZERO length chains, and add explaining comment.
7097 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7099 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7100 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7101 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7103 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7105 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7107 2017-11-15 Martin Liska <mliska@suse.cz>
7109 * tree-cfg.c (pass_warn_function_return::execute):
7110 Compare warn_return_type for greater than zero.
7112 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7116 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7117 to return true on Xeon and not on Xeon Phi.
7118 (ix86_check_avx256_register): Changed to ...
7119 (ix86_check_avx_upper_register): ... this. Add extra check for
7120 VALID_AVX512F_REG_OR_XI_MODE.
7121 (ix86_avx_u128_mode_needed): Changed
7122 ix86_check_avx256_register to ix86_check_avx_upper_register.
7123 (ix86_check_avx256_stores): Changed to ...
7124 (ix86_check_avx_upper_stores): ... this. Changed
7125 ix86_check_avx256_register to ix86_check_avx_upper_register.
7126 (ix86_avx_u128_mode_after): Changed
7127 avx_reg256_found to avx_upper_reg_found. Changed
7128 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7129 (ix86_avx_u128_mode_entry): Changed
7130 ix86_check_avx256_register to ix86_check_avx_upper_register.
7131 (ix86_avx_u128_mode_exit): Ditto.
7132 * config/i386/i386.h: (host_detect_local_cpu): New define.
7134 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7136 * config/arm/xgene1.md (xgene1): Split into automatons
7137 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7138 (xgene1_f_load): Adjust reservations and/or types.
7139 (xgene1_f_store): Likewise.
7140 (xgene1_load_pair): Likewise.
7141 (xgene1_store_pair): Likewise.
7142 (xgene1_fp_load1): Likewise.
7143 (xgene1_load1): Likewise.
7144 (xgene1_store1): Likewise.
7145 (xgene1_move): Likewise.
7146 (xgene1_alu): Likewise.
7147 (xgene1_simd): Likewise.
7148 (xgene1_bfm): Likewise.
7149 (xgene1_neon_load1): Likewise.
7150 (xgene1_neon_store1): Likewise.
7151 (xgene1_neon_logic): Likewise.
7152 (xgene1_neon_st1): Likewise.
7153 (xgene1_neon_ld1r): Likewise.
7154 (xgene1_alu_cond): Added.
7155 (xgene1_shift_reg): Likwise.
7156 (xgene1_bfx): Likewise.
7157 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7159 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7162 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7164 (can_widen_mult_without_libcall): New function.
7165 (expand_mul_overflow): If only checking unsigned mul overflow,
7166 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7167 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7168 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7169 (expand_DIVMOD): Formatting fix.
7170 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7171 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7172 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7174 PR tree-optimization/82977
7175 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7176 constructed temporary to strlen_to_stridx.put.
7178 2017-11-15 Martin Liska <mliska@suse.cz>
7180 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7181 * configure: Regenerate.
7183 2017-11-15 Martin Liska <mliska@suse.cz>
7186 * config/sh/sh-mem.cc: Use proper probability for
7189 2017-11-14 Jeff Law <law@redhat.com>
7191 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7192 the red zone for stack_clash_protection_final_dynamic_probe targets
7193 when the total dynamic stack size is zero bytes.
7195 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7196 blocks is post order.
7198 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7200 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7201 * final.c (rest_of_clean_state): Set it for the
7202 -fcompare-debug dump.
7203 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7204 class when TDF_COMPARE_DEBUG is set.
7206 * dwarf2out.c (gen_producer_string): Discard
7209 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7212 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7213 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7214 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7215 (__TGMATH_REAL_2_3): Remove macros.
7216 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7217 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7219 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7221 (remquo): Define using __TGMATH_REAL_3.
7223 2017-11-14 Jeff Law <law@redhat.com>
7225 * vr-values.c: New file with contents extracted from tree-vrp.c.
7226 * Makefile.in (OBJS): Add vr-values.o
7227 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7228 (set_value_range, set_and_canonicalize_value_range): Likewise.
7229 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7230 (value_range_constant_singleton, symbolic_range_p): Likewise.
7231 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7232 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7233 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7234 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7235 (find_case_label_range, find_case_label_index): Likewise.
7236 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7237 (range_int_cst_singleton_p, value_inside_range): Likewise.
7238 (get_single_symbol): Likewise.
7239 (switch_update): Move structure definition here.
7240 (to_remove_edges, to_update_switch_stmts): Provide externs.
7241 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7242 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7243 (vrp_val_is_min, set_value_range): Likewise.
7244 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7245 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7246 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7247 (range_is_nonnull, range_int_cst_p): Likewwise.
7248 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7249 (get_single_symbol, operand_less_p): Likewise
7250 (compare_values_warnv, compare_values): Likewise.
7251 (value_inside_range, value_range_constant_singleton): Likewise.
7252 (zero_nonzero_bitgs_from_vr): Likewise.
7253 (extract_range_from_binary_expr_1): Likewise.
7254 (overflow_comparison_p): Likewise.
7255 (to_remove_edges, to_update_switch_stmts): Likewise.
7256 (find_case_label-index, find_case_label_range): Likewise.
7257 (switch_update, set_value_range_to_nonnegative): Remove.
7258 (set_value_range_to_truthvalue): Likewise.
7259 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7260 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7261 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7262 (find_case_label_ranges, test_for_singularity): Likewise.
7263 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7264 (x_vr_values): Move to its remaining use site.
7266 2017-11-10 Jeff Law <law@redhat.com>
7268 * vr-values.h (VR_INITIALIZER): Move #define here.
7269 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7270 * Makefile.in (OBJS): Add tree-evrp.o
7271 * tree-vrp.h (assert_info): Move structure definition here.
7272 (set_value_range_to_varying): Prototype.
7273 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7274 (infer_value_range, register_edge_assert_for): Likewise.
7275 (stmt_interesting_for_vrp): Likewise.
7276 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7277 (set_value_range_to_varying): No longer static.
7278 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7279 (infer_value_range, register_edge_assert_for): Likewise.
7281 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7283 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7284 generate the XXBRD instruction.
7286 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7287 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7288 as being compatible if -mabi=ieeelongdouble.
7289 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7290 to setup float128 built-ins with hardware support.
7291 (BU_FLOAT128_HW_2): Likewise.
7292 (BU_FLOAT128_HW_3): Likewise.
7293 (BU_FLOAT128_HW_VSX_1): Likewise.
7294 (BU_FLOAT128_HW_VSX_2): Likewise.
7295 (scalar_extract_expq): Change float128 built-in functions to
7296 accommodate having both KFmode and TFmode functions. Use the
7297 KFmode variant as the default.
7298 (scalar_extract_sigq): Likewise.
7299 (scalar_test_neg_qp): Likewise.
7300 (scalar_insert_exp_q): Likewise.
7301 (scalar_insert_exp_qp): Likewise.
7302 (scalar_test_data_class_qp): Likewise.
7303 (sqrtf128_round_to_odd): Delete processing the round to odd
7304 built-in functions as special built-in functions, and define them
7305 as float128 built-ins. Use the KFmode variant as the default.
7306 (truncf128_round_to_odd): Likewise.
7307 (addf128_round_to_odd): Likewise.
7308 (subf128_round_to_odd): Likewise.
7309 (mulf128_round_to_odd): Likewise.
7310 (divf128_round_to_odd): Likewise.
7311 (fmaf128_round_to_odd): Likewise.
7312 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7313 support for KFmode and TFmode xststdcqp calls.
7314 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7315 point, switch the built-in handlers for the get/set float128
7316 exponent, get float128 mantissa, float128 test built-ins, and the
7317 float128 round to odd built-in functions. Eliminate creating the
7318 float128 round to odd built-in functions as special built-ins.
7319 (rs6000_init_builtins): Eliminate special creation of the float128
7320 round to odd built-in functions.
7321 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7322 function insns to support both TFmode and KFmode variants.
7323 (xsxsigqp_<mode>): Likewise.
7324 (xsiexpqpf_<mode>): Likewise.
7325 (xsiexpqp_<mode>): Likewise.
7326 (xststdcqp_<mode>): Likewise.
7327 (xststdcnegqp_<mode>): Likewise.
7328 (xststdcqp_<mode>): Likewise.
7330 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7332 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7333 unused path_in_freq_ptr parameter.
7334 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7336 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7338 * ipa-inline.c (edge_badness): Dump sreal frequency.
7339 (compute_inlined_call_time): Match natural implementaiton ...
7340 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7341 forgotten division by CGRAPH_FREQ_BASE.
7343 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7345 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7346 Solaris 11. Update comment.
7347 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7349 * config/sol2.h (STARTFILE_SPEC): Likewise.
7350 * configure: Regenerate.
7352 2017-11-14 Carl Love <cel@us.ibm.com>
7354 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7355 le_ and be_ prefixes to swap* variables. Remove
7356 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7359 2017-11-14 Jason Merrill <jason@redhat.com>
7361 Support GTY((cache)) on hash_map.
7362 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7363 (ggc_cache_remove): Override it instead of ggc_mx.
7364 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7365 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7366 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7367 (simple_cache_map_traits): Override maybe_mx.
7368 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7369 (hash_map): Friend gt_cleare_cache.
7370 (gt_cleare_cache): New.
7371 * tree.h (tree_cache_traits): New hash_map traits class.
7372 (tree_cache_map): New typedef.
7374 2017-11-14 Richard Biener <rguenther@suse.de>
7376 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7377 paramter and handling.
7378 (cleanup_control_flow_bb): Likewise.
7379 (cleanup_control_flow_pre): New helper performing a DFS walk
7380 to call cleanup_control_flow_bb in PRE order.
7381 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7382 via cleanup_control_flow_pre.
7384 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7386 * config/aarch64/aarch64-simd.md
7387 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7388 (*aarch64_simd_bsl<mode>_alt): Likewise.
7389 (aarch64_simd_bsldi_internal): New.
7390 (aarch64_simd_bsldi_alt): Likewise.
7392 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7394 * tracer.c (better_p): Do not compare frequencies.
7395 * reg-stack.c (better_edge): Likewise.
7396 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7399 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7401 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7402 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7403 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7404 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7405 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7406 * predict.c (maybe_hot_frequency_p): Inline to ...
7407 (maybe_hot_count_p): ... here; rewrite to counts.
7408 (counts_to_freqs): Rename to ...
7409 (update_max_bb_count): ... this one.
7410 (expensive_function_p): Use counts.
7411 (estimate_bb_frequencies): Update.
7412 (rebuild_frequencies): Update.
7413 * predict.h (counts_to_freqs): Rename to ...
7414 (update_max_bb_count): ... this one.
7415 * profile.c (compute_branch_probabilities): Add debug info
7416 * tree-inline.c (expand_call_inline): Update debug info.
7417 (optimize_inline_calls): Use update_max_bb_count..
7418 (tree_function_versioning): Use update_max_bb_count..
7419 * value-prof.c (gimple_value_profile_transformations):
7420 Do not use update_max_bb_count.
7422 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7424 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7425 always use frequencies.
7427 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7429 * bb-reorder.c: Remove frequencies from comments.
7430 (better_edge_p): Use profile counts.
7431 (find_traces): Dump profile counts.
7432 (rotate_loop): Use profile counts.
7433 (find_traces_1_round): Likewise.
7434 (connect_better_edge_p): Use counts instead of probabilities for
7436 (copy_bb_p): Drop early check for non-0 frequency.
7437 (sanitize_hot_paths): Update comments.
7439 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7441 * ipa-split.c (struct split_point): Add count.
7442 (consider_split): Do not compute incoming frequency; compute incoming
7443 count and store it to split_point.
7444 (split_function): Set count of the call to split part correctly.
7446 2017-11-13 Carl Love <cel@us.ibm.com>
7448 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7450 2017-11-13 Tom Tromey <tom@tromey.com>
7452 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7454 2017-11-13 Carl Love <cel@us.ibm.com>
7456 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7457 Add support for builtins:
7458 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7459 vector {un,}signed {char,int,short},
7460 vector {un,}signed {char,int,short}) arguments.
7461 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7462 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7463 Add BU_P9V_AV_2 expansions for the builtins.
7464 * config/rs6000/altivec.h (vec_first_match_index,
7465 vec_first_mismatch_index, vec_first_match_or_eos_index,
7466 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7467 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7468 new extern declaration.
7469 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7470 * config/rs6000/vsx.md (first_match_index_<mode>,
7471 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7472 first_mismatch_or_eos_index_<mode>): Add define expand.
7473 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7474 * doc/extend.texi: Update the built-in documenation file for the new
7477 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7479 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7480 into the min/max operations for _Float<N> and _Float<N>X types.
7482 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7485 * dwarf2out.c (do_eh_frame): New static variable.
7486 (dwarf2out_begin_prologue): Set it.
7487 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7489 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7491 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7493 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7495 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7498 * debug.h (dwarf2out_do_eh_frame): Declare.
7499 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7500 (dwarf2out_do_frame): Use it.
7501 (dwarf2out_do_cfi_asm): Likewise.
7502 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7503 (dwarf2out_assembly_start): Likewise.
7504 (dwarf2out_begin_prologue): Fix comment.
7505 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7506 if the target needs either debug or unwind DWARF2 info.
7507 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7508 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7509 -ftrapping-math, -ftrapv and -fwrapv.
7511 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7513 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7514 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7515 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7516 (estimate_edge_size_and_time): Likewise.
7517 (ipa_merge_fn_summary_after_inlining): Likewise.
7518 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7519 (compute_uninlined_call_time): Use sreal_frequency.
7520 (compute_inlined_call_time): Likewise.
7521 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7522 * profile-count.c: Include sreal.h.
7523 (profile_count::to_sreal_scale): New.
7524 * profile-count.h: Forward declare sreal.
7525 (profile_count::to_sreal_scale): Declare.
7527 2017-11-13 Nathan Sidwell <nathan@acm.org>
7529 * diagnostic.c (maybe_line_and_column): New.
7530 (diagnostic_get_location_text): Use it.
7531 (diagnostic_report_current_module): Likewise.
7532 (test_diagnostic_get_location_text): Add tests.
7534 2017-11-13 Luis Machado <luis.machado@linaro.org>
7536 * doc/md.texi (Specifying processor pipeline description): Fix
7537 incorrect latency for the div instruction example.
7539 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7541 PR tree-optimization/78821
7542 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7543 that bit_not_p is the same.
7544 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7545 (split_group): Count precisely bit_not_p bits in each statement.
7546 (invert_op): New function.
7547 (imm_store_chain_info::output_merged_store): Use invert_op to
7548 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7549 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7551 2017-11-13 Martin Liska <mliska@suse.cz>
7553 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7554 (struct source_info): Likewise.
7555 (add_branch_counts): Likewise.
7556 (add_line_counts): Likewise.
7557 (function_summary): Likewise.
7558 (output_intermediate_line): Likewise.
7559 (generate_results): Likewise.
7561 2017-11-13 Martin Liska <mliska@suse.cz>
7563 * gcov.c (struct block_info): Remove typedef for block_t.
7564 (struct line_info): Likewise.
7565 (line_info::has_block): Likewise.
7566 (EXIT_BLOCK): Likewise.
7567 (unblock): Likewise.
7568 (circuit): Likewise.
7569 (get_cycles_count): Likewise.
7570 (process_file): Likewise.
7571 (read_graph_file): Likewise.
7572 (solve_flow_graph): Likewise.
7573 (find_exception_blocks): Likewise.
7574 (add_line_counts): Likewise.
7575 (accumulate_line_info): Likewise.
7576 (output_line_details): Likewise.
7578 2017-11-13 Martin Liska <mliska@suse.cz>
7580 * gcov.c (struct arc_info): Remove typedef for arc_t.
7581 (struct line_info): Likewise.
7582 (add_branch_counts): Likewise.
7583 (output_branch_count): Likewise.
7584 (function_info::~function_info): Likewise.
7585 (circuit): Likewise.
7586 (output_intermediate_line): Likewise.
7587 (read_graph_file): Likewise.
7588 (solve_flow_graph): Likewise.
7589 (find_exception_blocks): Likewise.
7590 (add_line_counts): Likewise.
7591 (accumulate_line_info): Likewise.
7592 (output_line_details): Likewise.
7593 (output_function_details): Likewise.
7595 2017-11-13 Martin Liska <mliska@suse.cz>
7597 * gcov.c (struct function_info): Remove typedef for function_t.
7598 (struct source_info): Likewise.
7599 (source_info::get_functions_at_location): Likewise.
7600 (solve_flow_graph): Likewise.
7601 (find_exception_blocks): Likewise.
7602 (add_line_counts): Likewise.
7603 (output_intermediate_file): Likewise.
7604 (process_file): Likewise.
7605 (generate_results): Likewise.
7606 (release_structures): Likewise.
7607 (read_graph_file): Likewise.
7608 (read_count_file): Likewise.
7609 (accumulate_line_counts): Likewise.
7610 (output_lines): Likewise.
7612 2017-11-13 Martin Liska <mliska@suse.cz>
7614 * gcov.c (function_info::function_info): Remove num_counts
7615 and add vector<gcov_type>.
7616 (function_info::~function_info): Use the vector.
7617 (process_file): Likewise.
7618 (read_graph_file): Likewise.
7619 (read_count_file): Likewise.
7620 (solve_flow_graph): Likewise.
7622 2017-11-13 Martin Liska <mliska@suse.cz>
7624 * gcov.c (function_info::is_artificial): New function.
7625 (process_file): Erase all artificial early.
7626 (generate_results): Skip as all artificial are already
7629 2017-11-13 Martin Liska <mliska@suse.cz>
7631 * gcov.c (read_graph_file): Store to global vector of functions.
7632 (read_count_file): Iterate the vector.
7633 (process_file): Likewise.
7634 (generate_results): Likewise.
7635 (release_structures): Likewise.
7637 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7639 PR tree-optimization/82954
7640 * gimple-ssa-store-merging.c
7641 (imm_store_chain_info::coalesce_immediate_stores): If
7642 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7644 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7646 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7647 Upddate call to ENDIAN_LANE_N.
7648 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7649 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7650 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7651 and use aarch64_endian_lane_rtx.
7652 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7654 2017-11-12 Tom de Vries <tom@codesourcery.com>
7656 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7658 2017-11-12 Tom de Vries <tom@codesourcery.com>
7660 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7662 2017-11-12 Tom de Vries <tom@codesourcery.com>
7664 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7666 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7667 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7668 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7670 2017-11-11 Martin Sebor <msebor@redhat.com>
7673 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7676 * prefic.c (translate_name): Replace strncpy with memcpy to
7677 avoid -Wstringop-truncation.
7679 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7681 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7683 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7685 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7687 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7689 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7692 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7694 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7695 enable generating XXBRH if the value is in a vector register.
7696 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7697 value is in a vector register.
7698 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7699 register to register bswap64's instead of doing the GPR sequence
7700 used on previous machines.
7701 (bswapdi2_xxbrd): New insn.
7702 (bswapdi2_reg): Disallow on ISA 3.0.
7703 (register to register bswap64 splitter): Do not split the insn on
7704 ISA 3.0 systems that use XXBRD.
7706 2017-11-10 Martin Sebor <msebor@redhat.com>
7709 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7711 (find_subframework_file): Same.
7713 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7715 * auto-profile.c (afdo_indirect_call): Drop frequency.
7716 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7717 (cgraph_node::create_edge): Drop frequency argument.
7718 (cgraph_node::create_indirect_edge): Drop frequency argument.
7719 (cgraph_edge::make_speculative): Drop frequency arguments.
7720 (cgraph_edge::resolve_speculation): Do not update frequencies
7721 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7722 (cgraph_node::dump): Check consistency in IPA mode.
7723 (cgraph_edge::maybe_hot_p): Use IPA counter.
7724 (cgraph_edge::verify_count_and_frequency): Rename to ...
7725 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7726 (cgraph_node::verify_node): Update.
7727 * cgraph.h (struct cgraph_edge): Drop frequency.
7728 (cgraph_edge::frequency): New function.
7729 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7731 (cgraph_edge::rebuild_edges): Likewise.
7732 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7733 (duplicate_thunk_for_node): Do not pass frequency.
7734 (cgraph_node::create_clone): Scale only counts.
7735 (cgraph_node::create_virtual_clone): Do not pass frequency.
7736 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7737 (cgraph_node::create_version_clone): Do not pass frequency.
7738 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7739 (cgraph_node::expand_thunk): Do not pass frequency.
7740 (cgraph_node::create_wrapper): Do not pass frequency.
7741 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7743 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7744 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7745 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7746 (gather_caller_stats): Use frequency function.
7747 (ipcp_cloning_candidate_p): Use frequency function.
7748 (ipcp_propagate_stage): Use frequency function.
7749 (get_info_about_necessary_edges): Use frequency function.
7750 (update_profiling_info): Update only IPA profile.
7751 (update_specialized_profile): Use frequency functoin.
7752 (perhaps_add_new_callers): Update only IPA profile.
7753 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7754 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7755 (dump_ipa_call_summary): Use frequency function.
7756 (estimate_edge_size_and_time): Use frequency function.
7757 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7758 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7759 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7760 (update_noncloned_counts): ... ths one; scale counts only.
7761 (clone_inlined_nodes): Do not scale frequency.
7762 (inline_call): Do not pass frequency.
7763 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7764 (compute_inlined_call_time): Use IPA profile.
7765 (want_inline_small_function_p): Use IPA profile.
7766 (want_inline_self_recursive_call_p): Use IPA profile.
7767 (edge_badness): Use IPA profile.
7768 (lookup_recursive_calls): Use IPA profile.
7769 (recursive_inlining): Do not pass frequency.
7770 (resolve_noninline_speculation): Do not update frequency.
7771 (inline_small_functions): Collect max of IPA profile.
7772 (dump_overall_stats): Dump IPA porfile.
7773 (dump_inline_stats): Dump IPA porfile.
7774 (ipa_inline): Collect IPA stats.
7775 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7776 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7777 (ipa_propagate_frequency): Use frequency function.
7778 (ipa_profile): Cleanup.
7779 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7780 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7781 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7782 (input_node): Do not stream frequency.
7783 (input_edge): Do not stream frequency.
7784 (merge_profile_summaries): Scale only IPA profiles.
7785 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7786 * predict.c (drop_profile): Do not recompute frequency.
7787 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7788 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7789 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
7790 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
7791 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
7792 * tree-inline.c (copy_bb): Do not scale frequency.
7793 (expand_call_inline): Do not scale frequency.
7794 (tree_function_versioning): Do not scale frequency.
7795 * ubsan.c (ubsan_create_edge): Do not pass frequency.
7797 2017-11-10 Julia Koval <julia.koval@intel.com>
7799 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
7800 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
7801 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
7802 (_mm256_maskz_gf2p8affine_epi64_epi8)
7803 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
7804 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
7805 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
7806 (__builtin_ia32_vgf2p8affineqb_v32qi)
7807 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
7808 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
7810 2017-11-10 Tamar Christina <tamar.christina@arm.com>
7814 (arm_option_override): Refactor.
7815 (arm_option_reconfigure_globals): New.
7816 (arm_options_perform_arch_sanity_checks): New.
7817 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7819 (arm_options_perform_arch_sanity_checks): Likewise
7821 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7823 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7824 (power9-qpmul): New.
7825 * rs6000/rs6000.md ("type" attr): Add qmul.
7826 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7827 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7828 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7830 2017-11-10 Martin Sebor <msebor@redhat.com>
7833 * builtins.c (compute_objsize): Handle arrays that
7834 compute_builtin_object_size likes to fail for. Make extern.
7835 * builtins.h (compute_objsize): Declare.
7836 (check_strncpy_sizes): New function.
7837 (expand_builtin_strncpy): Call check_strncpy_sizes.
7838 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7839 -Wstringop-truncation.
7840 (gimple_fold_builtin_strncat): Same.
7841 * gimple.c (gimple_build_call_from_tree): Set call location.
7842 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
7843 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
7844 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
7845 (handle_builtin_strlen): Use strlen_to_stridx.
7846 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
7848 Use strlen_to_stridx.
7849 (pass_strlen::execute): Release strlen_to_stridx.
7850 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7851 (-Wstringop-truncation): Document new option.
7853 2017-11-10 Martin Liska <mliska@suse.cz>
7855 PR gcov-profile/82702
7856 * gcov.c (main): Handle intermediate files in a different
7858 (get_gcov_intermediate_filename): New function.
7859 (output_gcov_file): Remove support of intermediate files.
7860 (generate_results): Allocate intermediate file.
7861 (release_structures): Clean-up properly fn_end.
7862 (output_intermediate_file): Start iterating with line 1.
7864 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7866 PR tree-optimization/82929
7867 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7868 ops_swapped_p non-static data member.
7869 (store_immediate_info::store_immediate_info): Clear it.
7870 (imm_store_chain_info::coalesce_immediate_stores): If swapping
7871 ops set ops_swapped_p.
7872 (count_multiple_uses): Handle ops_swapped_p.
7874 2017-11-10 Martin Liska <mliska@suse.cz>
7876 * coverage.c (coverage_init): Stream information about
7877 support of has_unexecuted_blocks.
7878 * doc/gcov.texi: Document that.
7879 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7880 * gcov.c (read_graph_file): Likewise.
7881 (output_line_beginning): Fix a small issue with
7884 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7886 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7887 reference of trivial component.
7889 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7892 * gimple-ssa-store-merging.c
7893 (pass_store_merging::terminate_all_aliasing_chains): For
7894 gimple_store_p stmts also call refs_output_dependent_p.
7896 PR rtl-optimization/82913
7897 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7900 2017-11-09 Jeff Law <law@redhat.com>
7902 * vr-values.h: New file with vr_values class.
7903 * tree-vrp.c: Include vr-values.h
7904 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7905 data objects into the vr_values class.
7906 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7907 (get_value_range): Make it a member function within vr_values class.
7908 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7909 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7910 (op_with_constant_singleton_value_range): Likewise.
7911 (extract_range_for_var_from_comparison_expr): Likewise.
7912 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7913 (extract_range_from_binary_expr): Likewise.
7914 (extract_range_from_unary_expr): Likewise.
7915 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7916 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7917 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7918 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7919 (compare_name_with_value, compare_names): Likewise.
7920 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7921 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7922 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7923 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7924 (extract_range_from_phi_node): Likewise.
7925 (simplify_truth_ops_using_ranges): Likewise.
7926 (simplify_div_or_mod_using_ranges): Likewise.
7927 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7928 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7929 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7930 (simplify_float_conversion_using_ranges): Likewise.
7931 (simplify_internal_call_using_ranges): Likewise.
7932 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7933 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7934 poitner via x_vr_values for calls into gimple folder.
7935 (vrp_initialize_lattice): Make this the vr_values ctor.
7936 (vrp_free_lattice): Make this the vr_values dtor.
7937 (set_vr_value): New function.
7938 (class vrp_prop): Add vr_values data member. Add various member
7939 functions as well as member functions that delegate to vr_values.
7940 (check_array_ref): Make a member function within vrp_prop class.
7941 (search_for_addr_array, vrp_initialize): Likewise.
7942 (vrp_finalize): Likewise. Revamp to avoid direct access to
7943 vr_value, values_propagated, etc.
7944 (check_array_bounds): Extract vrp_prop class instance pointer from
7945 walk info structure. Use it to call member functions.
7946 (check_all_array_refs): Make a member function within vrp_prop class.
7947 Smuggle class instance pointer via walk info structure.
7948 (x_vr_values): New local static.
7949 (vrp_valueize): Use x_vr_values to get class instance.
7950 (vr_valueize_1): Likewise.
7951 (class vrp_folder): Add vr_values data member. Add various member
7952 functions as well as member functions that delegate to vr_values.
7953 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7954 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7955 class instance from vr_values. Use it to call member functions.
7956 (vrp_dom_walker): Add vr_values data member.
7957 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7958 instance via x_vr_values.
7959 (identify_jump_threads): Accept vr_values as argument. Store
7960 it into the walker structure.
7961 (evrp_dom_walker): Add vr_values class data member. Add various
7963 (evrp_dom_walker::try_find_new_range): Use vr_values data
7964 member to access the memory allocator.
7965 (evrp_dom_walker::before_dom_children): Store vr_values class
7966 instance into the vrp_folder class.
7967 (evrp_dom_walker::push_value_range): Rework to avoid direct
7968 access to num_vr_values and vr_value.
7969 (evrp_dom_walker::pop_value_range): Likewise.
7970 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7971 Use vr_values to get to dump_all_value_ranges member function.
7972 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7973 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7974 Pass vr_values class instance down to identify_jump_threads.
7975 Remove call to vrp_free_lattice.
7976 (debug_all_value_ranges): Remove.
7978 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7979 (vrp_folder): Likewise.
7981 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7982 Get it from the existing bitmap instead.
7983 (vrp_intersect_ranges_1): Likewise.
7985 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7987 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7989 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7990 set bit_not_p to it.
7991 (imm_store_chain_info::coalesce_immediate_stores): Break group
7992 if bit_not_p is different.
7993 (count_multiple_uses, split_group,
7994 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7995 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7996 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7997 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7998 to store_immediate_info ctor.
8000 2017-11-09 Jim Wilson <jimw@sifive.com>
8002 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8003 (scan_prog_file): Likewise.
8005 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8007 * bb-reorder.c (max_entry_frequency): Remove.
8008 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8009 connect_traces, push_to_next_round_p): Remove prototypes.
8010 (find_traces_1_round): Use counts only.
8011 (push_to_next_round_p): Likewise.
8012 (find_traces): Likewise.
8013 (rotate_loop): Likewise.
8014 (find_traces_1_round): Likewise.
8015 (connect_traces): Likewise.
8016 (edge_order): Likewise.
8018 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8020 * config/arm/arm.c (output_return_instruction): Add comments to
8021 indicate requirement for cmse_nonsecure_entry return to account
8022 for the size of clearing instruction output here.
8023 (thumb_exit): Likewise.
8024 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8025 return in hardfloat mode.
8027 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8029 * config/rs6000/rs6000.c (machine_function): Add a bool,
8030 "toc_is_wrapped_separately".
8031 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8032 if it wasn't explicitly set or unset, we are optimizing for speed, and
8033 doing separate shrink-wrapping.
8034 (rs6000_get_separate_components): Enable the TOC component if
8035 saving the TOC register in the prologue.
8036 (rs6000_components_for_bb): Handle the TOC component.
8037 (rs6000_emit_prologue_components): Store the TOC register where needed.
8038 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8039 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8041 2017-11-09 Martin Jambor <mjambor@suse.cz>
8043 * ipa-param-manipulation.c: New file.
8044 * ipa-param-manipulation.h: Likewise.
8045 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8046 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8047 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8048 (ipa_parm_adjustment): Likewise.
8049 (ipa_parm_adjustment_vec): Likewise.
8050 (ipa_get_vector_of_formal_parms): Moved declaration to
8051 ipa-param-manipulation.h.
8052 (ipa_get_vector_of_formal_parm_types): Likewise.
8053 (ipa_modify_formal_parameters): Likewise.
8054 (ipa_modify_call_arguments): Likewise.
8055 (ipa_combine_adjustments): Likewise.
8056 (ipa_dump_param_adjustments): Likewise.
8057 (ipa_modify_expr): Likewise.
8058 (ipa_get_adjustment_candidate): Likewise.
8059 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8060 ipa-param-manipulation.c.
8061 (ipa_get_vector_of_formal_parm_types): Likewise.
8062 (ipa_modify_formal_parameters): Likewise.
8063 (ipa_modify_call_arguments): Likewise.
8064 (ipa_modify_expr): Likewise.
8065 (get_ssa_base_param): Likewise.
8066 (ipa_get_adjustment_candidate): Likewise.
8067 (index_in_adjustments_multiple_times_p): Likewise.
8068 (ipa_combine_adjustments): Likewise.
8069 (ipa_dump_param_adjustments): Likewise.
8070 * tree-sra.c: Also include ipa-param-manipulation.h
8071 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8074 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8075 Alan Hayward <alan.hayward@arm.com>
8076 David Sherwood <david.sherwood@arm.com>
8078 * doc/sourcebuild.texi (vect_masked_store): Document.
8080 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8081 Alan Hayward <alan.hayward@arm.com>
8082 David Sherwood <david.sherwood@arm.com>
8084 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8086 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8087 Alan Hayward <alan.hayward@arm.com>
8088 David Sherwood <david.sherwood@arm.com>
8090 * doc/sourcebuild.texi (vect_variable_length): Document.
8092 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8093 Alan Hayward <alan.hayward@arm.com>
8094 David Sherwood <david.sherwood@arm.com>
8096 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8098 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8099 Alan Hayward <alan.hayward@arm.com>
8100 David Sherwood <david.sherwood@arm.com>
8102 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8104 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8105 Alan Hayward <alan.hayward@arm.com>
8106 David Sherwood <david.sherwood@arm.com>
8108 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8109 previously undocumented selectors.
8110 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8112 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8114 * doc/rtl.texi (const_vector): Say that elements can be
8115 const_wide_ints too.
8116 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8117 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8118 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8119 * optabs.c (expand_vector_broadcast): Likewise.
8121 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8122 Alan Hayward <alan.hayward@arm.com>
8123 David Sherwood <david.sherwood@arm.com>
8125 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8126 scaled index even if the unscaled address was invalid.
8127 Don't increase the complexity of using a scale in that case.
8129 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8130 Alan Hayward <alan.hayward@arm.com>
8131 David Sherwood <david.sherwood@arm.com>
8133 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8134 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8135 UNITS_PER_WORD bytes.
8136 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8137 based on REGMODE_NATURAL_SIZE of the inner mode.
8138 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8139 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8140 * expmed.c (lowpart_bit_field_p): Divide the value up into
8141 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8142 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8143 whether something is likely to occupy more than one register.
8145 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8148 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8149 profile_count::adjust_for_ipa_scaling.
8150 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8151 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8153 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8155 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8157 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8158 (split_group): Add total_orig and total_new arguments, estimate the
8159 number of statements related to the store group without store merging
8160 and with store merging.
8161 (imm_store_chain_info::output_merged_store): Adjust split_group
8162 callers, punt if estimated number of statements with store merging
8163 is not smaller than estimated number of statements without it.
8165 (handled_load): Remove has_single_use checks.
8166 (pass_store_merging::process_store): Likewise.
8168 2017-11-09 Richard Biener <rguenther@suse.de>
8170 PR tree-optimization/82902
8171 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8173 2017-11-09 Martin Liska <mliska@suse.cz>
8176 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8179 2017-11-09 Martin Liska <mliska@suse.cz>
8181 PR tree-optimization/82669
8182 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8184 2017-11-09 Martin Liska <mliska@suse.cz>
8186 PR gcov-profile/48463
8187 * coverage.c (coverage_begin_function): Output also end locus
8188 of a function and information whether the function is
8190 * gcov-dump.c (tag_function): Parse and print the information.
8191 * gcov.c (INCLUDE_MAP): Add include.
8192 (INCLUDE_SET): Likewise.
8193 (struct line_info): Move earlier in the source file because
8194 of vector<line_info> in function_info structure.
8195 (line_info::line_info): Likewise.
8196 (line_info::has_block): Likewise.
8197 (struct source_info): Add new member index.
8198 (source_info::get_functions_at_location): New function.
8199 (function_info::group_line_p): New function.
8200 (output_intermediate_line): New function.
8201 (output_intermediate_file): Use the mentioned function.
8202 (struct function_start): New.
8203 (struct function_start_pair_hash): Likewise.
8204 (process_file): Add code that identifies group functions.
8205 Assign lines either to global or function scope.
8206 (generate_results): Skip artificial functions.
8207 (find_source): Assign index for each source file.
8208 (read_graph_file): Read new flag artificial and end_line.
8209 (add_line_counts): Assign it either to global of function scope.
8210 (accumulate_line_counts): Isolate core of the function to
8211 accumulate_line_info and call it for both function and global
8213 (accumulate_line_info): New function.
8214 (output_line_beginning): Fix GNU coding style.
8215 (print_source_line): New function.
8216 (output_line_details): Likewise.
8217 (output_function_details): Likewise.
8218 (output_lines): Iterate both source (global) scope and function
8220 (struct function_line_start_cmp): New class.
8221 * doc/gcov.texi: Reflect changes in documentation.
8223 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8226 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8227 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8228 and similarly for not instead of neg.
8230 2017-11-08 Andi Kleen <ak@linux.intel.com>
8232 * config/i386/i386.opt: Add -mforce-indirect-call.
8233 * config/i386/predicates.md: Check for flag_force_indirect_call.
8234 * doc/invoke.texi: Document -mforce-indirect-call
8236 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8238 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8240 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8241 config/riscv/riscv.c (predict.h): New include.
8242 (riscv_slow_unaligned_access_p): No longer static.
8243 (riscv_block_move_straight): Add require.
8244 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8246 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8249 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8250 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8251 nonimmediate_operand predicate for operand 1 instead of
8254 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8256 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8258 * config/aarch64/constraints.md (Uml): New constraint.
8259 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8262 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8264 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8265 of two vec_duplicates into a vec_concat.
8267 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8269 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8270 Simplify vec_merge of vec_duplicate and vec_concat.
8271 * config/aarch64/constraints.md (Utq): New constraint.
8272 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8275 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8277 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8278 Simplify vec_merge of vec_duplicate and const_vector.
8279 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8281 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8282 mode iterator. Update predicate on operand 1 to
8283 handle non-const_vec constants. Delete constraints.
8284 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8286 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8288 PR tree-optimization/78821
8289 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8291 (store_operand_info::store_operand_info): Initialize it to false.
8292 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8293 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8294 store in the group, and if chain_info is non-NULL, to ignore altogether
8296 (compatible_load_p): Fail if bit_not_p does not match.
8297 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8298 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8299 (pass_store_merging::process_store): Adjust
8300 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8301 call terminate_all_aliasing_chains newly when adding a store into
8302 a chain with non-NULL chain_info.
8304 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8306 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8308 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8310 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8312 (aarch64_layout_frame): Initialise emit_frame_chain.
8313 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8314 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8316 2017-11-08 Martin Liska <mliska@suse.cz>
8318 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8319 of gimple_call_internal_p.
8321 2017-11-07 Tom de Vries <tom@codesourcery.com>
8323 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8325 2017-11-07 Tom de Vries <tom@codesourcery.com>
8327 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8330 2017-11-08 Martin Liska <mliska@suse.cz>
8333 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8335 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8337 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8339 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8342 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8343 (*cmp<mode>_ccz_1): New insn with $k alternative.
8346 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8347 RTX_COMM_COMPARE as commutative as well.
8348 (ix86_binary_operator_ok): Formatting fix.
8349 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8350 *<code><mode>3<mask_name><round_saeonly_name>,
8351 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8352 *<s>mul<mode>3_highpart<mask_name>,
8353 *vec_widen_umult_even_v16si<mask_name>,
8354 *vec_widen_umult_even_v8si<mask_name>,
8355 *vec_widen_umult_even_v4si<mask_name>,
8356 *vec_widen_smult_even_v16si<mask_name>,
8357 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8358 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8359 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8360 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8361 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8362 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8363 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8364 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8365 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8366 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8367 ix86_binary_operator_ok. Formatting fixes.
8368 (*<plusminus_insn><mode>3<mask_name><round_name>,
8369 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8372 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8374 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8375 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8376 eq and ne if TARGET_ISEL.
8377 (cmp): New code_iterator.
8378 (UNS, UNSU_, UNSIK): New code_attrs.
8379 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8380 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8382 ("eq<mode>3"): ... this.
8383 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8385 ("ne<mode>3"): ... this.
8387 2017-11-07 Julia Koval <julia.koval@intel.com>
8390 * common/config/i386/i386-common.c
8391 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8392 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8393 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8394 * config/i386/i386.opt: Ditto.
8395 * config/i386/i386.c (ix86_target_string): Ditto.
8396 (ix86_option_override_internal): Ditto.
8397 (ix86_init_mpx_builtins): Move MPX to args2.
8398 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8399 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8400 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8401 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8402 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8403 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8404 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8406 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8409 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8410 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8411 (zero-extendsidi peephole2): Remove peephole.
8413 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8416 * stor-layout.c: Include attribs.h.
8417 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8418 explicit lookup of "aligned" attribute.
8420 2017-11-07 Andrew Waterman <andrew@sifive.com>
8422 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8423 (riscv_expand_block_move): Likewise.
8424 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8426 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8427 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8428 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8429 (riscv_adjust_block_mem): Likewise.
8430 (riscv_block_move_loop): Likewise.
8431 (riscv_expand_block_move): Likewise.
8432 * config/riscv/riscv.md (movmemsi): New pattern.
8434 2017-11-07 Michael Clark <michaeljclark@mac.com>
8436 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8437 (MUSL_DYNAMIC_LINKER): Likewise.
8439 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8441 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8442 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8445 2017-11-07 Richard Biener <rguenther@suse.de>
8447 * match.pd: Fix build.
8449 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8450 Jackson Woodruff <jackson.woodruff@arm.com>
8452 PR tree-optimization/71026
8453 * match.pd: Canonicalize negate in division.
8455 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8458 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8460 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8462 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8463 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8465 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8467 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8468 non-scalar integral types.
8469 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8470 (-(A-B), -(~A)): New transformations.
8472 2017-11-07 Tom de Vries <tom@codesourcery.com>
8474 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8475 semicolon after "do {} while (0)".
8476 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8477 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8478 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8479 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8480 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8482 2017-11-07 Tom de Vries <tom@codesourcery.com>
8484 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8485 after "do {} while (0)".
8486 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8487 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8488 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8489 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8490 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8492 2017-11-07 Tom de Vries <tom@codesourcery.com>
8495 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8497 (arm_rtx_costs_internal): Add missing semicolon after
8498 HANDLE_NARROW_SHIFT_ARITH call.
8500 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8502 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8503 disable isel if it was not set explicitly.
8505 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8507 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8508 (add_type_attribute) likewise.
8510 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8512 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8513 of struct ix86_frame.
8514 (ix86_initial_elimination_offset): Likewise.
8515 (ix86_expand_split_stack_prologue): Likewise.
8517 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8519 * tree-vrp.h (enum value_range_type): Update stale comment.
8521 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8522 Alan Hayward <alan.hayward@arm.com>
8523 David Sherwood <david.sherwood@arm.com>
8525 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8526 (aarch64_expand_vec_perm_const): Take the number of units too.
8527 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8528 (aarch64_expand_vec_perm_const): Likewise.
8529 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8530 (vec_perm<mode>): Update accordingly.
8532 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8533 Alan Hayward <alan.hayward@arm.com>
8534 David Sherwood <david.sherwood@arm.com>
8536 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8537 Take the number of units too.
8538 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8539 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8540 but check for a vector mode before rather than after the call.
8541 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8542 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8543 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8544 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8545 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8546 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8547 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8548 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8549 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8550 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8551 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8552 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8553 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8554 (aarch64_sqdmull2_n<mode>): Update accordingly.
8556 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8557 Alan Hayward <alan.hayward@arm.com>
8558 David Sherwood <david.sherwood@arm.com>
8560 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8561 the number of units too.
8562 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8563 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8564 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8565 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8566 (vec_store_lanesxi<mode>): Update accordingly.
8568 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8569 Alan Hayward <alan.hayward@arm.com>
8570 David Sherwood <david.sherwood@arm.com>
8572 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8573 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8574 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8575 of units rather than the mode.
8576 * config/aarch64/iterators.md (nunits): New mode attribute.
8577 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8578 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8579 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8580 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8581 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8582 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8583 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8584 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8585 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8586 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8587 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8588 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8589 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8590 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8591 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8592 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8593 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8594 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8595 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8596 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8597 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8598 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8599 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8600 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8601 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8602 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8603 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8604 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8605 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8606 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8607 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8608 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8609 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8610 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8611 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8612 (aarch64_simd_vec_setv2di): Likewise.
8614 2017-11-06 Carl Love <cel@us.ibm.com>
8616 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8618 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8619 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8621 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8623 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8624 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8625 Add power 8 macro expansions.
8626 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8627 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8628 power 8 instructions. (VSX_XXBR): Add iterator.
8630 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8632 * config/arm/arm.md (predicable_short_it): Change default to "no",
8633 improve documentation, remove uses that are identical to the default.
8634 (enabled_for_depr_it): Rename to enabled_for_short_it.
8635 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8636 Remove default uses.
8637 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8638 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8639 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8640 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8642 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8645 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8646 float128 helper macros, which are no longer used after deleting
8647 the old 'q' built-in functions, and moving the round to odd
8648 built-in functions to being special built-in functions.
8649 (BU_FLOAT128_2): Likewise.
8650 (BU_FLOAT128_1_HW): Likewise.
8651 (BU_FLOAT128_2_HW): Likewise.
8652 (BU_FLOAT128_3_HW): Likewise.
8653 (FABSQ): Delete old 'q' built-in functions.
8654 (COPYSIGNQ): Likewise.
8655 (SQRTF128_ODD): Move round to odd built-in functions to be
8656 special built-in functions, so that we can handle
8657 -mabi=ieeelongdouble.
8658 (TRUNCF128_ODD): Likewise.
8659 (ADDF128_ODD): Likewise.
8660 (SUBF128_ODD): Likewise.
8661 (MULF128_ODD): Likewise.
8662 (DIVF128_ODD): Likewise.
8663 (FMAF128_ODD): Likewise.
8664 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8665 built-in names to 'f128'.
8666 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8667 old 'q' built-in functions, as the machine independent code for
8668 'f128' built-in functions handles this.
8669 (rs6000_expand_builtin): Add expansion for float128 round to odd
8670 functions, keying off on -mabi=ieeelongdouble of whether to use
8671 the KFmode or TFmode variant.
8672 (rs6000_init_builtins): Initialize the _Float128 round to odd
8674 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8675 _Float128 'q' built-in functions are now mapped into the new
8676 'f128' built-in functions.
8678 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8680 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8681 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8683 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8686 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8687 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8688 * toplev.c: Include "ipa-fnsummary.h".
8689 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8691 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8693 PR tree-optimization/82838
8694 * gimple-ssa-store-merging.c
8695 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8696 on a separate gimple_seq which is then appended to seq.
8698 2017-11-06 Jeff Law <law@redhat.com>
8701 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8702 (get_probe_interval): New functions.
8703 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8704 (ix86_adjust_stack_and_probe): Likewise.
8705 (output_adjust_stack_and_probe): Likewise.
8706 (ix86_emit_probe_stack_range): Likewise.
8707 (ix86_expand_prologue): Likewise.
8709 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8711 PR tree-optimization/82816
8712 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8713 if the modes of the two types are the same.
8714 (convert_plusminus_to_widen): Likewise.
8716 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8718 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8720 (usadv16qi): New define_expand.
8721 (usadv8hi): New define_expand.
8723 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8726 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8728 (inline_transform): Likewise.
8729 * predict.c (counts_to_freqs): Remove useless conditional.
8730 * profile-count.h (profile_count::apply_scale): Move sanity check.
8731 * tree-inline.c (copy_bb): Always scale.
8732 (copy_cfg_body): Likewise.
8734 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8737 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8739 (*cmp_ite0): Add enabled_for_depr_it attribute.
8740 (*cmp_ite1): Likewise.
8742 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8744 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8747 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8749 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8750 return the value by pointer.
8751 (extract_range_from_multiplicative_op_1): Update accordingly.
8752 Return as soon as an operation fails.
8754 2017-11-05 Tom de Vries <tom@codesourcery.com>
8757 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8758 (DEF_SANITIZER_BUILTIN): ... here.
8759 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8760 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8762 2017-11-05 Tom de Vries <tom@codesourcery.com>
8765 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8767 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8768 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8769 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8771 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8772 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8773 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8775 2017-11-05 Tom de Vries <tom@codesourcery.com>
8778 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8781 2017-11-04 Michael Clark <michaeljclark@mac.com>
8783 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8784 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8786 (*addsi3_extended): Likewise.
8787 (*addsi3_extended2): Likewise.
8788 (<optab>si3): Likewise.
8789 (<optab>di3): Likewise.
8790 (<optab><mode>3): Likewise.
8791 (<*optabe>si3_internal): Likewise.
8792 (zero_extendqi<SUPERQI:mode>2): Likewise.
8793 (*add<mode>hi3): Likewise.
8794 (*xor<mode>hi3): Likewise.
8795 (<optab>di3): Likewise.
8796 (*<optab>si3_extend): Likewise.
8797 (*sge<u>_<X:mode><GPR:mode>): Likewise.
8798 (*slt<u>_<X:mode><GPR:mode>): Likewise.
8799 (*sle<u>_<X:mode><GPR:mode>): Likewise.
8801 2017-11-04 Andrew Waterman <andrew@sifive.com>
8803 * config/riscv/riscv.c (riscv_option_override): Conditionally set
8804 TARGET_STRICT_ALIGN based upon -mtune argument.
8806 2017-11-04 Andrew Waterman <andrew@sifive.com>
8808 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
8810 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
8812 * config/i386/i386.c (choose_basereg): Use optional scratch
8813 register and add assertion.
8814 (x86_emit_outlined_ms2sysv_save): Use scratch register when
8815 needed, and don't allocate stack.
8816 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
8817 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8818 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8820 2017-11-03 Jeff Law <law@redhat.com>
8822 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8823 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8824 to probe at the start of a noreturn function.
8826 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8828 PR tree-optimization/78821
8829 * gimple-ssa-store-merging.c: Update the file comment.
8830 (MAX_STORE_ALIAS_CHECKS): Define.
8831 (struct store_operand_info): New type.
8832 (store_operand_info::store_operand_info): New constructor.
8833 (struct store_immediate_info): Add rhs_code and ops data members.
8834 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8835 and op1r arguments to the ctor, initialize corresponding data members.
8836 (struct merged_store_group): Add load_align_base and load_align
8838 (merged_store_group::merged_store_group): Initialize them.
8839 (merged_store_group::do_merge): Update them.
8840 (merged_store_group::apply_stores): Pick the constant for
8841 encode_tree_to_bitpos from one of the two operands, or skip
8842 encode_tree_to_bitpos if neither operand is a constant.
8843 (class pass_store_merging): Add process_store method decl. Remove
8844 bool argument from terminate_all_aliasing_chains method decl.
8845 (pass_store_merging::terminate_all_aliasing_chains): Remove
8846 var_offset_p argument and corresponding handling.
8847 (stmts_may_clobber_ref_p): New function.
8848 (compatible_load_p): New function.
8849 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
8850 if there is overlap and rhs_code is not INTEGER_CST. For
8851 non-overlapping stores terminate group if rhs is not mergeable.
8852 (get_alias_type_for_stmts): Change first argument from
8853 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8854 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8855 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
8857 (get_location_for_stmts): Change first argument from
8858 auto_vec<gimple *> & to vec<gimple *> &.
8859 (struct split_store): Remove orig_stmts data member, add orig_stores.
8860 (split_store::split_store): Create orig_stores rather than orig_stmts.
8861 (find_constituent_stmts): Renamed to ...
8862 (find_constituent_stores): ... this. Change second argument from
8863 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
8864 to info structures rather than the statements.
8865 (split_group): Rename ALLOW_UNALIGNED argument to
8866 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
8867 it. Adjust find_constituent_stores caller.
8868 (imm_store_chain_info::output_merged_store): Handle rhs_code other
8869 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
8870 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
8871 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
8872 (mem_valid_for_store_merging): New function.
8873 (handled_load): New function.
8874 (pass_store_merging::process_store): New method.
8875 (pass_store_merging::execute): Use process_store method. Adjust
8876 terminate_all_aliasing_chains caller.
8878 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8880 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8881 Return true for more constants, symbols and label references.
8882 (aarch64_valid_floating_const): Remove unused function.
8884 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8887 * config/aarch64/aarch64.c (aarch64_layout_frame):
8888 Undo forcing of LR at bottom of frame.
8890 2017-11-03 Jeff Law <law@redhat.com>
8893 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8894 for int_registers_saved.
8896 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8897 extracted from tree-ssa-dom.c.
8898 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8899 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8900 (record_equivalences_from_incoming_edge): Add additional argument
8901 to single_pred_edge_ignoring_loop_edges call.
8902 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8903 (uncprop_dom_walker::before_dom_children): Add additional argument
8904 to single_pred_edge_ignoring_loop_edges call.
8905 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8906 single_pred_edge_ignoring_loop_edges rather than open coding.
8907 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8909 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8911 * match.pd (-(-A)): Rewrite.
8913 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8915 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8916 (rs6000_emit_int_cmove): New declaration.
8917 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8918 (rs6000_emit_sISEL): Delete.
8919 (rs6000_emit_int_cmove): Make non-static.
8920 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8921 instead of rs6000_emit_sISEL.
8923 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8925 * asan.c (create_cond_insert_point): Maintain profile.
8926 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8928 * basic-block.h (struct basic_block_def): Remove frequency.
8929 (EDGE_FREQUENCY): Use to_frequency
8930 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8932 (find_traces): Update to use to_frequency.
8933 (find_traces_1_round): Likewise; use only IPA counts.
8934 (bb_to_key): Likewise.
8935 (connect_traces): Use IPA counts only.
8936 (copy_bb_p): Update to use to_frequency.
8937 (fix_up_crossing_landing_pad): Likewise.
8938 (sanitize_hot_paths): Likewise.
8939 * bt-load.c (basic_block_freq): Likewise.
8940 * cfg.c (init_flow): Set count_max to uninitialized.
8941 (check_bb_profile): Remove frequencies; check counts.
8942 (dump_bb_info): Do not dump frequencies.
8943 (update_bb_profile_for_threading): Update counts only.
8944 (scale_bbs_frequencies_int): Likewise.
8945 (MAX_SAFE_MULTIPLIER): Remove.
8946 (scale_bbs_frequencies_gcov_type): Update counts only.
8947 (scale_bbs_frequencies_profile_count): Update counts only.
8948 (scale_bbs_frequencies): Update counts only.
8949 * cfg.h (struct control_flow_graph): Add count-max.
8950 (update_bb_profile_for_threading): Update prototype.
8951 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8952 (find_many_sub_basic_blocks): Likewise.
8953 * cfgcleanup.c (try_forward_edges): Likewise.
8954 (try_crossjump_to_edge): Likewise.
8955 * cfgexpand.c (expand_gimple_cond): Likewise.
8956 (expand_gimple_tailcall): Likewise.
8957 (construct_init_block): Likewise.
8958 (construct_exit_block): Likewise.
8959 * cfghooks.c (verify_flow_info): Check consistency of counts.
8960 (dump_bb_for_graph): Do not dump frequencies.
8961 (split_block_1): Do not update frequencies.
8962 (split_edge): Do not update frequencies.
8963 (make_forwarder_block): Do not update frequencies.
8964 (duplicate_block): Do not update frequencies.
8965 (account_profile_record): Do not update frequencies.
8966 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8967 for global heuristics.
8968 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8969 (expected_loop_iterations_unbounded): Use counts only.
8970 * cfgloopmanip.c (scale_loop_profile): Simplify.
8971 (create_empty_loop_on_edge): Simplify
8973 (duplicate_loop_to_header_edge): Simplify
8974 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8975 (update_br_prob_note): Take care of removing note when profile
8977 (relink_block_chain): Do not dump frequency.
8978 (rtl_account_profile_record): Use to_frequency.
8979 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8980 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8981 (cgraph_update_edges_for_call_stmt_node): Likewise.
8982 (cgraph_edge::verify_count_and_frequency): Update.
8983 (cgraph_node::verify_node): Temporarily disable frequency verification.
8984 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8985 to_cgraph_frequency.
8986 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8987 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8989 (cgraph_node::expand_thunk): Update profile.
8990 * except.c (dw2_build_landing_pads): Do not update frequency.
8991 * final.c (compute_alignments): Use to_frequency.
8992 (dump_basic_block_info): Do not dump frequency.
8993 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8994 (dump_gimple_bb_header): Do not dump frequency.
8995 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8997 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8998 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8999 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9000 (init_before_recovery): Do not update frequency.
9001 (sched_create_recovery_edges): Do not update frequency.
9002 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9003 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9004 (ipa_cp_c_finalize): Set max_count to uninitialized.
9005 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9006 (param_change_prob): Use counts.
9007 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9009 * ipa-split.c (consider_split): Use to_frequency.
9010 (split_function): Use to_frequency.
9011 * ira-build.c (loop_compare_func): Likewise.
9012 (mark_loops_for_removal): Likewise.
9013 (mark_all_loops_for_removal): Likewise.
9014 * loop-doloop.c (doloop_modify): Do not update frequency.
9015 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9017 * lto-streamer-in.c (input_function): Update count_max.
9018 * omp-expand.c (expand_omp_taskreg): Update count_max.
9019 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9020 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9021 (maybe_hot_count_p): Use ipa counts only.
9022 (maybe_hot_bb_p): Simplify.
9023 (maybe_hot_edge_p): Simplify.
9024 (probably_never_executed): Do not take frequency argument.
9025 (probably_never_executed_bb_p): Do not pass frequency.
9026 (probably_never_executed_edge_p): Likewise.
9027 (combine_predictions_for_bb): Check that profile is nonzero.
9028 (propagate_freq): Do not set frequency.
9029 (drop_profile): Simplify.
9030 (counts_to_freqs): Simplify.
9031 (expensive_function_p): Use to_frequency.
9032 (propagate_unlikely_bbs_forward): Simplify.
9033 (determine_unlikely_bbs): Simplify.
9034 (estimate_bb_frequencies): Add hack to silence graphite issues.
9035 (compute_function_frequency): Use ipa counts.
9036 (pass_profile::execute): Update.
9037 (rebuild_frequencies): Use counts only.
9038 (force_edge_cold): Use counts only.
9039 * profile-count.c (profile_count::dump): Dump new count types.
9040 (profile_count::differs_from_p): Check compatiblity.
9041 (profile_count::to_frequency): New function.
9042 (profile_count::to_cgraph_frequency): New function.
9043 * profile-count.h (struct function): Declare.
9044 (enum profile_quality): Add profile_guessed_local and
9045 profile_guessed_global0.
9046 (class profile_proability): Decrease number of bits to 29;
9047 update from_reg_br_prob_note and to_reg_br_prob_note.
9048 (class profile_count: Update comment; decrease number of bits
9049 to 61. Check compatibility.
9050 (profile_count::compatible_p): New private member function.
9051 (profile_count::ipa_p): New member function.
9052 (profile_count::operator<): Handle global zero correctly.
9053 (profile_count::operator>): Handle global zero correctly.
9054 (profile_count::operator<=): Handle global zero correctly.
9055 (profile_count::operator>=): Handle global zero correctly.
9056 (profile_count::nonzero_p): New member function.
9057 (profile_count::force_nonzero): New member function.
9058 (profile_count::max): New member function.
9059 (profile_count::apply_scale): Handle IPA scalling.
9060 (profile_count::guessed_local): New member function.
9061 (profile_count::global0): New member function.
9062 (profile_count::ipa): New member function.
9063 (profile_count::to_frequency): Declare.
9064 (profile_count::to_cgraph_frequency): Declare.
9065 * profile.c (OVERLAP_BASE): Delete.
9066 (compute_frequency_overlap): Delete.
9067 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9068 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9069 * sched-ebb.c (rank): Use counts only.
9070 * shrink-wrap.c (handle_simple_exit): Use counts only.
9071 (try_shrink_wrapping): Use counts only.
9072 (place_prologue_for_one_component): Use counts only.
9073 * tracer.c (find_best_predecessor): Use to_frequency.
9074 (find_trace): Use to_frequency.
9075 (tail_duplicate): Use to_frequency.
9076 * trans-mem.c (expand_transaction): Do not update frequency.
9077 * tree-call-cdce.c: Do not update frequency.
9078 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9079 (gimple_merge_blocks): Likewise.
9080 (gimple_split_edge): Likewise.
9081 (gimple_duplicate_sese_region): Likewise.
9082 (gimple_duplicate_sese_tail): Likewise.
9083 (move_sese_region_to_fn): Likewise.
9084 (gimple_account_profile_record): Likewise.
9085 (insert_cond_bb): Likewise.
9086 * tree-complex.c (expand_complex_div_wide): Likewise.
9087 * tree-eh.c (lower_resx): Update profile.
9088 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9090 (initialize_cfun): Do not initialize frequencies
9091 (freqs_to_counts): Delete.
9092 (copy_cfg_body): Ignore count parameter.
9093 (copy_body): Update.
9094 (expand_call_inline): Update count_max.
9095 (optimize_inline_calls): Update count_max.
9096 (tree_function_versioning): Update count_max.
9097 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9098 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9100 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9101 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9102 (try_peel_loop): Likewise.
9103 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9105 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9106 (tree_transform_and_unroll_loop): Do not use frequencies
9107 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9108 Use reliable prediction only.
9109 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9110 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9111 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9112 probability scaling.
9113 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9114 not update frequency
9115 (any_remaining_duplicated_blocks): Likewise.
9116 (update_profile): Likewise.
9117 (estimated_freqs_path): Delete.
9118 (freqs_to_counts_path): Delete.
9119 (clear_counts_path): Delete.
9120 (ssa_fix_duplicate_block_edges): Likewise.
9121 (duplicate_thread_path): Likewise.
9122 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9123 * tree-tailcall.c (decrease_profile): Do not update frequency.
9124 (eliminate_tail_call): Likewise.
9125 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9126 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9127 (optimize_mask_stores): Likewise.
9128 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9129 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9130 (ubsan_expand_ptr_ifn): Update profile.
9131 * value-prof.c (gimple_ic): Simplify.
9132 * value-prof.h (gimple_ic): Update prototype.
9133 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9134 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9136 (want_inline_self_recursive_call_p): Likewise.
9137 (resolve_noninline_speculation): Only cummulate defined counts.
9138 (inline_small_functions): Use nonzero_p.
9139 (ipa_inline): Do not access freed node.
9141 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9143 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9144 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9146 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9148 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9149 non-legitimate address.
9151 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9153 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9154 (*lt0_<mode>di, *lt0_<mode>si): New.
9156 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9158 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9159 TARGET_PAIRED_FLOAT.
9161 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9162 Jim Wilson <jim.wilson@linaro.org>
9164 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9165 * config/aarch64/aarch64-tune.md: Regenerated.
9166 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9167 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9169 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9171 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9172 (arc_expand_prologue): Restore blink for millicode.
9173 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9175 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9178 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9179 gen_vec_duplicate after forcing the scalar into a register.
9181 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9183 * combine (try_combine): Print the insns input to try_combine to the
9186 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9189 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9190 Remove second argument from aarch64_process_target_attr call.
9191 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9193 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9195 (aarch64_handle_attr_arch): Remove second argument.
9196 (aarch64_handle_attr_cpu): Ditto.
9197 (aarch64_handle_attr_tune): Ditto.
9198 (aarch64_handle_attr_isa_flags): Ditto.
9199 (aarch64_process_one_target_attr): Ditto.
9200 (aarch64_process_target_attr): Ditto.
9201 (aarch64_option_valid_attribute_p): Remove second argument.
9202 on aarch64_process_target_attr call.
9204 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9206 * diagnostic.c: Include "selftest-diagnostic.h".
9207 (selftest::assert_location_text): New function.
9208 (selftest::test_diagnostic_get_location_text): New function.
9209 (selftest::diagnostic_c_tests): Call it.
9211 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9213 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9214 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9215 (class selftest::test_diagnostic_context): Move to...
9216 * selftest-diagnostic.c: New file.
9217 * selftest-diagnostic.h: New file.
9219 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9221 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9222 offset range for FT32B.
9223 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9224 * config/ft32/ft32.md: Add TARGET_NOPM.
9225 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9226 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9228 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9230 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9232 2017-11-02 Jeff Law <law@redhat.com>
9234 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9235 virtual keyword on FINAL OVERRIDE members.
9237 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9238 virtuals together. Add virtual destructor.
9239 (substitute_and_fold_engine): Similarly.
9241 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9243 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9245 2017-11-02 Richard Biener <rguenther@suse.de>
9247 PR tree-optimization/82795
9248 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9250 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9252 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9254 (gcc_GAS_CHECK_FEATURE): Remove.
9255 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9256 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9257 * configure: Regenerate.
9259 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9261 * config/arc/arc.c (hwloop_optimize): Account for empty
9264 2017-11-02 Richard Biener <rguenther@suse.de>
9267 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9268 Truncate ARRAY_REF index and element size.
9270 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9272 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9274 2017-11-01 Jeff Law <law@redhat.com>
9276 * tree-ssa-ccp.c (ccp_folder): New class derived from
9277 substitute_and_fold_engine.
9278 (ccp_folder::get_value): New member function.
9279 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9280 (ccp_fold_stmt): Remove prototype.
9281 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9282 * tree-ssa-copy.c (copy_folder): New class derived from
9283 substitute_and_fold_engine.
9284 (copy_folder::get_value): Renamed from get_value.
9285 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9286 * tree-vrp.c (vrp_folder): New class derived from
9287 substitute_and_fold_engine.
9288 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9289 (vrp_folder::get_value): New member function.
9290 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9291 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9292 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9293 provide a class interface to folder/substitute routines.
9294 (ssa_prop_fold_stmt_fn): Remove typedef.
9295 (ssa_prop_get_value_fn): Likewise.
9296 (subsitute_and_fold): Remove prototype.
9297 (replace_uses_in): Likewise.
9298 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9299 Renamed from replace_uses_in. Call the virtual member function
9300 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9301 (substitute_and_fold_dom_walker): Remove initialization of
9302 data member entries for calbacks. Add substitute_and_fold_engine
9303 member and initialize it.
9304 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9305 member functions for get_value, replace_phi_args_in c
9306 replace_uses_in, and fold_stmt calls.
9307 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9308 substitute_and_fold. Remove assert. Update ctor call.
9310 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9311 (ssa_prop_visit_phi_fn): Likewise.
9312 (class ssa_propagation_engine): New class to provide an interface
9314 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9316 (ssa_prop_visit_phi): Likewise.
9317 (ssa_propagation_engine::simulate_stmt): Moved into class.
9318 Call visit_phi/visit_stmt from the class rather than via
9319 file scoped static variables.
9320 (ssa_propagation_engine::simulate_block): Moved into class.
9321 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9322 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9323 set file scoped statics for the visit_stmt/visit_phi callbacks.
9324 * tree-complex.c (complex_propagate): New class derived from
9325 ssa_propagation_engine.
9326 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9327 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9328 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9330 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9331 ssa_propagation_engine.
9332 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9333 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9334 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9335 * tree-ssa-copy.c (copy_prop): New class derived from
9336 ssa_propagation_engine.
9337 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9338 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9339 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9340 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9341 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9342 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9343 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9345 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9347 PR rtl-optimization/82778
9348 PR rtl-optimization/82597
9349 * compare-elim.c (struct comparison): Add in_a_setter field.
9350 (find_comparison_dom_walker::before_dom_children): Remove killed
9351 bitmap and df_simulate_find_defs call, instead walk the defs.
9352 Compute last_setter and initialize in_a_setter. Merge definitions
9353 with first initialization for a few variables.
9354 (try_validate_parallel): Use insn_invalid_p instead of
9355 recog_memoized. Return insn rather than just the pattern.
9356 (try_merge_compare): Fix up comment. Don't uselessly test if
9357 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9359 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9362 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9363 Alan Hayward <alan.hayward@arm.com>
9364 David Sherwood <david.sherwood@arm.com>
9366 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9367 aarch64_hard_regno_nregs to get the number of registers
9370 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9371 Alan Hayward <alan.hayward@arm.com>
9372 David Sherwood <david.sherwood@arm.com>
9374 * config/aarch64/constraints.md (Upl): Rename to...
9376 * config/aarch64/aarch64.md
9377 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9380 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9381 Alan Hayward <alan.hayward@arm.com>
9382 David Sherwood <david.sherwood@arm.com>
9384 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9385 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9388 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9389 Alan Hayward <alan.hayward@arm.com>
9390 David Sherwood <david.sherwood@arm.com>
9392 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9393 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9394 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9396 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9397 of a single element.
9398 * config/aarch64/iterators.md: Add a comment above the permute
9399 unspecs to say that they are generated directly by
9400 aarch64_expand_vec_perm_const.
9401 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9403 2017-11-01 Nathan Sidwell <nathan@acm.org>
9405 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9407 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9409 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9410 and medany code models, and describe what they do.
9412 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9414 Revert accidental duplicate:
9416 * combine.c (can_change_dest_mode): Reject changes in
9417 REGMODE_NATURAL_SIZE.
9419 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9421 PR rtl-optimization/64682
9422 PR rtl-optimization/69567
9423 PR rtl-optimization/69737
9424 PR rtl-optimization/82683
9425 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9426 register mentioned in the note, drop the note, unless it came from I3,
9427 in which case it should go to I3 again.
9429 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9431 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9432 and return false if not.
9433 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9435 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9437 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9438 range is known to be empty.
9440 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9441 Alan Hayward <alan.hayward@arm.com>
9442 David Sherwood <david.sherwood@arm.com>
9444 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9445 and CONST_VECTOR_NUNITS instead of computing the number of units from
9446 the byte sizes of the vector and element.
9447 (simplify_binary_operation_1): Likewise.
9448 (simplify_const_binary_operation): Likewise.
9449 (simplify_ternary_operation): Likewise.
9451 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9452 Alan Hayward <alan.hayward@arm.com>
9453 David Sherwood <david.sherwood@arm.com>
9455 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9456 (int_mem_offset): ...this new function.
9457 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9458 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9459 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9462 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9463 Alan Hayward <alan.hayward@arm.com>
9464 David Sherwood <david.sherwood@arm.com>
9466 * lower-subreg.c (interesting_mode_p): New function.
9467 (compute_costs, find_decomposable_subregs, decompose_register)
9468 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9469 (resolve_clobber, dump_choices): Use it.
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 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9477 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9478 Alan Hayward <alan.hayward@arm.com>
9479 David Sherwood <david.sherwood@arm.com>
9481 * alias.c (find_base_value, find_base_term): Only process integer
9482 truncations. Check the precision rather than the size.
9484 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9485 Alan Hayward <alan.hayward@arm.com>
9486 David Sherwood <david.sherwood@arm.com>
9488 * machmode.h (is_narrower_int_mode): New function
9489 * optabs.c (expand_float, expand_fix): Use it.
9490 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9492 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9493 Alan Hayward <alan.hayward@arm.com>
9494 David Sherwood <david.sherwood@arm.com>
9496 * rtl.h (narrower_subreg_mode): New function.
9497 * ira-color.c (update_costs_from_allocno): Use it.
9499 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9500 Alan Hayward <alan.hayward@arm.com>
9501 David Sherwood <david.sherwood@arm.com>
9503 * optabs-query.h (convert_optab_p): New function, split out from...
9504 (convert_optab_handler): ...here.
9505 (widening_optab_handler): Delete.
9506 (find_widening_optab_handler): Remove permit_non_widening parameter.
9507 (find_widening_optab_handler_and_mode): Likewise. Provide an
9508 override that operates on mode class wrappers.
9509 * optabs-query.c (widening_optab_handler): Delete.
9510 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9511 parameter. Assert that the two modes are the same class and that
9512 the "from" mode is narrower than the "to" mode. Use
9513 convert_optab_handler instead of widening_optab_handler.
9514 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9515 instead of widening_optab_handler.
9516 * expr.c (expand_expr_real_2): Update calls to
9517 find_widening_optab_handler.
9518 * optabs.c (expand_widen_pattern_expr): Likewise.
9519 (expand_binop_directly): Take the insn_code as a parameter.
9520 (expand_binop): Only call find_widening_optab_handler for
9521 conversion optabs; use optab_handler otherwise. Update calls
9522 to find_widening_optab_handler and expand_binop_directly.
9523 Use convert_optab_handler instead of widening_optab_handler.
9524 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9525 find_widening_optab_handler and use scalar_mode rather than
9527 (convert_plusminus_to_widen): Likewise.
9529 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9530 Alan Hayward <alan.hayward@arm.com>
9531 David Sherwood <david.sherwood@arm.com>
9533 * machmode.h (fixed_size_mode): New class.
9534 * rtl.h (get_pool_mode): Return fixed_size_mode.
9535 * gengtype.c (main): Add fixed_size_mode.
9536 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9537 (get_raw_arg_mode): Likewise.
9538 * doc/tm.texi: Regenerate.
9539 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9540 * targhooks.c (default_get_reg_raw_mode): Likewise.
9541 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9542 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9543 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9544 (msp430_get_raw_result_mode): Likewise.
9545 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9546 * dbxout.c (dbxout_parms): Require fixed-size modes.
9547 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9548 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9549 * omp-low.c (lower_oacc_reductions): Likewise.
9550 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9551 (simplify_subreg): Update accordingly.
9552 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9553 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9554 that aren't fixed-size.
9555 (get_pool_mode): Return a fixed_size_mode.
9556 (output_constant_pool_2): Take a fixed_size_mode.
9558 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9559 Alan Hayward <alan.hayward@arm.com>
9560 David Sherwood <david.sherwood@arm.com>
9562 * doc/rtl.texi (vec_series): Document.
9563 (const): Say that the operand can be a vec_series.
9564 * rtl.def (VEC_SERIES): New rtx code.
9565 * rtl.h (const_vec_series_p_1): Declare.
9566 (const_vec_series_p): New function.
9567 * emit-rtl.h (gen_const_vec_series): Declare.
9568 (gen_vec_series): Likewise.
9569 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9570 (gen_vec_series): Likewise.
9571 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9572 * simplify-rtx.c (simplify_unary_operation): Handle negations
9574 (simplify_binary_operation_series): New function.
9575 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9576 (test_vector_ops_series): New function.
9577 (test_vector_ops): Call it.
9578 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9579 gen_const_vec_series.
9580 (altivec_lvsr): Likewise.
9581 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9583 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9584 Alan Hayward <alan.hayward@arm.com>
9585 David Sherwood <david.sherwood@arm.com>
9587 * doc/rtl.texi (const): Update description of address constants.
9588 Say that vector constants are allowed too.
9589 * common.md (E, F): Use CONSTANT_P instead of checking for
9591 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9592 checking for CONST_VECTOR.
9593 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9595 * expr.c (expand_expr_real_2): Check for vector modes instead
9596 of checking for CONST_VECTOR.
9597 * rtl.h (const_vec_p): New function.
9598 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9599 (unwrap_const_vec_duplicate): Handle them here too.
9601 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9602 David Malcolm <dmalcolm@redhat.com>
9603 Alan Hayward <alan.hayward@arm.com>
9604 David Sherwood <david.sherwood@arm.com>
9606 * rtl.h (vec_duplicate_p): New function.
9607 * selftest-rtl.c (assert_rtx_eq_at): New function.
9608 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9609 (assert_rtx_eq_at): Declare.
9610 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9611 * selftest-run-tests.c (selftest::run_tests): Call it.
9612 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9613 (simplify_unary_operation_1): Recursively handle vector duplicates.
9614 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9616 (simplify_subreg): Handle subregs of vector duplicates.
9617 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9618 (selftest::simplify_rtx_c_tests): New functions.
9620 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9621 Alan Hayward <alan.hayward@arm.com>
9622 David Sherwood <david.sherwood@arm.com>
9624 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9625 (gen_vec_duplicate): Likewise.
9626 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9628 (gen_const_vector): ...here.
9629 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9630 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9631 whose elements are all equal.
9632 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9633 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9634 (simplify_relational_operation): Likewise.
9635 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9637 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9638 (aarch64_expand_vector_init): Likewise.
9639 * config/arm/arm.c (neon_vdup_constant): Likewise.
9640 (neon_expand_vector_init): Likewise.
9641 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9642 (arm_block_set_unaligned_vect): Likewise.
9643 (arm_block_set_aligned_vect): Likewise.
9644 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9645 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9646 (expand_vec_perm_even_odd_pack): Likewise.
9647 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9648 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9649 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9650 gen_const_vec_duplicate.
9651 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9652 * config/mips/mips.c (mips_gen_const_int_vector): Use
9653 gen_const_vec_duplicate.
9654 (mips_expand_vector_init): Use CONST0_RTX.
9655 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9656 (define_split): Use gen_const_vec_duplicate.
9657 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9658 (define_split): Use gen_const_vec_duplicate.
9659 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9660 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9661 * config/spu/spu.c (spu_const): Likewise.
9663 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9664 Alan Hayward <alan.hayward@arm.com>
9665 David Sherwood <david.sherwood@arm.com>
9667 * combine.c (can_change_dest_mode): Reject changes in
9668 REGMODE_NATURAL_SIZE.
9670 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9672 * configure.ac (--enable-libssp): New.
9673 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9674 trying to determine target-specific default. Adjust indentation.
9675 * configure: Regenerated.
9676 * doc/install.texi (Configuration): Expand --disable-libssp
9679 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9681 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9684 2017-10-31 Martin Jambor <mjambor@suse.cz>
9687 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9689 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9691 * auto-profile.c (autofdo_source_profile::read): Use
9692 UNKNOWN_LOCATION rather than 0.
9693 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9694 (warning_at): ...this overload.
9695 (warning_at_rich_loc_n): Rename to...
9696 (warning_n): ...this overload.
9697 (error_at_rich_loc): Rename to...
9698 (error_at): ...this overload.
9699 (pedwarn_at_rich_loc): Rename to...
9700 (pedwarn): ...this overload.
9701 (permerror_at_rich_loc): Rename to...
9702 (permerror): ...this overload.
9703 (inform_at_rich_loc): Rename to...
9704 (inform): ...this overload.
9705 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9706 (diagnostic_n_impl_richloc): Rename to...
9707 (diagnostic_n_impl): ...this rich_location *-based decl.
9708 (inform_at_rich_loc): Rename to...
9709 (inform): ...this, and add an assertion.
9710 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9711 (warning_at_rich_loc): Rename to...
9712 (warning_at): ...this, and add an assertion.
9713 (warning_at_rich_loc_n): Rename to...
9714 (warning_n): ...this, and add an assertion.
9715 (warning_n): Update location_t-based implementation for removal of
9716 location_t-based diagnostic_n_impl.
9717 (pedwarn_at_rich_loc): Rename to...
9718 (pedwarn): ...this, and add an assertion.
9719 (permerror_at_rich_loc): Rename to...
9720 (permerror): ...this, and add an assertion.
9721 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9722 (error_at_rich_loc): Rename to...
9723 (error_at): ...this, and add an assertion.
9724 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9725 (driver::do_spec_on_infiles): Likewise.
9726 * substring-locations.c (format_warning_va): Update for renaming
9727 of inform_at_rich_loc.
9729 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9731 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9732 _Float<N>X built-in functions so that the variant without the
9733 "__builtin_" prefix is only enabled for the GNU C and Objective C
9734 languages when they are in non-strict ANSI/ISO mode.
9735 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9736 * target.def (floatn_builtin_p): Add a target hook to control
9737 whether _Float<N> and _Float<N>X built-in functions without the
9738 "__builtin_" prefix are enabled, and return true for C and
9739 Objective C in the default hook. Include langhooks.h in
9741 * targhooks.h (default_floatn_builtin_p): Likewise.
9742 * targhooks.c (default_floatn_builtin_p): Likewise.
9743 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9744 floatn_builtin_p target hook.
9745 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9747 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9748 Eric Botcazou <ebotcazou@adacore.com>
9750 PR rtl-optimization/81803
9751 * lra-constraints.c (curr_insn_transform): Also reload the whole
9752 register for a strict subreg no wider than a word if this is for
9753 a WORD_REGISTER_OPERATIONS target.
9755 2017-10-31 Jason Merrill <jason@redhat.com>
9757 * gdbinit.in: Skip over inlines from timevar.h.
9759 2017-10-31 Martin Liska <mliska@suse.cz>
9761 * doc/gcov.texi: Document new option.
9762 * gcov.c (print_usage): Likewise print it.
9763 (process_args): Support the argument.
9764 (format_count): New function.
9765 (format_gcov): Use the function.
9767 2017-10-31 Martin Liska <mliska@suse.cz>
9769 * gcov.c (struct name_map): do not use typedef.
9770 Define operator== and operator<.
9771 (name_search): Remove.
9772 (name_sort): Remove.
9773 (main): Do not allocate names.
9774 (process_file): Add vertical space.
9775 (generate_results): Use std::find.
9776 (release_structures): Do not release memory.
9777 (find_source): Use std::find.
9779 2017-10-31 Martin Liska <mliska@suse.cz>
9781 * gcov.c (struct line_info): Remove it's typedef.
9782 (line_info::line_info): Add proper ctor.
9783 (line_info::has_block): Do not use a typedef.
9784 (struct source_info): Do not use typedef.
9785 (circuit): Likewise.
9786 (get_cycles_count): Likewise.
9787 (output_intermediate_file): Iterate via vector iterator.
9788 (add_line_counts): Use std::vector methods.
9789 (accumulate_line_counts): Likewise.
9790 (output_lines): Likewise.
9792 2017-10-31 Martin Liska <mliska@suse.cz>
9794 * gcov.c (struct source_info): Remove typedef.
9795 (source_info::source_info): Add proper ctor.
9796 (accumulate_line_counts): Use struct, not it's typedef.
9797 (output_gcov_file): Likewise.
9798 (output_lines): Likewise.
9799 (main): Do not allocate an array.
9800 (output_intermediate_file): Use size of vector container.
9801 (process_file): Resize the vector.
9802 (generate_results): Do not preallocate, use newly added vector
9804 (release_structures): Do not release sources.
9805 (find_source): Use vector methods.
9806 (add_line_counts): Do not use typedef.
9808 2017-10-31 Martin Liska <mliska@suse.cz>
9810 * doc/gcov.texi: Document that.
9811 * gcov.c (add_line_counts): Mark lines with a non-executed
9813 (output_line_beginning): Handle such lines.
9814 (output_lines): Pass new argument.
9815 (output_intermediate_file): Print it in intermediate format.
9817 2017-10-31 Martin Liska <mliska@suse.cz>
9819 * color-macros.h: New file.
9820 * diagnostic-color.c: Factor out color related to macros to
9822 * doc/gcov.texi: Document -k option.
9823 * gcov.c (INCLUDE_STRING): Include string.h.
9824 (print_usage): Add -k option.
9825 (process_args): Parse it.
9826 (pad_count_string): New function.
9827 (output_line_beginning): Likewise.
9828 (DEFAULT_LINE_START): New macro.
9829 (output_lines): Support color output.
9831 2017-10-31 Martin Liska <mliska@suse.cz>
9833 PR gcov-profile/82633
9834 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9835 their interaction with GCOV infrastructure.
9836 * configure.ac: Add -fkeep-{inline,static}-functions to
9838 * configure: Regenerate.
9840 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
9843 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
9845 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
9848 * config/rs6000/rs6000.md (allocate_stack): Force update interval
9849 into a register if it does not fit into an immediate offset field.
9851 2017-10-31 Olivier Hainque <hainque@adacore.com>
9853 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9855 2017-10-31 Julia Koval <julia.koval@intel.com>
9857 * config.gcc: Add gfniintrin.h.
9858 * config/i386/gfniintrin.h: New.
9859 * config/i386/i386-builtin-types.def
9860 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
9861 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
9862 __builtin_ia32_vgf2p8affineinvqb_v32qi,
9863 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
9864 __builtin_ia32_vgf2p8affineinvqb_v16qi,
9865 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
9866 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
9867 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
9868 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
9869 V64QI_FTYPE_V64QI_V64QI_INT): New types.
9870 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
9871 * config/i386/immintrin.h: Include gfniintrin.h.
9872 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
9874 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9876 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9878 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9880 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9881 (arm_ashldi3_1bit): Remove pattern.
9882 (ashrdi3): Remove shift by 1 expansion.
9883 (arm_ashrdi3_1bit): Remove pattern.
9884 (lshrdi3): Remove shift by 1 expansion.
9885 (arm_lshrdi3_1bit): Remove pattern.
9886 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9887 cost of ashldi3 by 1.
9888 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9889 (<shift>di3_neon): Likewise.
9891 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9893 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9894 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9895 and (*aarch64_simd_mov<VQ:mode>).
9896 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9897 pattern alternative.
9898 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9899 attributes to match pattern alternative.
9901 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9903 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9904 * config/rs6000/emmintrin.h: New file.
9905 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9907 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9909 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9910 * (movdi_vfp_cortexa8): Remove pattern.
9912 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9914 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9916 (Specific, alpha*-dec-osf5.1): Remove.
9917 (Specific, mips-sgi-irix5): Remove.
9918 (Specific, mips-sgi-irix6): Remove.
9920 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9923 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9924 arguments to clear_bit_region_be.
9926 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9928 * gimplify.c: Include memmodel.h.
9930 2017-10-30 Martin Jambor <mjambor@suse.cz>
9932 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9933 condition whether loop should be executed at all.
9935 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9937 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9938 gimple folding of vec_madd() intrinsics.
9939 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9940 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9941 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9943 2017-10-30 Richard Biener <rguenther@suse.de>
9945 PR tree-optimization/82762
9947 2017-10-23 Richard Biener <rguenther@suse.de>
9949 PR tree-optimization/82129
9951 2017-08-01 Richard Biener <rguenther@suse.de>
9953 PR tree-optimization/81181
9954 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9955 (compute_antic): ... end of iteration here.
9957 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9959 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9960 -std=iso9899:2017 and -std=gnu17.
9961 * doc/standards.texi (C Language): Document C17 support.
9962 * doc/cpp.texi (Overview): Mention -std=c17.
9963 (Standard Predefined Macros): Document C11 and C17 values of
9964 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9965 * doc/extend.texi (Inline): Do not list individual options for
9966 standards newer than C99.
9967 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9969 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9972 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9974 * asan.c (asan_finish_file): Align asan globals array by shadow
9977 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9980 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9981 (struct store_immediate_info): Add bitregion_start and bitregion_end
9983 (store_immediate_info::store_immediate_info): Add brs and bre
9984 arguments and initialize bitregion_{start,end} from those.
9985 (struct merged_store_group): Add bitregion_start, bitregion_end,
9986 align_base and mask fields. Drop unnecessary struct keyword from
9987 struct store_immediate_info. Add do_merge method.
9988 (clear_bit_region_be): Use memset instead of loop storing zeros.
9989 (merged_store_group::do_merge): New method.
9990 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9991 stores as long as the surrounding bitregions have no gaps.
9992 (merged_store_group::merge_overlapping): Use do_merge.
9993 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9994 is byte aligned, rather than requiring that start and width are
9995 byte aligned. Drop unnecessary struct keyword from
9996 struct store_immediate_info. Allocate and populate also mask array.
9997 Make start of the arrays relative to bitregion_start rather than
9998 start and size them according to bitregion_{end,start} difference.
9999 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10000 struct store_immediate_info.
10001 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10002 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10003 struct keyword from struct store_immediate_info.
10004 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10005 between stores as long as the surrounding bitregions have no gaps.
10007 (struct split_store): Add orig non-static data member.
10008 (split_store::split_store): Initialize orig to false.
10009 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10010 if there is exactly a single original stmt. Change stmts argument
10011 to pointer from reference, if NULL, don't push anything to it. Add
10012 first argument, use it to optimize skipping over orig stmts that
10013 are known to be before bitpos already. Simplify.
10014 (split_group): Return unsigned int count how many stores are or
10015 would be needed rather than a bool. Add allow_unaligned argument.
10016 Change split_stores argument from reference to pointer, if NULL,
10017 only do a dry run computing how many stores would be produced.
10018 Rewritten algorithm to use both alignment and misalign if
10019 !allow_unaligned and handle bitfield stores with gaps.
10020 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10021 from bitregion_start instead of start. Compute allow_unaligned
10022 here, if true, do 2 split_group dry runs to compute which one
10023 produces fewer stores and prefer aligned if equal. Punt if
10024 new count is bigger or equal than original before emitting any
10025 statements, rather than during that. Remove no longer needed
10026 new_ssa_names tracking. Replace num_stmts with
10027 split_stores.length (). Use 32-bit stack allocated entries
10028 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10029 if possible. Handle bitfields with gaps.
10030 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10031 Compute bitregion_{start,end} for the stores and construct
10032 store_immediate_info with that. Formatting fixes.
10034 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10037 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10038 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10040 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10042 * gimplify.c: Include tm_p.h.
10044 * common.opt (gcoff): Re-add as ignored option.
10045 (gcoff1, gcoff2, gcoff3): Likewise.
10047 * Makefile.in (OBJS): Delete sdbout.o.
10048 (GTFILES): Delete $(srcdir)/sdbout.c.
10049 * debug.h: Delete sdb_debug_hooks.
10050 * final.c: Delete sdbout.h include.
10051 (final_scan_insn): Delete SDB_DEBUG check.
10052 (rest_of_clean_state): Likewise.
10053 * output.h: Delete sdb_begin_function_line.
10054 * sdbout.c: Delete.
10055 * sdbout.h: Delete.
10056 * toplev.c: Delete sdbout.h include.
10057 (process_options): Delete SDB_DEBUG check.
10058 * tree-core.h (tree_type_common): Delete pointer field of
10060 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10061 TYPE_SYMTAB_POINTER.
10062 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10063 (TYPE_SYMTAB_IS_POINTER): Delete.
10064 (TYPE_SYMTAB_IS_DIE): Renumber.
10065 * xcoffout.c: Refer to former sdbout.c file.
10066 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10068 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10069 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10070 (Debugging Options): Delete -gcoff.
10071 (-gstabs): Delete SDB reference.
10073 (-gcoff@var{level}): Delete.
10074 * doc/passes.texi (Debugging information output): Delete SDB and
10075 sdbout.c references.
10076 * doc/tm.texi: Regenerate.
10077 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10078 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10080 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10081 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10082 (SDB_DEBUGGING_INFO): Delete.
10083 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10084 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10085 * target.def (output_source_filename): Delete COFF reference.
10087 * common.opt (gcoff): Delete.
10088 (gxcoff+): Update Negative chain.
10089 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10091 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10092 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10093 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10094 and SDB references.
10095 (expand_function_start): Change sdb reference to past tense.
10096 (expand_function_end): Change sdb reference to past tense.
10097 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10098 * opts.c (debug_type_names): Delete coff entry.
10099 (common_handle_option): Delete OPT_gcoff case.
10100 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10102 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10103 * config/cris/cris.h: Delete SDB reference in comment.
10104 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10105 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10106 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10107 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10109 (ix86_expand_prologue): Likewise.
10110 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10111 * config/ia64/ia64.h: Likewise.
10112 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10113 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10115 * config/mmix/mmix.h: Likewise.
10116 * config/nds32/nds32.c: Likewise.
10117 * config/stormy/storym16.h: Likewise.
10118 * config/visium/visium.h: Likewise.
10119 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10121 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10123 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10124 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10125 FRAME_POINTER_REGNUM point at high end of local var area.
10127 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10129 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10130 Move comment around. Do not reset best_edge for a copiable
10131 destination if the copy would cause a partition change.
10132 (better_edge_p): Remove redundant check.
10134 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10136 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10138 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10140 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10141 for math functions that have _Float<N> and _Float<N>X variants.
10142 (mathfn_built_in_2): Add support for math functions that have
10143 _Float<N> and _Float<N>X variants.
10144 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10145 (expand_builtin_mathfn_ternary): Add support for fma with
10146 _Float<N> and _Float<N>X variants.
10147 (expand_builtin): Likewise.
10148 (fold_builtin_3): Likewise.
10149 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10150 create math function _Float<N> and _Float<N>X variants as external
10152 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10153 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10154 the __builtin_ prefix and if not strict ansi, without the prefix.
10155 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10156 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10157 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10158 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10159 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10160 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10161 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10162 function signatures for fma _Float<N> and _Float<N>X variants.
10163 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10164 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10165 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10166 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10167 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10168 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10169 * gencfn-macros.c (print_case_cfn): Add support for math functions
10170 that have _Float<N> and _Float<N>X variants.
10171 (print_define_operator_list): Likewise.
10172 (fltfn_suffixes): Likewise.
10174 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10175 for math functions that have _Float<N> and _Float<N>X variants.
10176 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10177 and _Float<N>X variants.
10178 (COPYSIGN): Likewise.
10181 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10182 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10184 (integer_valued_read_call_p): Likewise.
10185 * fold-const-call.c (fold_const_call_ss): Likewise.
10186 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10187 _Float<N> and _Float<N>X variants.
10188 (fold_const_call_ssss): Add support for fma _Float<N> and
10189 _Float<N>X variants.
10190 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10191 support for copysign and fma _Float<N> and _Float<N>X variants.
10192 (backprop::process_builtin_call_use): Likewise.
10193 * tree-call-cdce.c (can_test_argument_range); Add support for
10194 sqrt _Float<N> and _Float<N>X variants.
10195 (edom_only_function): Likewise.
10196 (get_no_error_domain): Likewise.
10197 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10198 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10199 copysign _Float<N> and _Float<N>X variants.
10200 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10201 handled by machine independent code.
10202 (FMAF128): Likewise.
10203 * doc/cpp.texi (Common Predefined Macros): Document defining
10204 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10205 fma _Float<N> and _Float<N>X variants.
10207 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10210 * config/i386/i386-modes.def (CCFPU): Remove definition.
10211 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10212 (ix86_cc_modes_compatible): Ditto.
10213 (ix86_expand_carry_flag_compare): Ditto.
10214 (ix86_expand_int_movcc): Ditto.
10215 (ix86_expand_int_addcc): Ditto.
10216 (ix86_reverse_condition): Ditto.
10217 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10218 Return true/false for unordered/ordered fp comparisons.
10219 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10220 (ix86_prepare_fp_compare_args): Update for rename.
10221 (ix86_expand_fp_compare): Update for rename. Generate unordered
10222 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10223 (ix86_expand_sse_compare_and_jump): Ditto.
10224 * config/i386/predicates.md (fcmov_comparison_operator):
10225 Remove CCFPU mode handling.
10226 (ix86_comparison_operator): Ditto.
10227 (ix86_carry_flag_operator): Ditto.
10228 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10229 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10230 (*cmpu<mode>_cc_i387): Ditto.
10231 (FPCMP): Remove mode iterator.
10232 (unord): Remove mode attribute.
10233 (unord_subst): New define_subst transformation
10234 (unord): New define_subst attribute.
10235 (unordered): Ditto.
10236 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10237 (*cmpi<unord>xf_i387): Ditto.
10238 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10239 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10240 using unord_subst transformation.
10241 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10242 (round_saeonly): Also handle CCFP mode.
10243 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10244 Remove UNSPEC_SAHF unspec handling.
10246 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10248 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10250 2017-10-27 Jeff Law <law@redhat.com>
10252 * gimple-ssa-sprintf.c: Include domwalk.h.
10253 (class sprintf_dom_walker): New class, derived from dom_walker.
10254 (sprintf_dom_walker::before_dom_children): New function.
10255 (struct call_info): Moved into sprintf_dom_walker class
10256 (compute_formath_length, handle_gimple_call): Likewise.
10257 (sprintf_length::execute): Call the dominator walker rather
10258 than walking the statements.
10260 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10261 gimple statement locations.
10262 (check_array_bounds): Corresponding changes. Get the statement's
10263 location directly from wi->stmt.
10265 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10268 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10270 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10272 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10273 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10275 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10278 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10279 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10280 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10281 maybe_get_pool_constant.
10282 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10285 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10287 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10288 2.26 caveat. Update gas and gld versions.
10289 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10292 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10294 * cgraph.h (set_malloc_flag): Declare.
10295 * cgraph.c (set_malloc_flag_1): New function.
10296 (set_malloc_flag): Likewise.
10297 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10298 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10300 (read_ipa_call_summary): Add support for reading is_return_callee.
10301 (write_ipa_call_summary): Stream is_return_callee.
10302 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10303 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10304 ipa-prop.h, ipa-fnsummary.h.
10305 (pure_const_names): Change to static.
10306 (malloc_state_e): Define.
10307 (malloc_state_names): Define.
10308 (funct_state_d): Add field malloc_state.
10309 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10310 (check_retval_uses): New function.
10311 (malloc_candidate_p): Likewise.
10312 (analyze_function): Add support for malloc attribute.
10313 (pure_const_write_summary): Stream malloc_state.
10314 (pure_const_read_summary): Add support for reading malloc_state.
10315 (dump_malloc_lattice): New function.
10316 (propagate_malloc): New function.
10317 (warn_function_malloc): New function.
10318 (ipa_pure_const::execute): Call propagate_malloc and
10319 ipa_free_fn_summary.
10320 (pass_local_pure_const::execute): Add support for malloc attribute.
10321 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10322 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10324 2017-10-27 Martin Liska <mliska@suse.cz>
10326 PR gcov-profile/82457
10327 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10328 for fork-like functions to be properly instrumented.
10330 2017-10-27 Richard Biener <rguenther@suse.de>
10332 PR middle-end/81659
10333 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10334 info when we redirected EH.
10336 2017-10-26 Michael Collison <michael.collison@arm.com>
10338 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10340 (<optab>_trunchf<GPI:mode>2: New pattern.
10341 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10342 * config/aarch64/iterators.md (wv): New mode attribute.
10343 (vf, VF): New mode attributes.
10344 (vgp, VGP): New mode attributes.
10345 (s): Update attribute with SImode and DImode prefixes.
10347 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10349 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10350 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10351 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10352 (nios2_option_overide): Initialize it. Don't allow R0-relative
10353 addressing with PIC.
10354 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10355 (nios2_symbolic_constant_p): Likewise.
10356 (nios2_legitimate_address_p): Likewise.
10357 (nios2_r0rel_section_name_p): New.
10358 (nios2_symbol_ref_in_r0rel_data_p): New.
10359 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10360 (r0rel_constant_p): New.
10361 (nios2_print_operand_address): Handle r0rel_constant_p.
10362 (nios2_cdx_narrow_form_p): Likewise.
10363 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10364 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10365 (Nios II Options): Document -mr0rel-sec.
10367 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10369 * config/nios2/nios2.c: Include xregex.h.
10370 (nios2_gprel_sec_regex): New.
10371 (nios2_option_overide): Initialize it. Don't allow GP-relative
10372 addressing with PIC.
10373 (nios2_small_section_name_p): Check for regex match.
10374 * config/nios2/nios2.opt (mgprel-sec=): New option.
10375 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10376 (Nios II Options): Document -mgprel-sec.
10378 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10380 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10382 2017-10-26 Tom de Vries <tom@codesourcery.com>
10384 PR tree-optimization/82707
10385 * gimple.c (gimple_copy): Fix unsharing of
10386 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10388 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10390 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10391 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10392 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10393 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10394 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10395 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10396 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10397 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10398 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10399 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10400 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10401 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10402 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10403 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10404 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10405 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10406 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10408 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10410 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10412 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10413 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10414 warning to rs6000.c. Remove the Undocumented flag, since it has
10416 (-mabi=ibmlongdouble): Likewise.
10417 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10418 already set, set the default format for long double.
10419 (rs6000_debug_reg_global): Print whether long double is IBM or
10421 (rs6000_option_override_internal): Rework setting long double
10422 format. Only warn if the user is changing the long double default
10423 and they did not use -Wno-psabi.
10424 * doc/invoke.texi (PowerPC options): Update the documentation for
10425 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10427 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10428 Alan Hayward <alan.hayward@arm.com>
10429 David Sherwood <david.sherwood@arm.com>
10431 * rtl.h (wider_subreg_mode): New function.
10432 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10433 rather than an unsigned int *.
10434 * ira-color.c (regno_max_ref_width): Replace with...
10435 (regno_max_ref_mode): ...this new variable.
10436 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10437 Use wider_subreg_mode.
10438 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10439 rather than an unsigned int *.
10440 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10441 (process_alt_operands): Likewise.
10442 (invariant_p): Likewise.
10443 * lra-spills.c (assign_mem_slot): Likewise.
10444 (add_pseudo_to_slot): Likewise.
10445 * lra.c (collect_non_operand_hard_regs): Likewise.
10446 (add_regs_to_insn_regno_info): Likewise.
10447 * reload1.c (regno_max_ref_width): Replace with...
10448 (regno_max_ref_mode): ...this new variable.
10449 (reload): Update accordingly. Update call to
10450 ira_sort_regnos_for_alter_reg.
10451 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10452 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10453 (scan_paradoxical_subregs): Likewise.
10455 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10457 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10458 (aarch64_frame): Add emit_frame_chain boolean.
10459 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10460 Move eh_return case to aarch64_layout_frame.
10461 (aarch64_layout_frame): Initialize emit_frame_chain.
10462 (aarch64_expand_prologue): Use emit_frame_chain.
10464 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10466 * config/aarch64/aarch64.c (aarch64_layout_frame):
10467 Ensure LR is always stored at the bottom of the callee-saves.
10468 Remove rarely used frame layout which saves callee-saves at top of
10469 frame, so the store of LR can be used as a valid probe in all cases.
10471 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10473 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10474 Improve unaligned TImode/TFmode base/offset split.
10476 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10477 Alan Hayward <alan.hayward@arm.com>
10478 David Sherwood <david.sherwood@arm.com>
10480 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10481 * combine.c (find_single_use_1): Likewise.
10482 (expand_field_assignment): Likewise.
10483 (move_deaths): Likewise.
10484 * lra-constraints.c (simplify_operand_subreg): Likewise.
10485 (curr_insn_transform): Likewise.
10486 * lra.c (collect_non_operand_hard_regs): Likewise.
10487 (add_regs_to_insn_regno_info): Likewise.
10488 * rtlanal.c (reg_referenced_p): Likewise.
10489 (covers_regno_no_parallel_p): Likewise.
10491 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10493 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10494 Don't print any bits outside the precision of the value.
10495 * wide-int.cc (test_printing): Add some new tests.
10497 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10499 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10500 supports -xbrace_comment option.
10501 * configure: Regenerate.
10502 * config.in: Regenerate.
10503 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10504 (ASM_CPU_SPEC): Use it.
10506 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10508 * target.def (static_rtx_alignment): New hook.
10509 * targhooks.h (default_static_rtx_alignment): Declare.
10510 * targhooks.c (default_static_rtx_alignment): New function.
10511 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10512 * doc/tm.texi: Regenerate.
10513 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10514 instead of targetm.constant_alignment. Remove call to
10515 set_mem_attributes.
10516 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10517 (cris_preferred_mininum_alignment): New function, split out from...
10518 (cris_constant_alignment): ...here.
10519 (cris_static_rtx_alignment): New function.
10520 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10522 (ix86_constant_alignment): ...here.
10523 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10524 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10525 (mmix_static_rtx_alignment): New function.
10526 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10527 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10529 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10532 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10533 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10535 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10537 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10538 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10539 mark as a sign-extending load.
10540 (local_pic_load_u): Define.
10542 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10544 PR middle-end/82062
10545 * fold-const.c (operand_equal_for_comparison_p): Also return true
10546 if ARG0 is a simple variant of ARG1 with narrower precision.
10547 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10549 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10551 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10553 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10554 scatter_static, scatter_per_elt.
10555 * x86-tune-costs.h: Add new cost entries.
10557 2017-10-25 Richard Biener <rguenther@suse.de>
10559 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10560 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10561 class pass_fre): Move to ...
10562 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10563 class pass_fre): ... here and adjust for statistics.
10565 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10568 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10569 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10570 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10571 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10574 2017-10-25 Richard Biener <rguenther@suse.de>
10576 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10577 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10578 eliminate_push_avail, eliminate_insert): Move inside...
10579 (class eliminate_dom_walker): ... this class in preparation
10581 (fini_eliminate): Remove by merging with ...
10582 (eliminate): ... this function. Adjust for class changes.
10583 (pass_pre::execute): Remove fini_eliminate call.
10584 (pass_fre::execute): Likewise.
10586 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10589 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10590 (VPERMI2, VPERMI2I): New mode iterators.
10591 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10592 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10594 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10595 mode iterator. Remove 3 old define_insn patterns.
10596 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10597 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10598 VPERMI2 mode iterator, remove the other two expanders.
10599 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10600 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10601 instructions, remove the other two patterns.
10602 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10603 mode iterator, remove the other two patterns.
10604 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10605 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10606 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10607 and adjust argument order accordingly.
10608 (ix86_expand_vec_perm): Adjust caller.
10609 (expand_vec_perm_1): Likewise.
10610 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10611 (expand_vec_perm_vpermt2_vpshub2): ... this.
10612 (ix86_expand_vec_perm_const_1): Adjust caller.
10613 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10616 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10617 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10618 (vec_shl_<mode>): Remove unused expander.
10619 (avx512bw_<shift_insn><mode>3): New define_insn.
10620 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10621 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10623 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10626 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10629 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10631 PR rtl-optimization/82396
10632 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10633 (autopref_multipass_init): Simplify initialization.
10634 (autopref_rank_data): Simplify sort order.
10635 * gcc/sched-int.h (autopref_multipass_data_): Remove
10636 multi_mem_insn_p, min_offset and max_offset.
10638 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10640 PR middle-end/60580
10641 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10642 Check special value of flag_omit_frame_pointer.
10643 (aarch64_can_eliminate): Likewise.
10644 (aarch64_override_options_after_change_1): Simplify handling of
10645 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10647 2017-10-24 Richard Biener <rguenther@suse.de>
10649 PR tree-optimization/82697
10650 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10651 zero for conditional load and unconditional store.
10653 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10655 * doc/install.texi: Document bootstrap-cet.
10657 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10660 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10661 ENDBR instruction at function entrance if function is only
10664 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10667 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10668 patterns to better describe from which operation the CF is computed.
10669 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10670 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10671 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10672 is 0, use _0 suffixed expanders instead of emitting a comparison
10675 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10677 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10678 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10680 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10682 PR middle-end/82569
10683 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10684 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10685 * loop-iv.c (iv_get_reaching_def): Likewise.
10686 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10687 variable is promoted and the partition contains undefined values.
10689 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10691 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10693 (nios2_address_cost): Define.
10694 (nios2_legitimize_address): Recognize (exp + constant) directly.
10695 (TARGET_ADDRESS_COST): Define.
10697 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10699 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10700 (nios2_symbolic_memory_operand_p): Declare.
10701 (nios2_split_large_constant): Declare.
10702 (nios2_split_symbolic_memory_operand): Declare.
10703 * config/nios2/nios2.c: Adjust includes.
10704 (nios2_symbolic_constant_allowed): New.
10705 (nios2_symbolic_constant_p): New.
10706 (nios2_plus_symbolic_constant_p): New.
10707 (nios2_valid_addr_expr_p): Recognize addresses involving
10708 symbolic constants.
10709 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10710 (nios2_symbolic_memory_operand_p): New.
10711 (nios2_large_constant_p): New.
10712 (nios2_split_large_constant): New.
10713 (nios2_split_plus_large_constant): New.
10714 (nios2_split_symbolic_memory_operand): New.
10715 (nios2_legitimize_address): Code refactoring. Handle addresses
10716 involving symbolic constants.
10717 (nios2_emit_move_sequence): Likewise.
10718 (nios2_print_operand): Improve error output.
10719 (nios2_print_operand_address): Handle LO_SUM.
10720 (nios2_cdx_narrow_form_p): Likewise.
10721 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10722 operands involving symbolic constants.
10723 (movhi_internal, movsi_internal): Likewise.
10724 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10725 (extendhisi2, extendqi<mode>2): Likewise.
10727 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10729 * tree-pass.h (PROP_rtl_split_insns): Define.
10730 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10732 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10734 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10736 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10739 * target.def (const_not_ok_for_debug_p): Default to
10740 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10741 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10742 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10743 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10744 which targetm.const_not_ok_for_debug_p returned true.
10745 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10747 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10749 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10750 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10751 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10753 (ix86_const_not_ok_for_debug_p): New function.
10754 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10755 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10757 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10760 * system.h: Conditionally include "unique-ptr.h" if
10761 INCLUDE_UNIQUE_PTR is defined.
10762 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10763 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10765 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10767 * config/rl78/rl78.md: New define_expand "subdi3".
10769 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10772 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10773 DF_REF_INSN if DF_REF_INSN_INFO is false.
10775 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10777 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10778 xmm_move instead of sse_move.
10779 (sse_store_index): New function.
10780 (ix86_register_move_cost): Be more sensible about mismatch stall;
10781 model AVX moves correctly; make difference between sse->integer and
10783 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10784 moves; make difference between SSE and AVX.
10785 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10786 and zmm_move. Increase size of sse load and store tables;
10787 add unaligned load and store tables; add ssemmx_to_integer.
10788 * x86-tune-costs.h: Update all entries according to real
10789 move latencies from Agner Fog's manual and chip documentation.
10791 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10794 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
10795 * config/i386/constraints.md (Wf): New constraint.
10796 * config/i386/i386.md (UNSPEC_SBB): New unspec.
10797 (cmp<dwi>_doubleword): Removed.
10798 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
10799 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
10800 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
10801 expand with cmp<dwi>_doubleword. For LTU and GEU use
10802 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
10804 * common.opt (gcolumn-info): Enable by default.
10805 * doc/invoke.texi (gcolumn-info): Document new default.
10807 2017-10-23 Richard Biener <rguenther@suse.de>
10809 PR tree-optimization/82672
10810 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10811 Fold the stmt if we propagated into it.
10813 2017-10-23 Richard Biener <rguenther@suse.de>
10815 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
10816 (bitmap_remove_expr_from_set): ... this. All callers call this
10817 for non-constant values.
10818 (bitmap_set_subtract): Rename to...
10819 (bitmap_set_subtract_expressions): ... this. Adjust and
10821 (bitmap_set_contains_value): Remove superfluous check.
10822 (bitmap_set_replace_value): Inline into single caller ...
10823 (bitmap_value_replace_in_set): ... here and simplify.
10824 (dependent_clean): Merge into ...
10825 (clean): ... this using an overload. Adjust.
10826 (prune_clobbered_mems): Adjust.
10827 (compute_antic_aux): Likewise.
10828 (compute_partial_antic_aux): Likewise.
10830 2017-10-23 Richard Biener <rguenther@suse.de>
10832 PR tree-optimization/82129
10834 2017-08-01 Richard Biener <rguenther@suse.de>
10836 PR tree-optimization/81181
10837 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10838 (compute_antic): ... end of iteration here.
10840 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10842 * target.def (starting_frame_offset): New hook.
10843 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
10844 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
10845 * doc/tm.texi.in: Regenerate.
10846 * hooks.h (hook_hwi_void_0): Declare.
10847 * hooks.c (hook_hwi_void_0): New function.
10848 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
10849 STARTING_FRAME_OFFSET.
10850 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10851 * reload1.c (reload): Likewise.
10852 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10853 instead of STARTING_FRAME_OFFSET.
10854 * function.c (try_fit_stack_local): Likewise.
10855 (assign_stack_local_1): Likewise
10856 (instantiate_virtual_regs): Likewise.
10857 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
10858 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
10859 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
10860 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
10861 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
10862 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
10863 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
10864 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
10865 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
10866 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
10867 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
10868 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
10869 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
10870 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
10871 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
10872 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
10873 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
10874 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10875 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10876 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10877 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10878 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10879 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10880 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10881 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10882 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10883 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10884 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10885 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10886 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10887 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10888 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10889 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10890 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10891 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10892 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10893 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10894 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10895 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10896 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10897 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10898 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10899 return a HOST_WIDE_INT.
10900 (avr_builtin_setjmp_frame_value): Use it instead of
10901 STARTING_FRAME_OFFSET.
10902 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10903 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10904 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10906 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10907 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10908 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10909 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10910 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10911 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10912 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10913 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10914 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10915 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10916 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10917 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10919 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10920 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10921 * config/mips/mips.c (mips_compute_frame_info): Refer to
10922 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10923 (mips_starting_frame_offset): New function.
10924 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10925 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10926 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10927 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10928 and return a HOST_WIDE_INT.
10929 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10930 (mmix_initial_elimination_offset): Refer to
10931 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10932 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10933 * config/pa/pa.c (pa_starting_frame_offset): New function.
10934 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10935 (pa_expand_prologue): Likewise.
10936 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10937 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10938 !FRAME_GROWS_DOWNWARD handling to...
10939 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10940 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10941 !FRAME_GROWS_DOWNWARD handling to...
10942 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10943 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10944 !FRAME_GROWS_DOWNWARD handling to...
10945 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10946 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10948 (rs6000_starting_frame_offset): New function.
10949 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10950 !FRAME_GROWS_DOWNWARD handling to...
10951 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10952 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10953 !FRAME_GROWS_DOWNWARD handling to...
10954 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10955 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10956 !FRAME_GROWS_DOWNWARD handling to...
10957 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10958 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10959 (rs6000_starting_frame_offset): New function.
10960 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10961 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10962 * config/vax/vax.c (vax_starting_frame_offset): New function.
10963 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10964 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10965 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10966 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10967 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10968 * system.h (STARTING_FRAME_OFFSET): Poison.
10970 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10972 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10973 SCALAR_TYPE_MODE instead of TYPE_MODE.
10975 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10976 Alan Hayward <alan.hayward@arm.com>
10977 David Sherwood <david.sherwood@arm.com>
10979 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10981 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10982 Alan Hayward <alan.hayward@arm.com>
10983 David Sherwood <david.sherwood@arm.com>
10985 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10987 2017-10-23 Richard Biener <rguenther@suse.de>
10989 PR tree-optimization/82129
10990 * tree-ssa-pre.c (bitmap_set_and): Remove.
10991 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10992 canonicalizing expressions in the set to those with lowest
10993 ID rather than taking that from the first edge.
10995 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10997 * combine.c (rtx_equal_for_field_assignment_p): Use
10998 byte_lowpart_offset.
11000 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11001 Alan Hayward <alan.hayward@arm.com>
11002 David Sherwood <david.sherwood@arm.com>
11004 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11005 to a SUBREG_PROMOTED_VAR.
11007 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11008 Alan Hayward <alan.hayward@arm.com>
11009 David Sherwood <david.sherwood@arm.com>
11011 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11012 (expand_debug_source_expr): Likewise.
11013 * combine.c (combine_simplify_rtx): Likewise.
11014 * cse.c (fold_rtx): Likewise.
11015 * optabs.c (expand_float): Likewise.
11016 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11017 (simplify_binary_operation_1): Likewise.
11019 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11020 Alan Hayward <alan.hayward@arm.com>
11021 David Sherwood <david.sherwood@arm.com>
11023 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11024 (record_promoted_value): Likewise.
11025 * expr.c (expand_expr_real_2): Likewise.
11026 * ree.c (update_reg_equal_equiv_notes): Likewise.
11027 (combine_set_extension): Likewise.
11028 * rtlanal.c (low_bitmask_len): Likewise.
11029 * simplify-rtx.c (neg_const_int): Likewise.
11030 (simplify_binary_operation_1): Likewise.
11032 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11033 Alan Hayward <alan.hayward@arm.com>
11034 David Sherwood <david.sherwood@arm.com>
11036 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11037 * regcprop.c (maybe_mode_change): Likewise.
11038 * reload1.c (alter_reg): Likewise.
11040 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11042 * inchash.h (inchash::hash::add_wide_int): New function.
11043 * lto-streamer-out.c (hash_tree): Use it.
11045 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11047 * inchash.h (inchash::hash::add_wide_int): Rename to...
11048 (inchash::hash::add_hwi): ...this.
11049 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11050 (polymorphic_call_target_hasher::hash): Likewise.
11051 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11052 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11053 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11054 * lto-streamer-out.c (hash_tree): Likewise.
11055 * optc-save-gen.awk: Likewise.
11056 * tree.c (add_expr): Likewise.
11058 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11061 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11062 for ordered inequality comparisons even with TARGET_IEEE_FP.
11064 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11067 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11068 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11069 Expand with cmp<dwi>_doubleword.
11071 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11073 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11074 List CET intrinsics.
11075 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11076 specific to -fcf-protection option.
11078 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11080 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11081 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11082 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11083 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11084 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11085 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11086 (extra_objs): Add cet.o for Linux/x86 targets.
11087 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11088 * config/i386/cet.c: New file.
11089 * config/i386/cetintrin.h: Likewise.
11090 * config/i386/t-cet: Likewise.
11091 * config/i386/cpuid.h (bit_SHSTK): New.
11092 (bit_IBT): Likewise.
11093 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11094 pass IBT and SHSTK bits.
11095 * config/i386/i386-builtin-types.def
11096 (VOID_FTYPE_UNSIGNED_PVOID): New.
11097 (VOID_FTYPE_UINT64_PVOID): Likewise.
11098 * config/i386/i386-builtin.def: Add CET intrinsics.
11099 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11100 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11101 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11102 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11104 * config/i386/i386.c (rest_of_insert_endbranch): New.
11105 (pass_data_insert_endbranch): Likewise.
11106 (pass_insert_endbranch): Likewise.
11107 (make_pass_insert_endbranch): Likewise.
11108 (ix86_notrack_prefixed_insn_p): Likewise.
11109 (ix86_target_string): Add -mibt, -mshstk flags.
11110 (ix86_option_override_internal): Add flag_cf_protection
11112 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11113 (ix86_print_operand): Add 'notrack' prefix output.
11114 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11115 (ix86_expand_builtin): Expand CET intrinsics.
11116 (x86_output_mi_thunk): Add 'endbranch' instruction.
11117 * config/i386/i386.h (TARGET_IBT): New.
11118 (TARGET_IBT_P): Likewise.
11119 (TARGET_SHSTK): Likewise.
11120 (TARGET_SHSTK_P): Likewise.
11121 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11122 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11123 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11124 (builtin_setjmp_setup): New pattern.
11125 (builtin_longjmp): Likewise.
11126 (rdssp<mode>): Likewise.
11127 (incssp<mode>): Likewise.
11128 (saveprevssp): Likewise.
11129 (rstorssp): Likewise.
11130 (wrss<mode>): Likewise.
11131 (wruss<mode>): Likewise.
11132 (setssbsy): Likewise.
11133 (clrssbsy): Likewise.
11134 (nop_endbr): Likewise.
11135 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11137 * config/i386/immintrin.h: Include <cetintrin.h>.
11138 * config/i386/linux-common.h
11139 (file_end_indicate_exec_stack_and_cet): New prototype.
11140 (TARGET_ASM_FILE_END): New.
11142 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11144 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11145 latencies instead of having separate table; make difference between
11146 integer and float costs.
11147 * i386.h (processor_costs): Remove scalar_stmt_cost,
11148 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11149 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11151 * x86-tune-costs.h: Remove entries which has been removed in
11152 procesor_costs from all tables; make cond_taken_branch_cost
11153 and cond_not_taken_branch_cost COST_N_INSNS based.
11155 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11157 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11159 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11161 * config/i386/i386.md (isa): Remove fma_avx512f.
11162 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11163 <avx512>_fmadd_<mode>_mask3<round_name>,
11164 <avx512>_fmsub_<mode>_mask<round_name>,
11165 <avx512>_fmsub_<mode>_mask3<round_name>,
11166 <avx512>_fnmadd_<mode>_mask<round_name>,
11167 <avx512>_fnmadd_<mode>_mask3<round_name>,
11168 <avx512>_fnmsub_<mode>_mask<round_name>,
11169 <avx512>_fnmsub_<mode>_mask3<round_name>,
11170 <avx512>_fmaddsub_<mode>_mask<round_name>,
11171 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11172 <avx512>_fmsubadd_<mode>_mask<round_name>,
11173 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11174 (*vec_widen_umult_even_v16si<mask_name>,
11175 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11176 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11178 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11180 * extend.texi: Add 'nocf_check' documentation.
11181 * gimple.texi: Add second parameter to
11182 gimple_build_call_from_tree.
11183 * invoke.texi: Add -fcf-protection documentation.
11184 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11186 2017-10-20 Richard Biener <rguenther@suse.de>
11188 PR tree-optimization/82473
11189 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11190 the largest input type.
11192 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11194 * c-attribs.c (handle_nocf_check_attribute): New function.
11195 (c_common_attribute_table): Add 'nocf_check' handling.
11196 * gimple-parser.c: Add second argument NULL to
11197 gimple_build_call_from_tree.
11198 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11199 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11201 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11202 * common.opt: Add fcf-protection flag.
11203 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11204 * flag-types.h: Add enum cf_protection_level.
11205 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11206 Add 'nocf_check' attribute propagation to gimple call.
11207 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11208 (gimple_build_call_from_tree): Update prototype.
11209 (gimple_call_nocf_check_p): New function.
11210 (gimple_call_set_nocf_check): Likewise.
11211 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11212 * ipa-icf.c: Add nocf_check attribute in statement hash.
11213 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11214 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11215 * toplev.c (process_options): Add flag_cf_protection handling.
11217 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11219 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11221 2017-10-20 Richard Biener <rguenther@suse.de>
11223 PR tree-optimization/82603
11224 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11225 remove false predicated stores.
11227 2017-10-20 Richard Biener <rguenther@suse.de>
11229 * graphite-isl-ast-to-gimple.c
11230 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11231 Remove return value and simplify, dump copied stmt after lhs
11233 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11234 Reduce dump verbosity.
11235 (gsi_insert_earliest): Likewise.
11236 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11237 * graphite.c (print_global_statistics): Adjust dumping.
11238 (print_graphite_scop_statistics): Likewise.
11239 (print_graphite_statistics): Do not dump loops here.
11240 (graphite_transform_loops): But here.
11242 2017-10-20 Nicolas Roche <roche@adacore.com>
11244 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11245 * configure: Regenerate.
11247 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11250 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11251 functions when optimizing replace GIMPLE_RETURN stmts with
11252 calls to __builtin_unreachable ().
11255 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11256 for -fsanitize=thread link of executables.
11257 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11258 link of executables.
11261 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11262 New mode iterators.
11263 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11264 define_insns for logical vector shifts to use VI248_AVX512BW
11265 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11266 condition, useless isa and prefix attributes. Change the first
11267 2 of these define_insns to ...
11268 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11269 define_insn for avx512vl.
11270 (<shift_insn><mode>3): ... and this, new define_insn without
11271 masking for non-avx512vl.
11274 * config/i386/sse.md (*andnot<mode>3,
11275 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11276 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11277 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11278 not applied use empty suffix even for TARGET_AVX512VL.
11279 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11280 is applied, supply evex,evex or evex,evex,evex instead of just
11283 2017-10-20 Julia Koval <julia.koval@intel.com>
11285 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11286 (OPTION_MASK_ISA_GFNI_UNSET): New.
11287 (ix86_handle_option): Handle OPT_mgfni.
11288 * config/i386/cpuid.h (bit_GFNI): New.
11289 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11290 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11291 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11292 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11293 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11294 * config/i386/i386.opt: Add mgfni.
11296 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11298 * config/msp430/msp430.c (msp430_option_override): Disable
11299 -fdelete-null-pointer-checks.
11300 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11302 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11304 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11305 of x87 and SSE instructions.
11307 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11309 * asan.c (create_cond_insert_point): Do not update edge count.
11310 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11311 (afdo_propagate_circuit): Likewise.
11312 (afdo_calculate_branch_prob): Likewise.
11313 (afdo_annotate_cfg): Likewise.
11314 * basic-block.h (struct edge_def): Remove count.
11315 (edge_def::count): New accessor.
11316 * bb-reorder.c (rotate_loop): Update.
11317 (find_traces_1_round): Update.
11318 (connect_traces): Update.
11319 (sanitize_hot_paths): Update.
11320 * cfg.c (unchecked_make_edge): Update.
11321 (make_single_succ_edge): Update.
11322 (check_bb_profile): Update.
11323 (dump_edge_info): Update.
11324 (update_bb_profile_for_threading): Update.
11325 (scale_bbs_frequencies_int): Update.
11326 (scale_bbs_frequencies_gcov_type): Update.
11327 (scale_bbs_frequencies_profile_count): Update.
11328 (scale_bbs_frequencies): Update.
11329 * cfganal.c (connect_infinite_loops_to_exit): Update.
11330 * cfgbuild.c (compute_outgoing_frequencies): Update.
11331 (find_many_sub_basic_blocks): Update.
11332 * cfgcleanup.c (try_forward_edges): Update.
11333 (try_crossjump_to_edge): Update
11334 * cfgexpand.c (expand_gimple_cond): Update
11335 (expand_gimple_tailcall): Update
11336 (construct_exit_block): Update
11337 * cfghooks.c (verify_flow_info): Update
11338 (redirect_edge_succ_nodup): Update
11339 (split_edge): Update
11340 (make_forwarder_block): Update
11341 (duplicate_block): Update
11342 (account_profile_record): Update
11343 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11344 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11345 * cfgloopmanip.c (scale_loop_profile): Update.
11347 (lv_adjust_loop_entry_edge): Update.
11348 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11349 (force_nonfallthru_and_redirect): Update.
11350 (purge_dead_edges): Update.
11351 (rtl_flow_call_edges_add): Update.
11352 * cgraphunit.c (init_lowered_empty_function): Update.
11353 (cgraph_node::expand_thunk): Update.
11354 * gimple-pretty-print.c (dump_probability): Update.
11355 (dump_edge_probability): Update.
11356 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11357 * haifa-sched.c (sched_create_recovery_edges): Update.
11358 * hsa-gen.c (convert_switch_statements): Update.
11359 * ifcvt.c (dead_or_predicable): Update.
11360 * ipa-inline-transform.c (inline_transform): Update.
11361 * ipa-split.c (split_function): Update.
11362 * ipa-utils.c (ipa_merge_profiles): Update.
11363 * loop-doloop.c (add_test): Update.
11364 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11365 * lto-streamer-in.c (input_cfg): Update.
11366 (input_function): Update.
11367 * lto-streamer-out.c (output_cfg): Update.
11368 * modulo-sched.c (sms_schedule): Update.
11369 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11370 * predict.c (maybe_hot_edge_p): Update.
11371 (unlikely_executed_edge_p): Update.
11372 (probably_never_executed_edge_p): Update.
11373 (dump_prediction): Update.
11374 (drop_profile): Update.
11375 (propagate_unlikely_bbs_forward): Update.
11376 (determine_unlikely_bbs): Update.
11377 (force_edge_cold): Update.
11378 * profile.c (compute_branch_probabilities): Update.
11379 * reg-stack.c (better_edge): Update.
11380 * shrink-wrap.c (handle_simple_exit): Update.
11381 * tracer.c (better_p): Update.
11382 * trans-mem.c (expand_transaction): Update.
11383 (split_bb_make_tm_edge): Update.
11384 * tree-call-cdce.c: Update.
11385 * tree-cfg.c (gimple_find_sub_bbs): Update.
11386 (gimple_split_edge): Update.
11387 (gimple_duplicate_sese_region): Update.
11388 (gimple_duplicate_sese_tail): Update.
11389 (gimple_flow_call_edges_add): Update.
11390 (insert_cond_bb): Update.
11391 (execute_fixup_cfg): Update.
11392 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11393 * tree-complex.c (expand_complex_div_wide): Update.
11394 * tree-eh.c (lower_resx): Update.
11395 (unsplit_eh): Update.
11396 (cleanup_empty_eh_move_lp): Update.
11397 * tree-inline.c (copy_edges_for_bb): Update.
11398 (freqs_to_counts): Update.
11399 (copy_cfg_body): Update.
11400 * tree-ssa-dce.c (remove_dead_stmt): Update.
11401 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11402 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11403 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11404 (unloop_loops): Update.
11405 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11406 * tree-ssa-loop-split.c (connect_loops): Update.
11407 (split_loop): Update.
11408 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11409 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11410 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11411 * tree-ssa-reassoc.c (branch_fixup): Update.
11412 * tree-ssa-tail-merge.c (replace_block_by): Update.
11413 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11414 (compute_path_counts): Update.
11415 (update_profile): Update.
11416 (recompute_probabilities): Update.
11417 (update_joiner_offpath_counts): Update.
11418 (estimated_freqs_path): Update.
11419 (freqs_to_counts_path): Update.
11420 (clear_counts_path): Update.
11421 (ssa_fix_duplicate_block_edges): Update.
11422 (duplicate_thread_path): Update.
11423 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11424 (case_bit_test_cmp): Update.
11425 (collect_switch_conv_info): Update.
11426 (gen_inbound_check): Update.
11427 (do_jump_if_equal): Update.
11428 (emit_cmp_and_jump_insns): Update.
11429 * tree-tailcall.c (decrease_profile): Update.
11430 (eliminate_tail_call): Update.
11431 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11432 (vect_do_peeling): Update.
11433 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11434 * ubsan.c (ubsan_expand_null_ifn): Update.
11435 (ubsan_expand_ptr_ifn): Update.
11436 * value-prof.c (gimple_divmod_fixed_value): Update.
11437 (gimple_mod_pow2): Update.
11438 (gimple_mod_subtract): Update.
11439 (gimple_ic): Update.
11440 (gimple_stringop_fixed_value): Update.
11442 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11445 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11447 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11449 PR rtl-optimization/82395
11450 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11451 based on non_spilled_static_chain_regno_p.
11453 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11455 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11456 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11457 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11459 2017-10-19 Martin Sebor <msebor@redhat.com>
11461 PR tree-optimization/82596
11462 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11464 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11466 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11467 (get_mem_refs_of_builtin_call): Likewise.
11468 * builtins.c (expand_builtin_apply): Adjust call to
11469 allocate_dynamic_stack_space.
11470 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11471 the third argument to allocate_dynamic_stack_space, otherwise -1.
11472 (expand_builtin): Deal with all alloca variants.
11473 (is_inexpensive_builtin): Likewise.
11474 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11475 * calls.c (special_function_p): Deal with all alloca variants.
11476 (initialize_argument_information): Adjust call to
11477 allocate_dynamic_stack_space.
11478 (expand_call): Likewise.
11479 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11480 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11481 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11482 use it for the stack usage computation.
11483 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11484 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11485 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11486 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11487 (in_loop_p): Remove first argument and useless check.
11488 (pass_walloca::execute): Remove useless test and adjust call to above.
11489 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11490 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11491 (gimplify_call_expr): Deal with all alloca variants.
11492 * hsa-gen.c (gen_hsa_alloca): Likewise.
11493 (gen_hsa_insns_for_call): Likewise.
11494 * ipa-pure-const.c (special_builtin_state): Likewise.
11495 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11496 * tree-object-size.c (alloc_object_size): Likewise.
11497 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11498 (call_may_clobber_ref_p_1): Likewise.
11499 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11500 (ccp_fold_stmt): Likewise.
11501 (optimize_stack_restore): Likewise.
11502 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11503 (mark_all_reaching_defs_necessary_1): Likewise.
11504 (propagate_necessity): Likewise.
11505 (eliminate_unnecessary_stmts): Likewise.
11506 * tree.c (build_common_builtin_nodes): Build
11507 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11508 (build_alloca_call_expr): New function.
11509 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11510 (CASE_BUILT_IN_ALLOCA): Likewise.
11511 (build_alloca_call_expr): Declare.
11512 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11514 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11517 * dwarf2out.c (new_die_raw): New static inline function.
11518 (new_die): Use it to create the DIE.
11519 (add_AT_external_die_ref): Likewise.
11520 (clone_die): Likewise.
11521 (clone_as_declaration): Likewise.
11522 (dwarf2out_vms_debug_main_pointer): Likewise.
11523 (base_type_die): Likewise. Remove early return for corner cases.
11524 Do not call add_pubtype on the DIE here.
11525 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11526 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11527 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11528 native order exists for base types, attach the DIE manually and call
11529 add_pubtype on it. Do not equate a reverse order DIE to the type.
11531 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11533 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11534 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11536 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11538 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11540 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11541 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11542 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11543 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11544 * opts.c (sanitizer_opts): Add builtin.
11545 * ubsan.c (instrument_builtin): New function.
11546 (pass_ubsan::execute): Call it.
11547 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11548 * doc/invoke.texi: Document -fsanitize=builtin.
11550 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11551 builtins, store max (log2 (align), 0) into uchar field instead of
11552 align into uptr field.
11553 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11554 store uchar 0 field instead of uptr 0 field.
11555 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11556 instead of passing one address of struct with 2 locations pass
11557 two addresses of structs with 1 location each.
11558 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11559 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11560 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11561 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11562 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11563 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11564 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11565 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11567 2017-10-19 Martin Liska <mliska@suse.cz>
11570 * file-find.c (remove_prefix): Remove.
11571 * file-find.h (remove_prefix): Likewise.
11572 * gcc-ar.c: Remove smartness of lookup.
11574 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11576 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11577 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11578 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11580 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11583 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11584 (setcc + and to xor + setcc): New peephole2.
11586 2017-10-19 Tom de Vries <tom@codesourcery.com>
11588 * doc/sourcebuild.texi (Test Directives, Variants of
11589 dg-require-support): Add dg-require-stack-size.
11591 2017-10-19 Martin Liska <mliska@suse.cz>
11594 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11595 that have a large alignment.
11596 (gimplify_target_expr): Likewise.
11598 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11600 PR rtl-optimization/82602
11601 * ira.c (rtx_moveable_p): Return false for volatile asm.
11603 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11606 * config/i386/i386-modes.def (CCGZ): New CC mode.
11607 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11608 * config/i386/predicates.md (ix86_comparison_operator):
11610 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11611 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11612 with double-word subtraction.
11613 (put_condition_code): Handle CCGZmode.
11615 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11617 * wide-int.cc (debug (const wide_int &)): New.
11618 (debug (const wide_int *)): New.
11619 (debug (const widest_int &)): New.
11620 (debug (const widest_int *)): New.
11622 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11624 PR middle-end/82556
11625 * lra-constraints.c (curr_insn_transform): Use non-input operand
11626 instead of output one for matched reload.
11628 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11630 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11631 (tree-ssa-loop-ivopts.h): New header file.
11632 (struct builtin_info): New fields.
11633 (classify_builtin_1): Compute and record base and offset parts for
11634 memset builtin partition by calling strip_offset.
11635 (offset_cmp, fuse_memset_builtins): New functions.
11636 (finalize_partitions): Fuse adjacent memset partitions by calling
11638 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11639 Expose the interface.
11640 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11642 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11644 PR tree-optimization/82574
11645 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11646 that data reference must be executed exactly once per iteration
11647 against the outermost loop in nest.
11648 (classify_partition): Update call to above function.
11650 2017-10-18 Richard Biener <rguenther@suse.de>
11652 PR tree-optimization/82591
11653 * graphite.c (graphite_transform_loops): Move code gen message
11655 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11656 Here. Handle scop_to_isl_ast failing.
11657 (scop_to_isl_ast): Limit the number of ISL operations.
11659 2017-10-18 Richard Biener <rguenther@suse.de>
11661 * graphite-isl-ast-to-gimple.c
11662 (translate_isl_ast_to_gimple::set_rename): Simplify.
11663 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11664 (graphite_copy_stmts_from_block): ... here.
11665 (copy_bb_and_scalar_dependences): Simplify.
11666 (add_parameters_to_ivs_params): Canonicalize.
11667 (generate_entry_out_of_ssa_copies): Simplify.
11668 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11669 by passing in ISL dimension.
11670 (parameter_index_in_region_1): Rename to ...
11671 (parameter_index_in_region): ... this.
11672 (extract_affine): Adjust assert, pass down parameter index.
11673 (add_param_constraints): Use range-info when available.
11674 (build_scop_context): Adjust.
11675 * sese.c (new_sese_info): Adjust.
11676 (free_sese_info): Likewise.
11677 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11678 Remove unused typedefs.
11679 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11681 2017-10-18 Martin Liska <mliska@suse.cz>
11683 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11685 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11687 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11688 (s390_sched_init): Do not reset s390_sched_state if we entered the
11689 current basic block via a fallthru edge and all others are unlikely.
11691 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11693 * config/s390/s390.c (NUM_SIDES): New variable.
11694 (LONGRUNNING_THRESHOLD): New variable.
11695 (LATENCY_FACTOR): New variable.
11696 (s390_sched_score): Decrease score for long-running instructions on
11698 (s390_sched_variable_issue): Perform bookkeeping for long-running
11701 2017-10-18 Richard Biener <rguenther@suse.de>
11703 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11704 Simplify with removal of the parameter rename map.
11705 (set_rename): Likewise.
11706 (should_copy_to_new_region): Likewise.
11707 (graphite_copy_stmts_from_block): Likewise.
11708 (copy_bb_and_scalar_dependences): Remove initialization of
11709 unused copied_bb_map.
11710 (copy_def): Remove.
11711 (copy_internal_parameters): Likewise.
11712 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11713 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11714 Use INTEGRAL_TYPE_P.
11715 (parameter_index_in_region_1): Rename to ...
11716 (assign_parameter_index_in_region): ... this. Assert we have
11717 a parameter we handle.
11718 (scan_tree_for_params): Adjust.
11719 * sese.h (parameter_rename_map_t): Remove.
11720 (struct sese_info_t): Remove unused parameter_rename_map and
11721 copied_bb_map members.
11722 * sese.c (new_sese_info): Adjust.
11723 (free_sese_info): Likewise.
11725 2017-10-18 Martin Liska <mliska@suse.cz>
11728 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11729 on an abnormal edge.
11731 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11733 * doc/invoke.texi (ffunction-sections and fdata-sections):
11736 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11738 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11739 the use statement can throw internally.
11741 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11743 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11744 any RTX present on the RHS of a SET.
11745 * compare-elim.c (try_eliminate_compare): Restore comment.
11747 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11749 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11750 may be also called on tcc_type nodes.
11751 * langhooks.c (lhd_tree_size): Likewise.
11753 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11755 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11756 format_warning_at_substring.
11757 (maybe_warn): Convert source_range * param to a location_t. Pass
11758 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11759 (format_directive): Remove code to extract source_ranges and
11760 source_range * in favor of just a location_t.
11761 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11763 * substring-locations.c (format_warning_va): Convert
11764 source_range * param to a location_t.
11765 (format_warning_at_substring): Likewise.
11766 * substring-locations.h (format_warning_va): Likewise.
11767 (format_warning_at_substring): Likewise.
11769 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11771 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11773 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11774 and scatter/gather ops.
11776 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11777 vec_gather_load and vec_scatter_store.
11778 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11779 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11780 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11781 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11782 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11783 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11785 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11787 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11788 Detect FTST instruction and handle its register pops. Only pop
11789 second operand if can_pop_second_op is true.
11790 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
11791 set can_pop_second_op to false in the compare_for_stack_reg call.
11793 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
11794 output_fp_compare for stack register operands.
11795 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
11796 instructions here. Do not emit stack register pops here. Assert
11797 that FCOMPP pops next to top stack register. Rewrite function.
11799 2017-10-17 Nathan Sidwell <nathan@acm.org>
11801 PR middle-end/82577
11802 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
11803 use DECL_ASSEMBLER_NAME_RAW.
11805 PR middle-end/82546
11806 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
11809 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
11810 Wilco Dijkstra <wilco.dijkstra@arm.com>
11812 * builtins.c (expand_builtin_update_setjmp_buf): Add a
11813 converstion to Pmode from the buf_addr.
11815 2017-10-17 Richard Biener <rguenther@suse.de>
11817 * graphite-dependences.c (scop_get_reads_and_writes): Change
11818 output parameters to references.
11820 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11822 PR 71026/tree-optimization
11823 * fold-const.c (distribute_real_division): Removed.
11824 (fold_binary_loc): Remove calls to distribute_real_divison.
11826 2017-10-17 Richard Biener <rguenther@suse.de>
11828 * graphite-scop-detection.c
11829 (scop_detection::stmt_has_simple_data_refs_p): Always use
11830 the full nest as region.
11831 (try_generate_gimple_bb): Likewise.
11832 * sese.c (scalar_evolution_in_region): Simplify now that
11833 SCEV can handle instantiation in regions.
11834 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11835 in the non-loop part of a function if requested.
11837 2017-10-17 Richard Biener <rguenther@suse.de>
11839 PR tree-optimization/82563
11840 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11842 (graphite_regenerate_ast_isl): Call it.
11843 * graphite-scop-detection.c (build_scops): Remove entry edge split.
11845 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11847 PR tree-optimization/82549
11848 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
11849 Formatting fixes. Instead of calling make_bit_field_ref with negative
11852 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11854 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11855 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
11856 _mm_maskz_reduce_ss): New.
11857 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
11858 __builtin_ia32_reducess_mask): Ditto..
11859 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
11860 * config/i386/sse.md (reduces<mode>): Renamed to ...
11861 (reduces<mode><mask_scalar_name>): ... this.
11862 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
11864 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
11865 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
11867 2017-10-16 David Malcolm <dmalcolm@redhat.com>
11869 * Makefile.in (OBJS): Add unique-ptr-tests.o.
11870 * selftest-run-tests.c (selftest::run_tests): Call
11871 selftest::unique_ptr_tests_cc_tests.
11872 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
11873 * unique-ptr-tests.cc: New file.
11875 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11878 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11880 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11881 (make_hard_regno_born, make_hard_regno_dead): Update
11882 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11884 2017-10-16 Jeff Law <law@redhat.com>
11886 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11888 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11890 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11892 2017-10-16 Olivier Hainque <hainque@adacore.com>
11894 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11895 with_cpu if we were configured for an e500v2 target cpu name.
11897 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11899 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11900 * doc/invoke.texi: Document +nodsp as a valid extension for
11903 2017-10-16 Martin Liska <mliska@suse.cz>
11905 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11906 (test_set_range): Likewise.
11907 (test_range_functions): Rename to ...
11908 (test_bit_in_range): ... this.
11909 (sbitmap_c_tests): Add new test.
11911 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11913 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11915 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11916 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11918 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11920 * config/aarch64/aarch64-builtins.c
11921 (aarch64_types_quadopu_lane_qualifiers): New.
11922 (TYPES_QUADOPU_LANE): New.
11923 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11924 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11925 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11926 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11927 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11928 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11929 (Vdottype, DOTPROD): New.
11930 (sur): Add SDOT and UDOT.
11932 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11934 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11935 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11936 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11937 Add TARGET_DOTPROD.
11938 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11939 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11940 Enable TARGET_DOTPROD.
11941 (cortex-a75.cortex-a55): Likewise.
11942 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11944 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11946 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11947 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11949 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11951 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11952 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11953 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11954 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11955 * config/arm/types.md (neon_dot, neon_dot_q): New.
11956 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11958 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11960 * config/arm/arm.h (TARGET_DOTPROD): New.
11961 * config/arm/arm.c (arm_arch_dotprod): New.
11962 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11963 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11964 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11965 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11966 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11967 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11968 * doc/invoke.texi (armv8.2-a): Document dotprod
11970 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11972 * i386.c (ix86_vec_cost): New function.
11973 (ix86_rtx_costs): Handle vector operations better.
11974 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11975 * x86-tune-costs.h: Add new costs to all tables.
11977 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11979 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11981 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11982 divsd, sqrtss and sqrtsd
11983 * x86-tune-costs.h: Add new entries to all costs.
11984 (znver1_cost): Fix to match real instruction latencies.
11986 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11987 Michael Collison <michael.collison@arm.com>
11989 * compare-elim.c: Include emit-rtl.h.
11990 (can_merge_compare_into_arith): New function.
11991 (try_validate_parallel): Likewise.
11992 (try_merge_compare): Likewise.
11993 (try_eliminate_compare): Call the above when no previous clobber
11995 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11998 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12000 PR middle-end/62263
12001 PR middle-end/82498
12002 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12003 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12005 PR middle-end/62263
12006 PR middle-end/82498
12007 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12008 to be any operand_equal_p operands. For & (B - 1) require
12009 B to be power of 2. Recognize
12010 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12012 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12015 * optabs.c (expand_memory_blockage): Fix call of
12016 targetm.have_memory_blockage.
12018 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12021 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12022 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12023 objects to extra_objs instead of overwriting it.
12025 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12027 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12028 Use any_fp_register_operand as operand[3] predicate. Simplify
12029 equality test for operands[2] and operands[4] memory location.
12030 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12031 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12032 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12033 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12034 any_fp_register_operand as operand[1] predicate. Simplify
12035 equality test for operands[0] and operands[3] memory location.
12036 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12037 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12038 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12040 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12042 * target-insns.def: Add memory_blockage.
12043 * optabs.c (expand_memory_blockage): New function.
12044 (expand_asm_memory_barrier): Rename ...
12045 (expand_asm_memory_blockage): ... to this.
12046 (expand_mem_thread_fence): Call expand_memory_blockage
12047 instead of expand_asm_memory_barrier.
12048 (expand_mem_singnal_fence): Ditto.
12049 (expand_atomic_load): Ditto.
12050 (expand_atomic_store): Ditto.
12051 * doc/md.texi (Standard Pattern Names For Generation):
12052 Document memory_blockage instruction pattern.
12054 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12056 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12057 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12058 * config/rl78/rl78.md: New define_expand "adddi3".
12060 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12062 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12065 2017-10-13 Jeff Law <law@redhat.com>
12067 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12069 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12072 * internal-fn.c (expand_mul_overflow): If both operands have
12073 the same highpart of -1 or 0 and the topmost bit of lowpart
12074 is different, overflow is if res <= 0 rather than res < 0.
12076 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12078 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12079 TARGET_P9_VECTOR code for unaligned_load case.
12081 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12083 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12085 2017-10-13 Nathan Sidwell <nathan@acm.org>
12087 * tree-core.h (tree_contains_struct): Make bool.
12088 * tree.c (tree_contains_struct): Likewise.
12089 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12090 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12091 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12092 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12093 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12095 2017-10-13 Richard Biener <rguenther@suse.de>
12097 * graphite-isl-ast-to-gimple.c
12098 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12099 parameters and dominance check.
12100 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12101 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12102 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12103 Do not update SSA form here or do intermediate IL verification.
12104 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12105 (graphite_initialize): Remove check on the number of loops in
12106 the function and inline into graphite_transform_loops.
12107 (graphite_finalize): Inline into graphite_transform_loops.
12108 (graphite_transform_loops): Perform SSA update and IL verification
12110 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12112 2017-10-13 Richard Biener <rguenther@suse.de>
12114 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12115 graphite_expression_type_precision): Avoid global constructor
12117 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12118 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12119 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12120 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12121 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12122 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12124 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12127 * config/i386/i386.h (ix86_red_zone_size): New.
12128 * config/i386/i386.md (push peephole2s): Replace
12129 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12131 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12132 Alan Hayward <alan.hayward@arm.com>
12133 David Sherwood <david.sherwood@arm.com>
12135 * combine.c (can_change_dest_mode): Reject changes in
12136 REGMODE_NATURAL_SIZE.
12138 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12139 Alan Hayward <alan.hayward@arm.com>
12140 David Sherwood <david.sherwood@arm.com>
12142 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12143 (expand_debug_source_expr): Likewise.
12144 * combine.c (combine_simplify_rtx): Likewise.
12145 * cse.c (fold_rtx): Likewise.
12146 * fwprop.c (canonicalize_address): Likewise.
12147 * targhooks.c (default_shift_truncation_mask): Likewise.
12149 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12150 Alan Hayward <alan.hayward@arm.com>
12151 David Sherwood <david.sherwood@arm.com>
12153 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12154 (widened_mode): Likewise.
12155 (expand_unop): Likewise.
12156 * ree.c (transform_ifelse): Likewise.
12157 (merge_def_and_ext): Likewise.
12158 (combine_reaching_defs): Likewise.
12159 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12161 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12162 Alan Hayward <alan.hayward@arm.com>
12163 David Sherwood <david.sherwood@arm.com>
12165 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12166 * combine.c (gen_lowpart_for_combine): Likewise.
12167 * dwarf2out.c (rtl_for_decl_location): Likewise.
12168 * final.c (alter_subreg): Likewise.
12169 * rtlhooks.c (gen_lowpart_general): Likewise.
12170 (gen_lowpart_if_possible): Likewise.
12172 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12173 Alan Hayward <alan.hayward@arm.com>
12174 David Sherwood <david.sherwood@arm.com>
12176 * calls.c (expand_call): Use subreg_lowpart_offset.
12177 * cse.c (cse_insn): Likewise.
12178 * regcprop.c (copy_value): Likewise.
12179 (copyprop_hardreg_forward_1): Likewise.
12181 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12184 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12185 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12186 =Q constraints to +Q and into insn condition add check
12187 that operands[0] and operands[1] are equal.
12188 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12189 =Q constraints to +Q and into insn condition add check
12190 that operands[0] is equal to either operands[1] or operands[2].
12193 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12194 instead of handling MINUS_EXPR twice (once for each argument),
12195 canonicalize operand order and handle just once, use rtype where
12196 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12199 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12200 any values of __C while still being pattern recognizable as a simple
12201 rotate instruction.
12203 2017-10-13 Richard Biener <rguenther@suse.de>
12205 PR tree-optimization/82451
12207 2017-10-02 Richard Biener <rguenther@suse.de>
12209 PR tree-optimization/82355
12210 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12211 a mapping for the enclosing loop but avoid generating one for
12212 the loop tree root.
12213 (copy_bb_and_scalar_dependences): Remove premature codegen
12214 error on PHIs in blocks duplicated into multiple places.
12215 * graphite-scop-detection.c
12216 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12217 in the region use it as loop and nest to analyze the DR in.
12218 (try_generate_gimple_bb): Likewise.
12219 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12220 (add_loop_constraints): For blocks in a loop not in the region
12221 create a dimension with a single iteration.
12222 * sese.h (gbb_loop_at_index): Remove assert.
12224 * cfgloop.c (loop_preheader_edge): For the loop tree root
12225 return the single successor of the entry block.
12226 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12227 Reset the SCEV hashtable and niters.
12228 * graphite-scop-detection.c
12229 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12230 assert that we only have POLYNOMIAL_CHREC that vary in loops
12231 contained in the region.
12232 (scop_detection::graphite_can_represent_expr): Adjust.
12233 (scop_detection::stmt_has_simple_data_refs_p): For loops
12234 not in the region set loop to NULL. The nest is now the
12235 entry edge to the region.
12236 (try_generate_gimple_bb): Likewise.
12237 * sese.c (scalar_evolution_in_region): Adjust for
12238 instantiate_scev change.
12239 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12240 Make nest parameter the edge into the region.
12241 (create_data_ref): Likewise.
12242 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12243 entry edge into a region and adjust instantiate_scev calls.
12244 (create_data_ref): Likewise.
12245 (graphite_find_data_references_in_stmt): Likewise.
12246 (find_data_references_in_stmt): Pass the loop preheader edge
12247 from the nest argument.
12248 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12249 parameter the edge into the region.
12250 (instantiate_parameters): Use the loop preheader edge as entry.
12251 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12253 (get_instantiated_value_entry): Make instantiate_below parameter
12254 the edge into the region.
12255 (instantiate_scev_name): Likewise. Adjust dominance checks,
12256 when we cannot use loop-based instantiation instantiate by
12257 walking use-def chains.
12258 (instantiate_scev_poly): Adjust.
12259 (instantiate_scev_binary): Likewise.
12260 (instantiate_scev_convert): Likewise.
12261 (instantiate_scev_not): Likewise.
12262 (instantiate_array_ref): Remove.
12263 (instantiate_scev_3): Likewise.
12264 (instantiate_scev_2): Likewise.
12265 (instantiate_scev_1): Likewise.
12266 (instantiate_scev_r): Do not blindly handle N-operand trees.
12267 Do not instantiate array-refs. Handle all constants and invariants.
12268 (instantiate_scev): Make instantiate_below parameter
12269 the edge into the region.
12270 (resolve_mixers): Use the loop preheader edge for the region
12271 parameter to instantiate_scev_r.
12272 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12274 2017-10-13 Richard Biener <rguenther@suse.de>
12276 PR tree-optimization/82525
12277 * graphite-isl-ast-to-gimple.c
12278 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12280 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12281 Fail code generation when we cannot represent the isl integer.
12282 (binary_op_to_tree): Elide modulo operations that are no-ops
12283 in the type we code generate. Remove now superfluous code
12286 2017-10-13 Richard Biener <rguenther@suse.de>
12288 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12289 (scop_detection::harmful_loop_in_region): Remove premature
12290 IV type restriction.
12291 (scop_detection::graphite_can_represent_scev): We can handle
12292 pointer IVs just fine.
12294 2017-10-13 Alan Modra <amodra@gmail.com>
12296 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12297 "Clobbers and Scratch Registers". Add paragraph on
12298 alternative to clobbers for scratch registers and OpenBLAS
12301 2017-10-13 Alan Modra <amodra@gmail.com>
12303 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12304 example of a memory input for a string of known length. Move
12305 commentary out of table. Add a number of new examples
12306 covering array memory inputs.
12308 2017-10-12 Martin Liska <mliska@suse.cz>
12310 PR tree-optimization/82493
12311 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12312 (test_range_functions): New function.
12313 (sbitmap_c_tests): Likewise.
12314 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12315 * selftest.h (sbitmap_c_tests): New function.
12317 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12319 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12321 * config/rs6000/amo.h: Fix spacing issue.
12323 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12326 * config/i386/i386.md (*ashl<mode>3_mask_1,
12327 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12328 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12331 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12333 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12334 (profile_probability): Set max_probability
12335 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12337 * profile-count.c (profile_probability::differs_from_p): Do not
12338 rely on max_probaiblity == 10000
12340 2017-10-12 Jeff Law <law@redhat.com>
12342 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12345 2017-10-12 Martin Sebor <msebor@redhat.com>
12349 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12350 (handle_alias_pairs): Call it.
12351 * common.opt (-Wattribute-alias): New option.
12352 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12353 * doc/invoke.texi (-Wattribute-alias): Document.
12355 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12358 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12360 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12362 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12363 (make_hard_regno_born, make_hard_regno_dead): Update
12364 bb_killed_pseudos and bb_gen_pseudos.
12366 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12368 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12370 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12372 * config/alpha/alpha.c (alpha_split_conditional_move):
12373 Use std::swap instead of manually swapping.
12374 (alpha_stdarg_optimize_hook): Ditto.
12375 (alpha_canonicalize_comparison): Ditto.
12377 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12379 * tree-loop-distribution.c (struct builtin_info): New struct.
12380 (struct partition): Refactor fields into struct builtin_info.
12381 (partition_free): Free struct builtin_info.
12382 (build_size_arg_loc, build_addr_arg_loc): Delete.
12383 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12384 information from struct builtin_info.
12385 (find_single_drs): New function refactored from classify_partition.
12386 Also moved builtin validity checks to this function.
12387 (compute_access_range, alloc_builtin): New functions.
12388 (classify_builtin_st, classify_builtin_ldst): New functions.
12389 (classify_partition): Refactor code into functions find_single_drs,
12390 classify_builtin_st and classify_builtin_ldst.
12391 (distribute_loop): Don't do runtime alias check when distributing
12393 (find_seed_stmts_for_distribution): New function.
12394 (pass_loop_distribution::execute): Refactor code finding seed
12395 stmts into above function. Support distribution for the innermost
12396 two-level loop nest. Adjust dump information.
12398 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12400 * tree-loop-distribution.c: Adjust the general comment.
12401 (NUM_PARTITION_THRESHOLD): New macro.
12402 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12403 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12404 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12405 in call to build_partition_graph.
12406 (finalize_partitions): New parameter. Make loop distribution more
12407 conservative by fusing more partitions.
12408 (distribute_loop): Don't do runtime alias check in case of loop nest
12410 (find_seed_stmts_for_distribution): New function.
12411 (prepare_perfect_loop_nest): New function.
12412 (pass_loop_distribution::execute): Refactor code finding seed stmts
12413 and loop nest into above functions. Support loop nest distribution.
12414 Adjust dump information accordingly.
12416 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12418 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12419 and set PTYPE_SEQUENTIAL for merged partition.
12421 2017-10-12 Richard Biener <rguenther@suse.de>
12423 PR tree-optimization/69728
12425 2017-09-19 Richard Biener <rguenther@suse.de>
12427 PR tree-optimization/69728
12428 * graphite-sese-to-poly.c (schedule_error): New global.
12429 (add_loop_schedule): Handle empty domain by failing the
12431 (build_original_schedule): Handle schedule_error.
12433 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12434 domain by returning an unchanged schedule.
12436 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12438 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12439 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12441 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12443 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12446 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12449 * expr.c (store_field): Don't optimize away bitsize == 0 store
12450 from CALL_EXPR with addressable return type.
12452 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12454 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12455 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12456 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12457 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12458 TARGET_ISEL instead of TARGET_ISEL<sel>.
12460 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12462 * config/rs6000/rs6000.c
12463 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12465 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12467 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12469 (reg_or_cint_operand): Fix comment.
12470 (reg_or_zero_operand): New predicate.
12471 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12472 * config/rs6000/rs6000.c (output_isel): Delete.
12473 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12474 instead of reg_or_cint_operand. Output instruction directly (not via
12476 (isel_unsigned_<mode>): Ditto.
12477 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12478 gpc_reg_operand. Add an instruction alternative for this. Output
12479 instruction directly.
12480 (*isel_reversed_unsigned_<mode>): Ditto.
12482 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12484 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12485 (TARGET_CANONICALIZE_COMPARISON): Define.
12487 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12490 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12491 Check whether the dest is REG before adding REG_EQUIV note.
12493 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12496 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12498 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12499 (make_hard_regno_born, make_hard_regno_dead): Update
12500 bb_killed_pseudos and bb_gen_pseudos.
12502 2017-10-11 Nathan Sidwell <nathan@acm.org>
12504 * incpath.h (enum incpath_kind): Name enum, prefix values.
12505 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12506 * incpath.c (heads, tails): Use INC_MAX.
12507 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12508 (merge_include_chains, split_quote_chain,
12509 register_include_chains): Update incpath_kind names.
12510 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12511 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12513 (add_framework_path, darwin_register_objc_includes): Likewise.
12514 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12516 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12518 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12519 Do not use float_operator operator predicate.
12520 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12521 * config/i386/predicates.md (float_operator): Remove predicate.
12523 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12525 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12526 (*jccxf_i387): Ditto.
12527 (*jcc<mode>_i387): Ditto.
12528 (*jccu<mode>_i387): Ditto.
12529 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12530 (*jcc_*_i387 splitters): Remove.
12531 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12532 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12533 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12536 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12538 * profile-count.h (slow_safe_scale_64bit): New function.
12539 (safe_scale_64bit): New inline.
12540 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12541 * profile-count.c: Include wide-int.h
12542 (slow_safe_scale_64bit): New.
12544 2017-10-11 Nathan Sidwell <nathan@acm.org>
12546 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12547 HAS_DECL_ASSEMBLER_NAME_P.
12548 * gimple-expr.c (gimple_decl_printable_name: Check
12549 HAS_DECL_ASSEMBLER_NAME_P too.
12550 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12551 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12552 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12553 * passes.c (rest_of_decl_compilation): Check
12554 HAS_DECL_ASSEMBLER_NAME_P too.
12555 * recog.c (verify_changes): Likewise.
12556 * tree-pretty-print.c (dump_decl_name): Likewise.
12557 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12559 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12560 (SET_DECL_ASSEMBLER_NAME): Use it.
12561 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12562 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12563 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12565 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12567 * config.gcc (i386, x86_64): Add extra objects.
12568 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12569 (ix86_min_insn_size): Declare.
12570 (ix86_issue_rate): Declare.
12571 (ix86_adjust_cost): Declare.
12572 (ia32_multipass_dfa_lookahead): Declare.
12573 (ix86_macro_fusion_p): Declare.
12574 (ix86_macro_fusion_pair_p): Declare.
12575 (ix86_bd_has_dispatch): Declare.
12576 (ix86_bd_do_dispatch): Declare.
12577 (ix86_core2i7_init_hooks): Declare.
12578 (ix86_atom_sched_reorder): Declare.
12579 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12580 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12581 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12582 (rip_relative_addr_p): Rename to ...
12583 (ix86_rip_relative_addr_p): ... this one; export.
12584 (memory_address_length): Update.
12585 (ix86_issue_rate): Move to x86-tune-sched.c.
12586 (ix86_flags_dependent): Move to x86-tune-sched.c.
12587 (ix86_agi_dependent): Move to x86-tune-sched.c.
12588 (exact_dependency_1): Move to x86-tune-sched.c.
12589 (exact_store_load_dependency): Move to x86-tune-sched.c.
12590 (ix86_adjust_cost): Move to x86-tune-sched.c.
12591 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12592 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12593 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12594 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12595 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12596 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12597 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12598 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12599 (min_insn_size): Rename to ...
12600 (ix86_min_insn_size): ... this one; export.
12601 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12602 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12603 (core2i7_first_cycle_multipass_backtrack): Move to
12604 x86-tune-sched-core.c.
12605 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12606 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12607 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12608 (ix86_avoid_jump_mispredicts): Update.
12609 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12610 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12611 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12612 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12613 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12614 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12615 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12616 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12617 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12618 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12619 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12620 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12621 (BIG): Move to ix86-tune-sched-bd.c.
12622 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12623 (enum insn_path): Move to ix86-tune-sched-bd.c.
12624 (get_mem_group): Move to ix86-tune-sched-bd.c.
12625 (is_cmp): Move to ix86-tune-sched-bd.c.
12626 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12627 (is_branch): Move to ix86-tune-sched-bd.c.
12628 (is_prefetch): Move to ix86-tune-sched-bd.c.
12629 (init_window): Move to ix86-tune-sched-bd.c.
12630 (allocate_window): Move to ix86-tune-sched-bd.c.
12631 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12632 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12633 (process_end_window): Move to ix86-tune-sched-bd.c.
12634 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12635 (find_constant): Move to ix86-tune-sched-bd.c.
12636 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12637 (has_immediate): Move to ix86-tune-sched-bd.c.
12638 (get_insn_path): Move to ix86-tune-sched-bd.c.
12639 (get_insn_group): Move to ix86-tune-sched-bd.c.
12640 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12641 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12642 (add_insn_window): Move to ix86-tune-sched-bd.c.
12643 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12644 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12645 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12646 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12647 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12648 (do_dispatch): Move to ix86-tune-sched-bd.c.
12649 (has_dispatch): Move to ix86-tune-sched-bd.c.
12650 * config/i386/t-i386: Add new object files.
12651 * config/i386/x86-tune-costs.h: New file.
12652 * config/i386/x86-tune-sched-atom.c: New file.
12653 * config/i386/x86-tune-sched-bd.c: New file.
12654 * config/i386/x86-tune-sched-core.c: New file.
12655 * config/i386/x86-tune-sched.c: New file.
12657 2017-10-11 Liu Hao <lh_mouse@126.com>
12659 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12660 the generic version below instead.
12661 (should_colorize): Recognize Windows consoles as terminals
12663 * pretty-print.c [__MINGW32__] (write_all): New function.
12664 [__MINGW32__] (find_esc_head): Likewise.
12665 [__MINGW32__] (find_esc_terminator): Likewise.
12666 [__MINGW32__] (eat_esc_sequence): Likewise.
12667 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12669 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12672 2017-10-11 Richard Biener <rguenther@suse.de>
12674 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12675 Properly call analyze_scalar_evolution with the loop of the stmt.
12677 2017-10-11 Richard Biener <rguenther@suse.de>
12679 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12680 * tree-core.h (tree_base): Add chrec_var union member.
12681 * tree.h (CHREC_VAR): Remove.
12682 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12683 * tree-chrec.h (build_polynomial_chrec): Adjust.
12684 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12685 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12687 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12689 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12690 * match.pd: ... here.
12691 ((T) X == (T) Y): Relax condition.
12693 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12695 PR tree-optimization/82472
12696 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12698 (break_alias_scc_partitions): Update postorder number.
12700 2017-10-11 Martin Liska <mliska@suse.cz>
12703 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12705 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12707 2017-10-11 Martin Liska <mliska@suse.cz>
12709 * print-rtl.c (print_insn): Move declaration of idbuf
12710 to same scope as name.
12712 2017-10-11 Martin Liska <mliska@suse.cz>
12717 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12718 volatile arguments.
12720 2017-10-11 Martin Liska <mliska@suse.cz>
12723 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12724 volatile arguments.
12726 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12728 * config.gcc (default_gnu_indirect_function): Default to yes for
12729 arm*-*-linux* with glibc.
12731 2017-10-11 Richard Biener <rguenther@suse.de>
12733 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12734 default-defs and types we do not want to analyze.
12735 (interpret_loop_phi): Replace unreachable code with an assert.
12736 (compute_scalar_evolution_in_loop): Remove and inline ...
12737 (analyze_scalar_evolution_1): ... here, replacing condition with
12738 what makes the intent clearer. Remove handling of cases
12739 get_scalar_evolution now handles.
12741 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12743 PR rtl-optimization/81434
12744 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12745 comment for main loop. In sched_group_found if, also add checks for
12746 pass and min_cost_group.
12748 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12750 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12751 (rs6000_insn_cost): New function.
12752 * config/rs6000/rs6000.md (cost): New attribute.
12754 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12755 H.J. Lu <hongjiu.lu@intel.com>
12759 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12760 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12761 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12762 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12763 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12764 OPTION_MASK_ISA_MMX and some other bit set require both
12765 mmx and the ISAs without the mmx bit.
12766 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12767 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12768 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12769 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12770 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12771 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12772 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12773 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12774 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12775 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12776 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12777 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12778 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12779 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12780 Add OPTION_MASK_ISA_MMX.
12782 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12784 * config.gcc (armv7*-*-freebsd*): New target.
12785 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12787 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12789 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
12790 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
12793 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
12795 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
12796 a template parameter.
12797 (WIDE_INT_REF_FOR): Update accordingly.
12798 * tree.h (wi::int_traits <const_tree>): Delete.
12799 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
12800 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
12801 (wi::tree_to_wide_ref): New typedef.
12802 (wi::to_wide): New function.
12803 * calls.c (get_size_range): Use wi::to_wide when operating on
12804 trees as wide_ints.
12805 * cgraph.c (cgraph_node::create_thunk): Likewise.
12806 * config/i386/i386.c (ix86_data_alignment): Likewise.
12807 (ix86_local_alignment): Likewise.
12808 * dbxout.c (stabstr_O): Likewise.
12809 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
12810 * expr.c (const_vector_from_tree): Likewise.
12811 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
12812 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
12813 (fold_negate_expr_1, int_const_binop_1, const_binop)
12814 (fold_convert_const_int_from_real, optimize_bit_field_compare)
12815 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
12816 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
12817 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12818 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12819 (fold_not_const, round_up_loc): Likewise.
12820 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12821 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12822 (alloca_call_type): Likewise.
12823 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12824 * godump.c (go_output_typedef): Likewise.
12825 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12826 * internal-fn.c (get_min_precision): Likewise.
12827 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12828 * ipa-polymorphic-call.c
12829 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12830 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12831 (ipa_modify_call_arguments): Likewise.
12832 * match.pd: Likewise.
12833 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12834 * print-tree.c (print_node_brief, print_node): Likewise.
12835 * stmt.c (expand_case): Likewise.
12836 * stor-layout.c (layout_type): Likewise.
12837 * tree-affine.c (tree_to_aff_combination): Likewise.
12838 * tree-cfg.c (group_case_labels_stmt): Likewise.
12839 * tree-data-ref.c (dr_analyze_indices): Likewise.
12840 (prune_runtime_alias_test_list): Likewise.
12841 * tree-dump.c (dequeue_and_dump): Likewise.
12842 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
12843 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
12844 * tree-pretty-print.c (dump_generic_node): Likewise.
12845 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
12846 (simple_iv_with_niters): Likewise.
12847 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
12848 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
12849 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
12850 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12851 (refine_value_range_using_guard, number_of_iterations_ne_max)
12852 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12853 (get_cst_init_from_scev, record_nonwrapping_iv)
12854 (scev_var_range_cant_overflow): Likewise.
12855 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
12856 * tree-ssa-pre.c (compute_avail): Likewise.
12857 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
12858 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
12859 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
12860 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
12861 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
12862 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
12863 (dump_case_nodes, try_switch_expansion): Likewise.
12864 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
12865 (vect_do_peeling): Likewise.
12866 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
12867 * tree-vect-stmts.c (vectorizable_load): Likewise.
12868 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
12869 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
12870 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
12871 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
12872 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
12873 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
12874 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12875 (evrp_dom_walker::before_dom_children): Likewise.
12876 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12877 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12878 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12879 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12880 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12881 (get_range_pos_neg): Likewise.
12882 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12883 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12884 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12885 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12886 * config/avr/avr.c (avr_fold_builtin): Likewise.
12887 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12888 * config/msp430/msp430.c (msp430_attr): Likewise.
12889 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12890 * config/powerpcspe/powerpcspe-c.c
12891 (altivec_resolve_overloaded_builtin): Likewise.
12892 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12893 (rs6000_expand_ternop_builtin): Likewise.
12894 * config/rs6000/rs6000-c.c
12895 (altivec_resolve_overloaded_builtin): Likewise.
12896 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12897 (rs6000_expand_ternop_builtin): Likewise.
12898 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12900 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12902 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12903 when copying loop nest with only one inner loop.
12905 2017-10-10 Richard Biener <rguenther@suse.de>
12907 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12908 blocks if SCEV is active.
12909 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12911 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12912 (scev_initialize): Assert we are not yet initialized.
12914 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12916 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12917 inner loop's exit stmt by making it always exit the loop, otherwise
12918 we would generate an infinite empty loop.
12920 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12922 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12923 renaming variables in new preheader if it's deleted.
12925 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12927 * tree-loop-distribution.c (struct partition): Remove unused field
12928 loops of the structure.
12929 (partition_alloc, partition_free): Ditto.
12930 (build_rdg_partition_for_vertex): Ditto.
12932 2017-10-09 Jeff Law <law@redhat.com>
12934 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12935 return type to match prototype and documentation.
12937 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12939 * config/rs6000/rs6000.c (processor_costs): Move to ...
12940 * config/rs6000/rs6000.h: ... here.
12941 (rs6000_cost): Declare.
12943 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12945 * except.c (setjmp_fn): New global variable.
12946 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12947 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12948 if DONT_USE_BUILTIN_SETJMP is defined.
12950 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12952 * target.def (insn_cost): New hook.
12953 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12954 * doc/tm.texi: Regenerate.
12955 * rtlanal.c (insn_cost): Use the new hook.
12957 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12959 * combine.c (combine_validate_cost): Compute the new insn_cost,
12960 not just pattern_cost.
12961 (try_combine): Adjust comment.
12963 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12965 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12967 * combine.c (uid_insn_cost): Adjust comment.
12968 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12970 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12971 * dse.c (find_shift_sequence): Ditto.
12972 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12973 (bb_valid_for_noce_process_p): Use pattern_cost.
12974 * rtl.h (insn_rtx_cost): Delete.
12975 (pattern_cost): New prototype.
12976 (insn_cost): New prototype.
12977 * rtlanal.c (insn_rtx_cost): Rename to...
12978 (pattern_cost): ... this.
12981 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12983 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12984 (*jcc<mode>_0_r_i387): Ditto.
12985 (*jccxf_r_i387): Ditto.
12986 (*jcc<mode>_r_i387): Ditto.
12987 (*jccu<mode>_r_i387): Ditto.
12988 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12989 (*jcc): Rename from *jcc_1.
12991 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12993 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12994 deferred rescans after the lvx/stvx recombination pre-pass.
12996 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12998 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12999 memory operation instruction support.
13000 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13001 (rs6000-ibm-aix[789]*): Likewise.
13002 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13003 Document new functions.
13005 2017-10-09 Richard Biener <rguenther@suse.de>
13007 PR tree-optimization/82397
13008 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13009 equality only for semantically equal trees.
13011 2017-10-09 Richard Biener <rguenther@suse.de>
13013 PR tree-optimization/82449
13014 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13015 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13016 allow constant addresses.
13017 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13020 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13022 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13025 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13028 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13031 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13034 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13036 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13039 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13040 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13042 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13044 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13045 (WI_BINARY_PREDICATE_RESULT): Likewise.
13046 (wi::binary_traits::operator_result): New type.
13047 (wi::binary_traits::predicate_result): Likewise.
13048 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13049 (generic_wide_int::operator==, generic_wide_int::operator!=)
13050 (generic_wide_int::operator&, generic_wide_int::and_not)
13051 (generic_wide_int::operator|, generic_wide_int::or_not)
13052 (generic_wide_int::operator^, generic_wide_int::operator+
13053 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13055 (operator~, unary operator-, operator==, operator!=, operator&)
13056 (operator|, operator^, operator+, binary operator-, operator*): New
13058 * expr.c (get_inner_reference): Use wi::bit_and_not.
13059 * fold-const.c (fold_binary_loc): Likewise.
13060 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13061 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13062 (bit_value_binop): Likewise.
13063 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13064 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13065 (extract_range_from_binary_expr_1): Likewise.
13066 (masked_increment): Likewise.
13067 (simplify_bit_ops_using_ranges): Likewise.
13069 2017-10-09 Martin Jambor <mjambor@suse.cz>
13072 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13073 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13074 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13076 (hsa_fixup_mov_insn_type): New function.
13077 (hsa_op_with_type::get_in_type): Use it.
13078 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13079 immediates in an assert.
13080 (hsa_op_with_type::extend_int_to_32bit): New method.
13081 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13082 types. Convert to dest type if necessary.
13083 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13084 (reg_for_gimple_ssa): Pass false as min32int to
13085 hsa_type_for_scalar_tree_type.
13086 (gen_hsa_addr): Fixup type when creating addresable temporary.
13087 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13088 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13089 necessary. Call hsa_fixup_mov_insn_type.
13090 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13091 extend operands and convert to dest type if necessary.
13092 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13093 to dest type if necessary.
13094 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13095 if conversion nt necessary and size matches.
13096 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13097 to dest type if necessary.
13098 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13099 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13101 (gen_hsa_clrsb): Likewise.
13102 (gen_hsa_ffs): Likewise.
13103 (gen_hsa_divmod): Extend operands and convert to dest type if
13105 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13107 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13109 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13110 Remove empty default arguments. Use a brace block as output
13112 (conditional return): Ditto.
13114 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13115 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13117 (group_ending_nop): Ditto.
13118 (doloop_end): Ditto.
13119 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13120 (splitters for those): Ditto.
13122 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13124 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13125 a conditional jump (and the compare for it) so that pc_rtx is the
13127 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13128 for the deleted and renamed ctr<mode>_internal[234] patterns.
13129 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13130 Delete second conditional return pattern.
13131 (ctr<mode>_internal2): Delete this second bdnz pattern.
13132 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13133 (ctr<mode>_internal4): Delete this second bdz pattern.
13135 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13137 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13138 (always_initialized_rtx_for_ssa_name_p): New predicate.
13139 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13140 (finish_out_of_ssa): Free new field of SA.
13141 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13142 * tree-ssa-coalesce.c: Include tree-ssa.h.
13143 (get_parm_default_def_partitions): Remove extern keyword.
13144 (get_undefined_value_partitions): New function.
13145 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13146 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13147 uninitialized bits.
13148 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13150 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13152 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13154 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13156 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13157 for targets that preffer 128bit.
13159 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13161 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13163 2017-10-08 Olivier Hainque <hainque@adacore.com>
13165 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13166 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13168 (thumb_set_return_address): Likewise.
13170 2017-10-08 Olivier Hainque <hainque@adacore.com>
13172 * common/config/arm/arm-common.c (arm_except_unwind_info):
13173 Handle DWARF2_UNWIND_INFO.
13175 2017-10-07 Michael Collison <michael.collison@arm.com>
13177 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13180 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13182 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13183 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13184 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13185 defined, force the creation of a new block for a dispatch label.
13187 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13189 * invoke.texi (Wsuggest-attribute=cold): Document.
13190 * common.opt (Wsuggest-attribute=cold): New
13191 * ipa-pure-const.c (warn_function_cold): New function.
13192 * predict.c (compute_function_frequency): Use it.
13193 * predict.h (warn_function_cold): Declare.
13195 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13197 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13200 2017-10-06 Martin Liska <mliska@suse.cz>
13202 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13203 keyword for member functions.
13204 (struct sanopt_tree_couple): New struct.
13205 (struct sanopt_tree_couple_hash): New function.
13206 (struct sanopt_ctx): Add new hash_map.
13207 (has_dominating_ubsan_ptr_check): New function.
13208 (record_ubsan_ptr_check_stmt): Likewise.
13209 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13210 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13211 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13213 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13216 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13217 aarch64_simd_valid_immediate on CONST_VECTORs.
13218 (aarch64_reg_or_bic_imm): Likewise.
13220 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13222 PR rtl-optimization/82396
13223 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13225 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13227 * graphite-dependences.c (scop_get_reads): Move code to...
13228 (scop_get_must_writes): Move code to...
13229 (scop_get_may_writes): Move code to...
13230 (scop_get_reads_and_writes): ... here.
13231 (scop_get_dependences): Call scop_get_reads_and_writes.
13233 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13235 PR tree-optimization/82434
13236 * fold-const.h (can_native_encode_type_p,
13237 can_native_encode_string_p): Remove.
13238 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13239 don't encode anything, just return what would be otherwise returned.
13240 (native_encode_fixed, native_encode_complex, native_encode_vector):
13242 (native_encode_string): Likewise. Inline by hand
13243 can_native_encode_string_p.
13244 (can_native_encode_type_p): Remove.
13245 (can_native_encode_string_p): Remove.
13246 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13247 STRING_CSTs using can_native_encode_string_p, test all
13248 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13249 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13250 argument from native_encode_expr.
13251 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13252 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13255 2017-10-06 Richard Biener <rguenther@suse.de>
13257 PR tree-optimization/82397
13258 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13259 operand_equal_p but rely on data_ref_compare_tree for detecting
13261 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13262 to match up with dr_group_sort_cmp.
13264 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13267 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13269 * config/s390/s390-builtin-types.def: Regenerate.
13271 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13274 * config/s390/s390-builtin-types.def: Regenerate.
13275 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13276 Change flag from B_VXE to B_VX.
13277 (s390_vec_min_dbl): Remove B_VXE flag.
13279 2017-10-06 Richard Biener <rguenther@suse.de>
13281 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13282 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13283 translate_isl_ast_to_gimple::is_valid_rename,
13284 translate_isl_ast_to_gimple::get_rename,
13285 translate_isl_ast_to_gimple::get_def_bb_for_const,
13286 translate_isl_ast_to_gimple::get_new_name,
13287 translate_isl_ast_to_gimple::collect_all_ssa_names,
13288 translate_isl_ast_to_gimple::copy_loop_phi_args,
13289 translate_isl_ast_to_gimple::collect_all_ssa_names,
13290 translate_isl_ast_to_gimple::copy_loop_phi_args,
13291 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13292 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13293 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13294 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13295 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13296 translate_isl_ast_to_gimple::copy_cond_phi_args,
13297 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13298 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13299 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13300 translate_isl_ast_to_gimple::rename_uses,
13301 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13302 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13303 (set_rename_for_each_def): Likewise.
13304 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13305 here. Handle rewriting SCEV analyzable uses here.
13306 (copy_bb_and_scalar_dependences): Generate code for PHI
13308 (graphite_regenerate_ast_isl): Adjust.
13309 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13310 (add_write, add_read): New functions.
13311 (build_cross_bb_scalars_def): Use it and simplify.
13312 (build_cross_bb_scalars_use): Likewise.
13313 (graphite_find_cross_bb_scalar_vars): Inline into...
13314 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13315 simulating out-of-SSA. Compute liveout and add dependencies.
13316 (build_scops): Force an empty entry block.
13317 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13319 (sese_build_liveouts): Declare.
13320 (sese_trivially_empty_bb_p): Likewise.
13321 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13322 compute liveout and debug_liveout.
13323 (sese_bad_liveouts_use): Remove.
13324 (sese_reset_debug_liveouts_bb): Likewise.
13325 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13326 (sese_build_liveouts): Build liveout and debug_liveout and store
13328 (new_sese_info): Adjust.
13329 (free_sese_info): Likewise.
13330 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13331 do not build liveout here.
13332 (move_sese_in_condition): Adjust region entry.
13333 (scev_analyzable_p): Match up with chrec_apply requirements.
13334 (sese_trivially_empty_bb_p): New.
13335 * tree-into-ssa.c (get_reaching_def): Properly support generating
13336 default-defs for incremental rewrite of anonymous names.
13338 2017-10-06 Richard Biener <rguenther@suse.de>
13340 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13341 precision do not perform modulo reduction.
13343 2017-10-06 Richard Biener <rguenther@suse.de>
13345 PR tree-optimization/82436
13346 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13347 conservatively choose the vectorization factor when checking
13348 whether we can perform the required load permutation.
13349 (vect_transform_slp_perm_load): Assert when we may not fail.
13351 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13353 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13354 message for incompatible -msdata=* and -mcall-* options.
13356 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13358 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13359 rate for post-reload scheduling.
13361 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13363 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13365 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13367 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13368 to improve monte carlo in scimark.
13370 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13372 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13373 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13374 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13375 pentium4_cost, nocona_cost): Set reassociation width to 1.
13376 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13377 width to 2 for fp operations and 1 otherwise.
13378 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13380 (atom_cost): Set reassociation width to 2.
13381 (slm_cost, generic_cost): Set fp reassociation width
13382 to 2 and 1 otherwise.
13383 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13384 (core_cost): Set fp reassociation width to 4 and vector to 2.
13385 (ix86_reassociation_width): Rewrite using cost table; special case
13386 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13387 and TARGET_AVX128_OPTIMAL.
13388 * config/i386/i386.h (processor_costs): Add
13389 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13390 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13391 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13392 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13393 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13394 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13396 2017-10-05 Nathan Sidwell <nathan@acm.org>
13398 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13400 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13402 * config/arm/arm.c (arm_test_fpu_data): New.
13403 (arm_run_selftests): Call arm_test_fpu_data.
13405 2017-10-04 Nathan Sidwell <nathan@acm.org>
13407 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13409 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13411 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13413 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13414 check type for aarch64_simd_valid_immediate.
13415 (aarch64_output_simd_mov_immediate): Update prototype.
13416 (aarch64_simd_valid_immediate): Update prototype.
13417 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13418 support for ORR-immediate.
13419 (and<mode>3): modified pattern to add support for BIC-immediate.
13420 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13421 now checks for valid immediate for BIC and ORR based on new enum
13423 (aarch64_output_simd_mov_immediate): Function now used to output
13424 BIC/ORR imm as well based on new enum argument.
13425 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13427 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13428 (aarch64_reg_or_bic_imm): Likewise.
13430 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13432 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13433 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13435 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13439 PR rtl-optimization/82396
13440 * haifa-sched.c (autopref_multipass_init): Simplify
13442 (autopref_rank_data): Simplify sort order.
13443 * sched-int.h (autopref_multipass_data_): Remove
13444 multi_mem_insn_p, min_offset and max_offset.
13446 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13448 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13450 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13452 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13453 vect_doubleint_cvt.
13455 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13457 * doc/sourcebuild.texi: Document vect_long_mult.
13459 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13461 PR tree-optimization/82413
13462 * fold-const.c (build_range_check): Use widest_int when comparing
13463 the maximum ETYPE value with HIGH.
13465 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13467 PR rtl-optimization/82396
13468 * haifa-sched.c (autopref_multipass_init): Simplify
13470 (autopref_rank_data): Simplify sort order.
13471 * sched-int.h (autopref_multipass_data_): Remove
13472 multi_mem_insn_p, min_offset and max_offset.
13474 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13476 PR tree-optimization/82381
13477 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13478 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13481 PR tree-optimization/82374
13482 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13483 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13484 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13485 current_function_decl to the new decl.
13487 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13489 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13490 helper macro for IEEE float128 hardware built-in functions.
13491 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13493 (TRUNCF128_ODD): Likewise.
13494 (ADDF128_ODD): Likewise.
13495 (SUBF128_ODD): Likewise.
13496 (MULF128_ODD): Likewise.
13497 (DIVF128_ODD): Likewise.
13498 (FMAF128_ODD): Likewise.
13499 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13500 UNSPEC_TRUNC_ROUND_TO_ODD.
13501 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13502 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13503 floating point round to odd instructions.
13504 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13505 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13506 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13507 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13508 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13509 (trunc<mode>sf2_hw): Change the truncate with round to odd
13510 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13511 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13512 to odd hardware instructions.
13513 (sub<mode>3_odd): Likewise.
13514 (mul<mode>3_odd): Likewise.
13515 (div<mode>3_odd): Likewise.
13516 (sqrt<mode>2_odd): Likewise.
13517 (fma<mode>4_odd): Likewise.
13518 (fms<mode>4_odd): Likewise.
13519 (nfma<mode>4_odd): Likewise.
13520 (nfms<mode>4_odd): Likewise.
13521 (trunc<mode>df2_odd): Change the truncate with round to odd
13522 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13524 * doc/extend.texi (PowerPC built-in functions): Update documentation
13525 for existing IEEE float128-bit built-in functions. Add built-in
13526 functions that generate the IEEE 128-bit floating point round to
13529 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13531 PR rtl-optimization/77729
13532 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13533 to (X&(C1&~C2))|C2 transformations.
13535 2017-10-03 Martin Jambor <mjambor@suse.cz>
13537 PR tree-optimization/82363
13538 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13539 mismatch, mark lacc written regardless of racc.
13541 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13543 PR tree-optimization/82381
13544 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13545 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13546 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13547 fallthrough into reassoc_stmt_dominates_stmt_p.
13550 * combine.c (combine_instructions): Don't combine in unreachable
13553 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13556 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13557 function to not use the have_cpu variable. Do not set cpu_index,
13558 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13559 or the default cpu.
13560 (rs6000_valid_attribute_p): Remove duplicate initializations of
13561 old_optimize and func_optimize.
13562 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13563 (rs6000_activate_target_options): Make global.
13564 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13567 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13569 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13570 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13571 Set *poffset to 0 and *psize and *pmax_size to -1 if
13572 *poffset + *psize overflows in HOST_WIDE_INT.
13574 PR tree-optimization/82387
13575 PR tree-optimization/82388
13576 PR tree-optimization/82389
13577 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13578 instead of live_bytes non-NULL.
13580 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13583 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13586 2017-10-02 Richard Biener <rguenther@suse.de>
13588 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13589 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13590 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13592 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13594 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13595 requested precision matches the type's.
13596 * calls.c (alloc_max_size): Calculate the new candidate size as
13597 a widest_int and use wi::to_widest when comparing it with the
13598 current candidate size.
13599 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13600 zero rather than integer_zero_node.
13601 * match.pd: Check for a no-op conversion before using wi::add
13602 rather than after. Use tree_to_uhwi when summing small shift
13603 counts into an unsigned int.
13605 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13606 Alan Hayward <alan.hayward@arm.com>
13607 David Sherwood <david.sherwood@arm.com>
13610 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13611 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13612 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13613 POINTER_AND_FP_REGS.
13615 2017-10-02 Richard Biener <rguenther@suse.de>
13617 PR tree-optimization/82355
13618 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13619 a mapping for the enclosing loop but avoid generating one for
13620 the loop tree root.
13621 (copy_bb_and_scalar_dependences): Remove premature codegen
13622 error on PHIs in blocks duplicated into multiple places.
13623 * graphite-scop-detection.c
13624 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13625 in the region use it as loop and nest to analyze the DR in.
13626 (try_generate_gimple_bb): Likewise.
13627 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13628 (add_loop_constraints): For blocks in a loop not in the region
13629 create a dimension with a single iteration.
13630 * sese.h (gbb_loop_at_index): Remove assert.
13632 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13634 * omp-expand.c (adjust_context_scope): New function.
13635 (expand_parallel_call): Call adjust_context_scope.
13637 2017-10-01 Jeff Law <law@redhat.com>
13639 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13640 dom_opt_dom_walker class with direct access to private members.
13641 Add comments. Call test_for_singularity.
13642 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13643 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13644 m_dummy_cond anymore.
13645 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13647 (pass_dominator:execute): Build the dummy_cond here and pass it
13648 to the dom_opt_dom_walker ctor.
13649 (test_for_singularity): New function.
13651 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13652 Maya Rashish <coypu@sdf.org>
13654 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13655 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13656 (alpha*-*-netbsd*) Use nbsd_tm_file.
13657 (arm*-*-netbsdelf*) Likewise.
13658 (i[34567]86-*-netbsdelf*) Likewise.
13659 (x86_64-*-netbsd*) Likewise.
13660 (mips*-*-netbsd*) Likewise.
13661 (powerpc-*-netbsd*) Likewise.
13662 (sh*-*-netbsd*) Likewise.
13663 (sparc-*-netbsdelf*) Likewise.
13664 (sparc64-*-netbsd*) Likewise.
13665 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13667 (vax-*-netbsdelf*) Likewise.
13668 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13669 (UINT_FAST8_TYPE) Likewise.
13670 (INT_FAST16_TYPE) Check CHAR_FAST16.
13671 (UINT_FAST16_TYPE) Likewise.
13673 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13676 * config/i386/i386.md
13677 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13678 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13679 *divmodsi4_zext_2): New define_insn_and_split.
13680 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13681 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13682 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13683 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13684 New define_insn_and_split.
13685 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13686 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13687 operands[1] having DImode when mode is SImode.
13689 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13690 always SImode for DIV and MOD in REG_EQUAL notes.
13692 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13694 PR middle-end/82319
13695 * match.pd: Fix handling of NaNs in pattern.
13697 2017-09-29 Jeff Law <law@redhat.com>
13699 * sbitmap.c (bitmap_bit_in_range_p): New function.
13700 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13701 * tree-ssa-dse.c (live_bytes_read): New function.
13702 (dse_classify_store): Ignore reads of dead bytes.
13704 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13705 typos and whitespace errors.
13706 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13707 * config/s390/s390.c (s390_emit_prologue): Likewise.
13709 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13712 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13713 with a symbol for LRA.
13715 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13717 PR rtl-optimization/82338
13718 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13720 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13722 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13723 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13724 (qsort_chk): Declare.
13725 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13726 (qsort_chk): New function.
13728 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13730 PR tree-optimization/82337
13731 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13732 phi definition if the PHI result appears in an abnormal PHI.
13733 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13734 basis appears in an abnormal PHI.
13736 2017-09-29 Richard Biener <rguenther@suse.de>
13738 * graphite-isl-ast-to-gimple.c
13739 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13740 (binary_op_to_tree): Use it.
13741 (get_rename_from_scev): Likewise.
13742 (copy_loop_phi_nodes): Likewise.
13743 (copy_bb_and_scalar_dependences): Likewise.
13744 (translate_pending_phi_nodes): Likewise.
13746 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13749 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13750 for movabsq $(i32 << shift), r64.
13752 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13754 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13755 index when encoding %esp as %rsp to avoid 0x67 prefix.
13757 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13759 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13760 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13762 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13764 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13765 Extensions with more than 16 double VFP registers.
13766 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13767 to_clear_mask and all code related to it. Replace the remaining
13768 entry by a sbitmap and adapt code accordingly.
13770 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13772 * brig-builtins.def: Change pure attributes to const.
13774 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13776 * config.gcc (default_gnu_indirect_function): Default to yes for
13777 sparc*-*-linux* with glibc.
13779 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13781 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13782 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13783 when creating .init_array and .fini_array sections with priority
13786 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13789 * config/aarch64/aarch64.c
13790 (aarch64_builtin_support_vector_misalignment): Always return false
13791 when misalignment is unknown.
13793 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13795 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
13796 this function to return false if the definition used by the swap
13797 instruction is artificial, or if the memory address from which the
13798 constant value is loaded is not represented by a base address held
13799 in a register or if the base address register is a frame or stack
13800 pointer. Additionally, return false if the base address of the
13801 loaded constant is a SYMBOL_REF but is not considered to be a
13803 (replace_swapped_load_constant): New function.
13804 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
13805 loaded constant vector with a load of a swapped constant vector.
13807 2017-09-27 Carl Love <cel@us.ibm.com>
13809 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
13810 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
13811 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
13814 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
13816 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
13817 first, always call autopref_rank_data otherwise.
13819 2017-09-27 Richard Biener <rguenther@suse.de>
13821 * graphite-scop-detection.c (find_scop_parameters): Move
13822 loop bound handling ...
13823 (gather_bbs::before_dom_children): ... here, avoiding the need
13824 to build scop_info->loop_nest.
13825 (record_loop_in_sese): Remove.
13826 * sese.h (sese_info_t::loop_nest): Remove.
13827 * sese.c (new_sese_info): Do not allocate loop_nest.
13828 (free_sese_info): Do not free loop_nest.
13830 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13833 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13834 lhs from calls if the lhs has addressable type.
13836 2017-09-27 Richard Biener <rguenther@suse.de>
13838 * graphite.h (scop::max_alias_set): New member.
13839 * graphite-scop-detection.c: Remove references to non-existing
13840 --param in comments.
13841 (build_alias_sets): Record the maximum alias set used for drs.
13842 (build_scops): Support zero as unlimited for
13843 --param graphite-max-arrays-per-scop.
13844 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
13845 and inline into ...
13846 (build_poly_sr_1): ... here. Compute alias set based on the
13847 maximum alias set used for drs rather than
13848 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
13850 2017-09-27 Richard Biener <rguenther@suse.de>
13852 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13853 --param loop-block-tile-size=0 to disable tiling.
13855 2017-09-27 Richard Biener <rguenther@suse.de>
13857 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
13858 (graphite-max-nb-scop-params): Document special value zero.
13859 * domwalk.h (dom_walker::STOP): New symbolical constant.
13860 (dom_walker::dom_walker): Add optional parameter for bb to
13862 (dom_walker::~dom_walker): Declare.
13863 (dom_walker::before_dom_children): Document STOP return value.
13864 (dom_walker::m_user_bb_to_rpo): New member.
13865 (dom_walker::m_bb_to_rpo): Likewise.
13866 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
13867 mapping here if not provided by the user.
13868 (dom_walker::~dom_walker): Free bb to RPO mapping if not
13869 provided by the user.
13870 (dom_walker::STOP): Define.
13871 (dom_walker::walk): Do not compute bb to RPO mapping here.
13872 Support STOP return value from before_dom_children to stop
13874 * graphite-optimize-isl.c (optimize_isl): If the schedule
13875 is the same still generate code if -fgraphite-identity
13876 or -floop-parallelize-all are given.
13877 * graphite-scop-detection.c: Include cfganal.h.
13878 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13880 (gather_bbs::before_dom_children): Return STOP for BBs
13882 (build_scops): Compute bb to RPO mapping and pass it to
13883 the domwalk. Treat --param graphite-max-nb-scop-params=0
13884 as not limiting the number of params.
13885 * graphite.c (graphite_initialize): Remove limit on the
13886 number of basic-blocks in a function.
13887 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13888 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13889 default value of 10.
13891 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13893 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13894 Adjust code to eliminate needing to do the shift right 32-bits
13895 operation after XSCVDPSPN.
13897 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13899 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13900 ((X / Y) != 0 -> X >= Y): Likewise.
13902 2017-09-26 Carl Love <cel@us.ibm.com>
13904 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13905 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13906 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13907 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13908 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13909 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13910 definitions and overloading.
13911 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13912 statement for P9V_BUILTIN_XST_LEN_R.
13913 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13914 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13915 define_expand and define_insn for the instructions and builtins.
13916 * doc/extend.texi: Update the built-in documentation file for the new
13917 built-in functions.
13918 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13919 define_insn for the instructions
13921 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13924 * gcc/config/netbsd-protos.h: New file.
13925 * gcc/config/netbsd.c: New file.
13926 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13927 * gcc/config/t-netbsd: New file.
13928 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13929 (tmake_file) Add t-netbsd.
13930 (extra_objs) Add netbsd.o.
13932 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13936 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13938 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13940 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13941 sign extension from a vector register to a GPR by doing a 32-bit
13942 direct move and then an EXTSW.
13943 (extendsi<mode>2 splitter): Likewise.
13944 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13945 right or vector extract after doing XSCVDPSPN. Use
13946 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13948 (movdi_from_sf_zero_ext): Likewise.
13949 (reload_gpr_from_vsxsf): Likewise.
13950 (p8_mfvsrd_4_disf): Delete, no longer used.
13951 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13952 then needing to move the SFmode to a GPR to use the XSCVDPSP
13953 instruction instead of FRSP and XSCVDPSPN.
13954 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13955 it is adjacent to the other XSCVSPDP insns.
13956 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13957 SFmode to be in traditional Altivec registers.
13958 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13959 (vsx_xscvspdpn): Likewise.
13960 (vsx_xscvspdpn_scalar): Likewise.
13962 2017-09-26 Martin Jambor <mjambor@suse.cz>
13964 * tree-sra.c (compare_access_positions): Put integral types first,
13965 stabilize sorting of integral types, remove conditions putting
13966 non-full-precision integers last.
13967 (sort_and_splice_var_accesses): Disable scalarization if a
13968 non-integert would be represented by a non-full-precision integer.
13970 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13972 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13973 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13974 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13975 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13976 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13977 conditionals inside the function instead of around it. Call
13978 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13979 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13981 2017-09-26 Richard Biener <rguenther@suse.de>
13983 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13985 (scop_detection::build_scop_breadth): ... this. Removed.
13986 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13987 (scop_detection::harmful_stmt_in_bb): Likewise.
13988 (scop_detection::graphite_can_represent_stmt): Likewise.
13989 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13990 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13991 (scop_detection::can_represent_loop_1): ... this. Removed.
13992 (scop_detection::harmful_loop_in_region): Simplify after inlining
13993 the above and remove more quadraticness.
13994 (build_scops): Adjust.
13995 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13998 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14001 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14002 REGNO (base) == SP_REG if base is a REG.
14004 PR middle-end/35691
14005 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14006 if it is different SSA_NAME.
14007 (optimize_range_tests_cmp_bitwise): New function.
14008 (optimize_range_tests): Call it.
14010 2017-09-26 Richard Biener <rguenther@suse.de>
14012 PR tree-optimization/82321
14013 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14014 for the def being inside the loop.
14016 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14018 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14020 * config/s390/s390-builtins.def: Fix constraint on op4.
14022 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14024 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14025 independent expanders.
14026 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14027 ("vec_ordered", "vec_unordered"): New expanders.
14029 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14031 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14034 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14036 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14037 vec_unpacks_lo_v16qi.
14038 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14040 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14042 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14043 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14044 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14046 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14048 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14050 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14052 ("*vec_slb<mode>"): New insn pattern.
14053 ("vec_shr_<mode>"): New expander.
14054 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14055 and force the shift count operand to V16QImode.
14056 ("vec_srb<mode>"): Set shift count mode to V16QI.
14058 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14060 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14061 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14062 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14064 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14067 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14069 2017-09-26 Richard Biener <rguenther@suse.de>
14071 PR tree-optimization/82320
14072 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14075 2017-09-25 Jeff Law <law@redhat.com>
14077 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14078 prototype for new argument.
14079 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14080 mostly extracted from rs6000_emit_allocate_stack.
14081 (rs6000_emit_probe_stack_range_stack_clash): New function.
14082 (rs6000_emit_allocate_stack): Call
14083 rs6000_emit_probe_stack_range_stack_clash as needed.
14084 (rs6000_emit_probe_stack_range): Add additional argument
14085 to call to gen_probe_stack_range{si,di}.
14086 (output_probe_stack_range): New.
14087 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14088 (output_probe_stack_range_stack_clash): New.
14089 (rs6000_emit_prologue): Emit notes into dump file as requested.
14090 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14091 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14092 Add additional operand and pass it to output_probe_stack_range.
14094 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14096 PR tree-optimization/82163
14097 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14098 (checking_verify_loop_closed_ssa): New parameter.
14099 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14100 (check_loop_closed_ssa_stmt): Delete.
14101 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14102 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14103 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14106 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14108 * brig-builtins.def: Treat HSAIL barrier builtins as
14109 setjmp/longjump style functions.
14111 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14113 * target.def (constant_alignment): New hook.
14114 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14115 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14116 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14117 * doc/tm.texi: Regenerate.
14118 * targhooks.h (default_constant_alignment): Declare.
14119 (constant_alignment_word_strings): Likewise.
14120 * targhooks.c (default_constant_alignment): New function.
14121 (constant_alignment_word_strings): Likewise.
14122 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14123 instead of CONSTANT_ALIGNMENT.
14124 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14125 (force_const_mem): Likewise.
14126 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14127 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14128 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14129 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14130 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14132 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14133 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14134 constant_alignment_word_strings.
14135 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14136 (CONSTANT_ALIGNMENT): Likewise.
14137 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14138 (arm_constant_alignment): New function.
14139 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14140 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14141 constant_alignment_word_strings.
14142 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14143 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14144 constant_alignment_word_strings.
14145 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14146 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14147 (cris_constant_alignment): New function.
14148 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14149 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14150 (epiphany_constant_alignment): New function.
14151 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14152 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14153 constant_alignment_word_strings.
14154 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14155 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14156 constant_alignment_word_strings.
14157 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14158 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14159 constant_alignment_word_strings.
14160 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14161 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14162 * config/i386/i386.c (ix86_constant_alignment): Make static.
14163 Use the same interface as the target hook.
14164 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14165 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14166 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14167 constant_alignment_word_strings.
14168 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14169 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14170 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14171 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14172 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14173 constant_alignment_word_strings.
14174 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14175 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14176 constant_alignment_word_strings.
14177 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14178 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14179 constant_alignment_word_strings.
14180 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14181 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14183 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14184 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14185 * config/mips/mips.c (mips_constant_alignment): New function.
14186 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14187 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14188 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14189 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14190 (mmix_constant_alignment): Make static. Use the same interface
14191 as the target hook.
14192 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14193 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14194 constant_alignment_word_strings.
14195 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14196 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14197 constant_alignment_word_strings.
14198 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14199 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14200 constant_alignment_word_strings.
14201 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14202 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14203 (rs6000_constant_alignment): New function.
14204 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14205 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14206 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14207 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14208 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14209 (rs6000_constant_alignment): New function.
14210 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14211 * config/s390/s390.c (s390_constant_alignment): New function.
14212 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14213 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14214 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14215 constant_alignment_word_strings.
14216 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14217 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14218 (sparc_constant_alignment): New function.
14219 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14220 * config/spu/spu.c (spu_constant_alignment): New function.
14221 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14222 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14223 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14224 constant_alignment_word_strings.
14225 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14226 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14227 constant_alignment_word_strings.
14228 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14229 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14230 constant_alignment_word_strings.
14231 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14232 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14233 (visium_constant_alignment): New function.
14234 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14235 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14236 (xtensa_constant_alignment): New function.
14237 * system.h (CONSTANT_ALIGNMENT): Poison.
14239 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14241 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14242 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14243 (rs6000_builtin_valid_without_lhs): New helper function.
14244 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14245 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14247 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14249 * target.h (vec_perm_indices): Use unsigned short rather than
14251 (auto_vec_perm_indices): Likewise.
14252 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14253 Use unsigned int rather than unsigned char.
14254 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14256 2017-09-25 Richard Biener <rguenther@suse.de>
14258 * cfgloop.h (sort_sibling_loops): Declare.
14259 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14260 (sort_sibling_loops): New function sorting the sibling loop list
14262 * graphite.c (graphite_transform_loops): Sort sibling loops.
14264 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14266 * target.def (vec_perm_const_ok): Change sel parameter to
14268 * optabs-query.c (can_vec_perm_p): Update accordingly.
14269 * doc/tm.texi: Regenerate.
14270 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14271 auto_vec_perm_indices and remove separate nelt field.
14272 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14273 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14274 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14275 (aarch64_expand_vec_perm_const): Update accordingly.
14276 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14277 to vec_perm_indices.
14278 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14279 auto_vec_perm_indices and remove separate nelt field.
14280 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14281 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14282 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14284 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14285 to vec_perm_indices.
14286 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14287 sel to vec_perm_indices.
14288 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14289 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14290 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14292 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14295 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14298 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14299 on the FUNCTION_DECL function context if it has a DIE that is a
14302 2017-09-25 Richard Biener <rguenther@suse.de>
14304 PR tree-optimization/82285
14305 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14308 2017-09-25 Tom de Vries <tom@codesourcery.com>
14312 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14315 2017-09-25 Richard Biener <rguenther@suse.de>
14317 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14318 ISL errors other than isl_error_quota happen. Dump if the
14319 schedule is the same.
14320 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14321 errors instead of aborting inside ISL.
14323 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14326 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14327 of libgcc_eh for m64.
14328 * config/i386/darwin64.h: Likewise.
14330 2017-09-25 Richard Biener <rguenther@suse.de>
14332 PR middle-end/82144
14333 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14334 attribute for incomplete types nor twice for complete ones.
14336 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14339 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14340 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14343 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14346 * config/i386/i386.opt (mprefer-avx256): Use
14347 ix86_target_flags variable.
14348 * config/i386/i386.c (ix86_target_string): Move
14349 -mprefer-avx256 to flag2_opts.
14351 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14353 PR middle-end/35691
14354 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14355 and x != -1 | y != -1 into (x & y) != -1.
14357 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14359 * config.gcc: Add new case statement to set
14360 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14361 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14362 s390x-*-linux* case statements. Added aarch64 to the list of
14363 supported architectures.
14365 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14367 PR tree-optimization/82289
14368 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14369 STMT_VINFO_RELEVANT_P.
14371 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14372 Alan Hayward <alan.hayward@arm.com>
14373 David Sherwood <david.sherwood@arm.com>
14375 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14376 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14377 (extract_range_from_binary_expr_1): Don't call
14378 extract_range_from_multiplicative_op_1 if !range_int_cst_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 * target.def (preferred_vector_alignment): New hook.
14385 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14387 * doc/tm.texi: Regenerate.
14388 * targhooks.h (default_preferred_vector_alignment): Declare.
14389 * targhooks.c (default_preferred_vector_alignment): New function.
14390 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14392 (DR_TARGET_ALIGNMENT): New macro.
14393 (aligned_access_p): Update commentary.
14394 (vect_known_alignment_in_bytes): New function.
14395 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14397 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14398 Calculate the misalignment based on the target alignment rather than
14400 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14401 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14402 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14403 the target alignment, rather than masking the element misalignment
14404 with the number of elements in a vector. Also use the target
14405 alignment when calculating the maximum number of peels.
14406 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14407 instead of TYPE_ALIGN_UNIT.
14408 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14409 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14410 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14411 (vect_create_data_ref_ptr): Likewise.
14412 (vect_setup_realignment): Realign by ANDing with
14413 -DR_TARGET_MISALIGNMENT.
14414 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14415 the number of peels based on DR_TARGET_ALIGNMENT.
14416 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14417 with the guaranteed alignment boundary when deciding whether
14419 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14420 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14421 (ensure_base_align): Remove stmt_info parameter. Get the
14422 target base alignment from DR_TARGET_ALIGNMENT.
14423 (vectorizable_store): Update call accordingly. Interpret
14424 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14426 (vectorizable_load): Likewise.
14428 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14429 Alan Hayward <alan.hayward@arm.com>
14430 David Sherwood <david.sherwood@arm.com>
14432 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14433 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14434 (vect_enhance_data_refs_alignment): Likewise.
14436 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14438 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14439 error. Only quit immediately if parsing is complete.
14440 (BEGIN): Initialize fatal_err and parse_done.
14441 (begin fpu, end fpu): Check number of arguments.
14442 (begin arch, end arch): Likewise.
14443 (begin cpu, end cpu): Likewise.
14444 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14445 (optalias): Likewise.
14447 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14449 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14450 * config/arm/arm-isa.h: Delete. Move definitions to ...
14451 * arm-cpus.in: ... here. Use new feature and fgroup values.
14452 * config/arm/arm.c (arm_option_override): Use lower case for feature
14454 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14455 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14456 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14458 (print_isa_bits_for): New function.
14459 (gen_isa): New function.
14460 (gen_comm_data): Use print_isa_bits_for.
14461 (define feature): New keyword.
14462 (define fgroup): New keyword.
14463 * config/arm/t-arm (TM_H): Remove.
14464 (GTM_H): Add arm-isa.h.
14465 (arm-isa.h): Add rule to generate file.
14466 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14467 case for feature bit names.
14469 2017-09-22 Richard Biener <rguenther@suse.de>
14471 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14473 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14474 print of no dependency loops ...
14475 * graphite.c (graphite_transform_loops): ... here.
14476 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14478 (same_close_phi_node, remove_duplicate_close_phi,
14479 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14480 (canonicalize_loop_closed_ssa): ... here and simplify.
14481 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14482 (optimize_isl): Use dump_printf_loc to tell when we stopped
14483 optimizing because of an ISL timeout.
14485 2017-09-22 Richard Biener <rguenther@suse.de>
14487 PR tree-optimization/82291
14488 * tree-if-conv.c (predicate_mem_writes): Make sure to
14489 remove writes in blocks predicated with false.
14491 2017-09-22 Richard Biener <rguenther@suse.de>
14493 * sese.c: Include cfganal.h.
14494 (if_region_set_false_region): Remove.
14495 (create_if_region_on_edge): Likewise.
14496 (move_sese_in_condition): Re-implement without destroying
14499 2017-09-22 Richard Biener <rguenther@suse.de>
14501 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14502 Verify both BBs contain loop PHI nodes before dispatching to
14503 copy_loop_phi_args.
14504 (graphite_regenerate_ast_isl): Do not recompute dominators,
14505 do not verify three times. Restructure for clarity.
14506 * graphite-scop-detection.c (same_close_phi_node,
14507 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14508 defined_in_loop_p, canonicalize_loop_closed_ssa,
14509 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14510 checking and SSA rewrite, move to ...
14511 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14512 (graphite_initialize): Do not pass in ctx, do not reset the
14513 SCEV cache, compute only dominators.
14514 (graphite_transform_loops): Allocate ISL ctx after
14515 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14516 Maintain post-dominators only around build_scops.
14517 * sese.c (if_region_set_false_region): Make static. Free
14518 and recompute dominators.
14519 (move_sese_in_condition): Assert we don't get called with
14520 post-dominators computed.
14521 * sese.h (if_region_set_false_region): Remove.
14523 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14525 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14526 mode attribute for TARGET_AVX512VL.
14528 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14530 * config/i386/i386.opt (mprefer-avx256): New option.
14531 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14533 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14534 for TARGET_PREFER_AVX256.
14535 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14537 2017-09-21 Jeff Law <law@redhat.com>
14539 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14540 Fix dump output if the only stack space is for pushed registers.
14542 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14544 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14547 2017-09-21 Martin Sebor <msebor@redhat.com>
14550 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14551 code (in C++) or code that triggers warnings.
14553 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14555 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14557 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14559 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14560 * sched-rgn.c: Ditto.
14561 * sel-sched-ir.c: Ditto.
14563 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14565 * toplev.h (set_random_seed): Adjust return type.
14566 * toplev.c (init_local_tick): Move eager initialization of random_seed
14567 to get_random_seed. Adjust comment.
14568 (init_random_seed): Inline to get_random_seed, delete.
14569 (get_random_seed): Initialize random_seed lazily.
14570 (set_random_seed): Do not return previous value.
14571 (print_switch_value): Do not call get_random_seed.
14573 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14575 * cgraph.c (delete_function_version): New, broken out from...
14576 (cgraph_node::delete_function_version): ...here. Rename to
14577 cgraph_node::delete_function_version_by_decl. Update all uses.
14578 (cgraph_node::remove): Call delete_function_version.
14580 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14583 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14584 VAR_DECLs to which addressable non-volatile parameters are mapped
14585 and for id->retvar after the return value assignment. Clear
14586 id->retval and id->retbnd after inlining.
14588 2017-09-21 Richard Biener <rguenther@suse.de>
14590 PR tree-optimization/82276
14591 PR tree-optimization/82244
14592 * tree-vrp.c (build_assert_expr_for): Set
14593 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14595 (remove_range_assertions): Revert earlier change.
14597 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14600 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14602 2017-09-21 Richard Biener <rguenther@suse.de>
14604 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14605 Restore valid IL after code generation errors.
14606 * graphite.c (graphite_transform_loops): Diagnose code
14607 generation issues as MSG_MISSED_OPTIMIZATION and continue
14608 with processing SCOPs.
14610 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14611 Alan Hayward <alan.hayward@arm.com>
14612 David Sherwood <david.sherwood@arm.com>
14614 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14615 rather than choosing between plus_constant and gen_rtx_<CODE>.
14616 * expr.c (emit_push_insn): Likewise.
14617 (expand_expr_real_2): Likewise.
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 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14625 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14626 Alan Hayward <alan.hayward@arm.com>
14627 David Sherwood <david.sherwood@arm.com>
14629 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14630 calling tree_to_uhwi.
14632 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14633 Alan Hayward <alan.hayward@arm.com>
14634 David Sherwood <david.sherwood@arm.com>
14636 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14637 INTEGER_CST rather than a negative test for ADDR_EXPR.
14639 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14640 Alan Hayward <alan.hayward@arm.com>
14641 David Sherwood <david.sherwood@arm.com>
14643 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14644 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14646 2017-09-21 Richard Biener <rguenther@suse.de>
14648 PR tree-optimization/71351
14649 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14650 graphite_create_new_loop_guard): Remove, fold remaining parts
14652 (translate_isl_ast_node_for): ... here and simplify.
14654 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14657 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14658 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14659 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14660 alternative always use QI mode, for -Os imov (=R,R) alternative
14661 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14664 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14665 Jeff Law <law@redhat.com>
14667 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14668 (allocate_stack_space): New function, partially extracted from
14669 s390_emit_prologue.
14670 (s390_emit_prologue): Track offset to most recent stack probe.
14671 Code to allocate space moved into allocate_stack_space.
14672 Dump actions when no stack is allocated.
14673 (s390_prologue_plus_offset): New function.
14674 (s390_emit_stack_probe): Likewise.
14676 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14678 * common.opt (Wa, Wl, Wp, g, gz=): Add
14680 (gno-column-info): Remove.
14681 (gcolumn-info): Drop RejectNegative.
14682 (gno-): New prefix.
14683 (gno-record-gcc-switches): Remove.
14684 (grecord-gcc-switches): Drop RejectNegative.
14685 (gno-split-dwarf): Remove.
14686 (gsplit-dwarf): Drop RejectNegative.
14687 (gno-strict-dwarf): Remove.
14688 (gstrict-dwarf): Drop RejectNegative.
14689 * config/darwin.opt (gfull, gused): Add RejectNegative.
14690 * dwarf2out.c (gen_producer_string): Drop
14691 gno-record-gcc-switches handler.
14692 * optc-gen.awk: Add g to prefixes with negative forms.
14693 * opts-common.c (remapping_prefix_p): New.
14694 (find_opt): Check it.
14695 (generate_canonical_option): Test g prefix.
14696 (option_map): Add -gno- mapping.
14697 (add_misspelling_candidates): Check remapping_prefix_p.
14699 2017-09-20 Jeff Law <law@redhat.com>
14701 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14702 thinko in stack clash protection support.
14704 * explow.c (compute_stack_clash_protection_loop_data): Use
14705 CONST_INT_P instead of explicit test. Verify object is a
14706 CONST_INT_P before looking at INTVAL.
14707 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14708 instead of explicit test.
14710 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14713 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14714 address instead of to r1 and r11.
14716 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14718 * config.gcc: Support "knm".
14719 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14720 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14722 * config/i386/i386.c (m_KNM): Define.
14723 (processor_target_table): Add "knm".
14725 (ix86_option_override_internal): Add "knm".
14726 (ix86_issue_rate): Add PROCESSOR_KNM.
14727 (ix86_adjust_cost): Ditto.
14728 (ia32_multipass_dfa_lookahead): Ditto.
14729 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14730 (fold_builtin_cpu): Add M_INTEL_KNM.
14731 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14732 (processor_type): Add PROCESSOR_KNM.
14733 * config/i386/x86-tune.def: Add m_KNM.
14734 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14736 2017-09-20 Richard Biener <rguenther@suse.de>
14738 PR tree-optimization/80213
14739 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14740 are allowed in empty BBs as well.
14741 (canonicalize_loop_closed_ssa): Also look for other complex
14743 (scop_detection::get_sese): Include the loop-closed PHI block
14745 (scop_detection::merge_sese): Remove code adding extra blocks.
14746 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14747 (build_scops): Assert the final returned scop is invalid.
14749 2017-09-20 Richard Biener <rguenther@suse.de>
14751 PR tree-optimization/82264
14752 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14754 (vn_phi_lookup): Likewise.
14755 (vn_phi_insert): Likewise.
14757 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14759 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14760 that fits into uhwi or shwi, add DW_AT_const_value regardless
14761 of early_dwarf without going through RTL, using add_AT_unsigned
14764 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14765 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14766 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14767 (DEBUG_MACRO_SECTION): Likewise.
14768 (DEBUG_LTO_MACRO_SECTION): Likewise.
14769 (DEBUG_STR_DWO_SECTION): Likewise.
14770 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14771 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14772 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14773 (DEBUG_LTO_LINE_STR_SECTION): Define.
14774 (init_sections_and_labels): Initialize debug_line_str_section
14775 variable. Initialize debug_loc_section for -gdwarf-5 to
14776 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14778 2017-09-20 Richard Biener <rguenther@suse.de>
14780 * graphite-sese-to-poly.c (extract_affine): Properly handle
14781 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14783 2017-09-20 Richard Biener <rguenther@suse.de>
14785 PR tree-optimization/81373
14786 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14787 Force SESE live-out defs to be handled even if they are
14790 2017-09-19 Jeff Law <law@redhat.com>
14792 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
14793 nothing for stack adjustments with REG_STACK_CHECK.
14794 * sched-deps.c (parse_add_or_inc): Reject insns with
14795 REG_STACK_CHECK from dependency breaking.
14796 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
14797 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
14798 * reg-notes.def (STACK_CHECK): New note.
14800 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
14801 (ix86_expand_prologue): Dump stack clash info as needed.
14802 Call ix86_adjust_stack_and_probe_stack_clash as needed.
14804 * function.c (dump_stack_clash_frame_info): New function.
14805 * function.h (dump_stack_clash_frame_info): Prototype.
14806 (enum stack_clash_probes): New enum.
14808 * config/alpha/alpha.c (alpha_expand_prologue): Also check
14809 flag_stack_clash_protection.
14810 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14811 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
14812 (arm_frame_pointer_required): Likewise.
14813 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
14814 (ia64_expand_prologue): Likewise.
14815 * config/mips/mips.c (mips_expand_prologue): Likewise.
14816 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
14817 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14818 (sparc_flat_expand_prologue): Likewise.
14819 * config/spu/spu.c (spu_expand_prologue): Likewise.
14821 * explow.c: Include "params.h".
14822 (anti_adjust_stack_and_probe_stack_clash): New function.
14823 (get_stack_check_protect): Likewise.
14824 (compute_stack_clash_protection_loop_data): Likewise.
14825 (emit_stack_clash_protection_loop_start): Likewise.
14826 (emit_stack_clash_protection_loop_end): Likewise.
14827 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14828 Use anti_adjust_stack_and_probe_stack_clash.
14829 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14830 (emit_stack_clash_protection_loop_start): Likewise.
14831 (emit_stack_clash_protection_loop_end): Likewise.
14832 * rtl.h (get_stack_check_protect): Prototype.
14833 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14834 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14835 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14837 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14839 * doc/tm.texi: Rebuilt.
14840 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14841 get_stack_check_protect.
14842 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
14843 * config/arm/arm.c (arm_expand_prologue): Likewise.
14844 (arm_frame_pointer_required): Likewise.
14845 * config/i386/i386.c (ix86_expand_prologue): Likewise.
14846 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
14847 * config/mips/mips.c (mips_expand_prologue): Likewise.
14848 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
14849 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
14850 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14851 (sparc_flat_expand_prologue): Likewise.
14853 * common.opt (-fstack-clash-protection): New option.
14854 * flag-types.h (enum stack_check_type): Note difference between
14855 -fstack-check= and -fstack-clash-protection.
14856 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
14857 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
14858 * toplev.c (process_options): Issue warnings/errors for cases
14859 not handled with -fstack-clash-protection.
14860 * doc/invoke.texi (-fstack-clash-protection): Document new option.
14861 (-fstack-check): Note additional problem with -fstack-check=generic.
14862 Note that -fstack-check is primarily for Ada and refer users
14863 to -fstack-clash-protection for stack-clash-protection.
14864 Document new params for stack clash protection.
14866 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
14868 * config/i386/i386.c (ix86_split_long_move): Do not handle
14869 address used for LEA in a special way.
14871 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
14873 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14875 2017-09-19 Martin Sebor <msebor@redhat.com>
14878 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14879 of incompatible types.
14881 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14883 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14884 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14885 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14886 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14888 2017-09-19 Richard Biener <rguenther@suse.de>
14890 PR tree-optimization/82244
14891 * tree-vrp.c (remove_range_assertions): Do not propagate
14892 a constant to abnormals but replace the assert with a copy.
14894 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14896 PR rtl-optimization/57878
14897 PR rtl-optimization/68988
14898 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14899 avoidance test involving non_reload_pseudos. Move frequency test
14900 below the general fragmentation avoidance test.
14902 2017-09-19 Richard Biener <rguenther@suse.de>
14904 PR tree-optimization/69728
14905 * graphite-sese-to-poly.c (schedule_error): New global.
14906 (add_loop_schedule): Handle empty domain by failing the
14908 (build_original_schedule): Handle schedule_error.
14910 2017-09-19 Richard Biener <rguenther@suse.de>
14912 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14913 Do not iterate to sibling loops but only to siblings of inner
14916 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14919 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14920 that the registers are different.
14922 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14924 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14925 to processor_model and "amdfam17h" to arch_names_table.
14926 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14928 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14931 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14933 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14934 Alan Hayward <alan.hayward@arm.com>
14935 David Sherwood <david.sherwood@arm.com>
14937 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14939 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14941 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14942 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14943 vect_schedule_slp_instance.
14944 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14945 Update call to vect_slp_analyze_node_operations. Simplify return
14947 (vect_slp_analyze_bb_1): Update call accordingly.
14948 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14949 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14950 (vect_schedule_slp): Update call accordingly.
14952 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14953 Alan Hayward <alan.hayward@arm.com>
14954 David Sherwood <david.sherwood@arm.com>
14956 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14957 with types that aren't in fact scalar.
14959 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14961 * tree-vect-slp.c (vect_record_max_nunits): New function,
14963 (vect_build_slp_tree_1): ...here.
14964 (vect_build_slp_tree_2): Call it for phis too.
14966 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14968 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14969 to vect_get_vec_def_for_operand when getting the mask operand.
14971 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14972 Alan Hayward <alan.hayward@arm.com>
14973 David Sherwood <david.sherwood@arm.com>
14975 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14978 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14979 Alan Hayward <alan.hayward@arm.com>
14980 David Sherwood <david.sherwood@arm.com>
14982 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14983 calculation for vector booleans.
14985 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14986 Alan Hayward <alan.hayward@arm.com>
14987 David Sherwood <david.sherwood@arm.com>
14989 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14991 (vect_transform_stmt): ...here.
14992 (vect_analyze_stmt): Use it instead of calling
14993 vectorizable_live_operation directly.
14995 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14997 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14998 non-SIMT targets in acc vector loops.
15000 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15002 * configure.ac: Add arc and check if assembler supports gdwarf2.
15003 * configure: Regenerate.
15005 2017-09-18 Richard Biener <rguenther@suse.de>
15007 PR tree-optimization/82220
15008 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15009 epilogue niters from the min_profitable_iters compute.
15011 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15014 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15015 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15016 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15018 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15021 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15022 switching to a new text section.
15024 2017-09-18 Richard Biener <rguenther@suse.de>
15026 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15028 (build_alias_set): Reject aliases with no access function.
15030 2017-09-18 Richard Biener <rguenther@suse.de>
15032 PR tree-optimization/79622
15033 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15035 (build_cross_bb_scalars_use): Likewise.
15037 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15039 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15041 2017-09-18 Alan Modra <amodra@gmail.com>
15044 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15045 stack_pointer_rtx for count 0. Update comments. Break up
15046 large rtl expression.
15048 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15050 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15051 Increase to 20 bytes.
15052 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15053 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15054 or avx version of the stub.
15056 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15059 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15060 compute the minimum stack alignment. Also update preferred stack
15061 boundary for leaf functions.
15063 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15065 PR tree-optimization/82228
15066 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15069 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15071 * common/config/nds32/nds32-common.c
15072 (nds32_option_optimization_table): Refine formatting.
15073 (nds32_option_optimization_table): Use -fsched-pressure and
15074 -fomit-frame-pointer for specific optimization level.
15076 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15078 * config/nds32/nds32.c: Refine formatting and comments.
15079 * config/nds32/nds32.h: Likewise.
15080 * config/nds32/nds32.md: Likewise.
15081 * config/nds32/nds32-cost.c: Likewise.
15082 * config/nds32/nds32-isr.c: Likewise.
15083 * config/nds32/nds32-md-auxiliary.c: Likewise.
15084 * config/nds32/nds32-multiple.md: Likewise.
15085 * config/nds32/nds32-predicates.c: Likewise.
15087 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15088 Jakub Jelinek <jakub@redhat.com>
15090 Add support for -std=c++2a.
15091 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15093 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15095 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15098 * doc/extend.texi (Common Function Attributes): Add
15099 references to ARM, AArch64, and S/390 specific attributes.
15100 (Function Specific Option Pragmas): Add AArch64 and S/390
15101 to list of back ends that support the target pragma.
15103 2017-09-15 Nathan Sidwell <nathan@acm.org>
15105 * doc/standards.texi: Fix C++17 description. Update URLs for
15108 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15110 * common.opt (Wcast-align=strict): New warning option.
15111 * doc/invoke.texi: Document -Wcast-align=strict.
15113 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15115 * cgraph.h (cgraph_thunk_info): Add comments.
15116 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15117 assert for VIRTUAL_* arguments stricter.
15119 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15121 PR tree-optimization/71026
15122 * match.pd: Move RDIV patterns from fold-const.c
15123 * fold-const.c (distribute_real_division): Removed.
15124 (fold_binary_loc): Remove calls to distribute_real_divison.
15126 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15128 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15129 c++1z and gnu++1z as deprecated. Change other references to
15130 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15131 Change -Wc++1z-compat to -Wc++17-compat.
15132 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15133 * dwarf2out.c (highest_c_language): Handle C++17.
15134 (gen_compile_unit_die): Likewise.
15136 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15138 PR rtl-optimization/82192
15139 * combine.c (make_extraction): Don't look through non-paradoxical
15140 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15143 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15144 Alan Hayward <alan.hayward@arm.com>
15145 David Sherwood <david.sherwood@arm.com>
15147 * target.def (function_arg_offset): New hook.
15148 * targhooks.h (default_function_arg_offset): Declare.
15149 * targhooks.c (default_function_arg_offset): New function.
15150 * function.c (locate_and_pad_parm): Use
15151 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15152 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15153 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15154 * doc/tm.texi: Regenerate.
15155 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15156 * config/spu/spu.c (spu_function_arg_offset): New function.
15157 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15158 * system.h (FUNCTION_ARG_OFFSET): Poison.
15160 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15161 Alan Hayard <alan.hayward@arm.com>
15162 David Sherwood <david.sherwood@arm.com>
15164 * target.def (truly_noop_truncation): New hook.
15165 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15166 than TRULY_NOOP_TRUNCATION.
15167 * hooks.h (hook_bool_uint_uint_true): Declare.
15168 * hooks.c (hook_bool_uint_uint_true): New function.
15169 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15170 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15171 * doc/tm.texi: Regenerate.
15172 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15173 rather than TRULY_NOOP_TRUNCATION in comments.
15174 (simplify_comparison): Likewise.
15175 (record_truncated_value): Likewise.
15176 * expmed.c (extract_bit_field_1): Likewise.
15177 (extract_split_bit_field): Likewise.
15178 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15179 instead of TRULY_NOOP_TRUNCATION.
15180 * function.c (assign_parm_setup_block): Likewise.
15181 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15182 * rtlhooks.c: Include target.h.
15183 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15184 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15185 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15186 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15187 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15188 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15189 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15190 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15191 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15192 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15193 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15194 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15195 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15196 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15197 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15198 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15199 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15200 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15201 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15202 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15203 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15204 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15205 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15206 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15207 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15208 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15209 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15210 rather than TRULY_NOOP_TRUNCATION in comments.
15211 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15212 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15213 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15214 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15215 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15216 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15217 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15218 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15219 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15220 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15221 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15222 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15223 rather than TRULY_NOOP_TRUNCATION in comments.
15224 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15225 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15226 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15227 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15228 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15229 TRULY_NOOP_TRUNCATION condition.
15230 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15231 (TRULY_NOOP_TRUNCATION): Delete.
15232 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15233 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15234 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15235 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15236 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15237 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15238 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15239 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15240 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15241 rather than TRULY_NOOP_TRUNCATION in comments.
15242 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15243 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15244 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15245 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15246 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15247 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15249 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15252 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15253 (*cmp_ior): Likewise.
15254 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15255 (*ior_scc_scc_cmp): Likewise.
15256 (*and_scc_scc): Likewise.
15257 (*and_scc_scc_cmp): Likewise.
15259 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15260 Alan Hayard <alan.hayward@arm.com>
15261 David Sherwood <david.sherwood@arm.com>
15263 * target.def (can_change_mode_class): New hook.
15264 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15265 (hard_regno_nregs): Likewise.
15266 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15267 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15268 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15269 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15270 (LOAD_EXTEND_OP): Update accordingly.
15271 * doc/tm.texi: Regenerate.
15272 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15273 CANNOT_CHANGE_MODE_CLASS.
15274 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15275 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15276 * combine.c (simplify_set): Update accordingly.
15277 * emit-rtl.c (validate_subreg): Likewise.
15278 * recog.c (general_operand): Likewise.
15279 * regcprop.c (mode_change_ok): Likewise.
15280 * reload1.c (choose_reload_regs): Likewise.
15281 (inherit_piecemeal_p): Likewise.
15282 * rtlanal.c (simplify_subreg_regno): Likewise.
15283 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15284 instead of CANNOT_CHANGE_MODE_CLASS.
15285 (reload_cse_simplify_operands): Likewise.
15286 * reload.c (push_reload): Use targetm.can_change_mode_class
15287 instead of CANNOT_CHANGE_MODE_CLASS.
15288 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15289 REG_CANNOT_CHANGE_MODE_P.
15290 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15291 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15292 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15293 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15294 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15295 (arm_can_change_mode_class): New function.
15296 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15297 than CANNOT_CHANGE_MODE_CLASS in comments.
15298 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15299 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15300 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15301 (ix86_can_change_mode_class): ...this new function, inverting the
15302 sense of the return value.
15303 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15304 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15305 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15306 (ia64_can_change_mode_class): New function.
15307 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15308 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15309 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15310 (m32c_can_change_mode_class): ...this new function, inverting the
15311 sense of the return value.
15312 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15313 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15314 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15315 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15316 (mips_can_change_mode_class): ...this new function, inverting the
15317 sense of the return value.
15318 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15319 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15320 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15321 (msp430_can_change_mode_class): New function.
15322 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15323 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15324 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15325 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15326 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15327 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15328 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15329 (pa_cannot_change_mode_class): Replace with...
15330 (pa_can_change_mode_class): ...this new function, inverting the
15331 sense of the return value.
15332 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15333 than CANNOT_CHANGE_MODE_CLASS in comments.
15334 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15335 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15336 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15337 (pdp11_cannot_change_mode_class): Replace with...
15338 (pdp11_can_change_mode_class): ...this new function, inverting the
15339 sense of the return value.
15340 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15341 * config/powerpcspe/powerpcspe-protos.h
15342 (rs6000_cannot_change_mode_class_ptr): Delete.
15343 * config/powerpcspe/powerpcspe.c
15344 (rs6000_cannot_change_mode_class_ptr): Delete.
15345 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15346 (rs6000_option_override_internal): Assign to
15347 targetm.can_change_mode_class instead of
15348 rs6000_cannot_change_mode_class_ptr.
15349 (rs6000_cannot_change_mode_class): Replace with...
15350 (rs6000_can_change_mode_class): ...this new function, inverting the
15351 sense of the return value.
15352 (rs6000_debug_cannot_change_mode_class): Replace with...
15353 (rs6000_debug_can_change_mode_class): ...this new function.
15354 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15355 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15356 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15357 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15358 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15360 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15361 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15362 (rs6000_option_override_internal): Assign to
15363 targetm.can_change_mode_class instead of
15364 rs6000_cannot_change_mode_class_ptr.
15365 (rs6000_cannot_change_mode_class): Replace with...
15366 (rs6000_can_change_mode_class): ...this new function, inverting the
15367 sense of the return value.
15368 (rs6000_debug_cannot_change_mode_class): Replace with...
15369 (rs6000_debug_can_change_mode_class): ...this new function.
15370 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15371 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15372 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15373 (s390_can_change_mode_class): ...this new function, inverting the
15374 sense of the return value.
15375 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15376 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15377 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15378 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15379 (sh_cannot_change_mode_class): Replace with...
15380 (sh_can_change_mode_class): ...this new function, inverting the
15381 sense of the return value.
15382 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15383 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15384 (sparc_can_change_mode_class): New function.
15385 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15386 * config/spu/spu.c (spu_can_change_mode_class): New function.
15387 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15388 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15389 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15390 (visium_can_change_mode_class): New function.
15391 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15393 2017-09-15 Richard Biener <rguenther@suse.de>
15395 PR tree-optimization/82217
15396 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15397 but not undefined case.
15399 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15402 * postreload.c (reload_cse_simplify_operands): Skip
15403 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15405 2017-09-15 Richard Biener <rguenther@suse.de>
15407 PR tree-optimization/68823
15408 * graphite-scop-detection.c (build_alias_set): If we have a
15409 possible dependence check whether we can handle them by just
15410 looking at the DRs DR_ACCESS_FNs.
15411 (build_scops): If build_alias_set fails, fail the SCOP.
15413 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15415 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15416 to support float128 built-in functions that require the ISA 3.0
15418 (BU_FLOAT128_3_HW): Likewise.
15419 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15420 built-in functions.
15421 (FMAF128): Likewise.
15423 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15424 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15425 floating point instructions.
15426 (rs6000_invalid_builtin): Likewise.
15427 (rs6000_builtin_mask_names): Likewise.
15428 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15429 (RS6000_BTM_FLOAT128_HW): Likewise.
15430 (RS6000_BTM_COMMON): Likewise.
15431 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15433 * doc/extend.texi (RS/6000 built-in functions): Document the
15434 IEEE 128-bit floating point square root and fused multiply-add
15435 built-in functions.
15437 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15439 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15440 reg (r2) isn't in the set of registers defined in the prologue.
15442 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15443 Alan Hayward <alan.hayward@arm.com>
15444 David Sherwood <david.sherwood@arm.com>
15446 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15447 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15448 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15449 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15450 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15452 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15453 max_vectorization_factor.
15454 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15456 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15457 Alan Hayward <alan.hayward@arm.com>
15458 David Sherwood <david.sherwood@arm.com>
15460 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15461 (vect_worthwhile_without_simd_p): Declare.
15462 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15463 (vectorizable_reduction): Use it.
15464 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15465 (vectorizable_operation): Likewise.
15467 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15468 Alan Hayward <alan.hayward@arm.com>
15469 David Sherwood <david.sherwood@arm.com>
15471 * tree-vectorizer.h (vect_get_num_copies): New function.
15472 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15473 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15474 (vectorizable_induction): Likewise.
15475 (vectorizable_live_operation): Likewise.
15476 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15477 (vectorizable_bswap): Likewise.
15478 (vectorizable_call): Likewise.
15479 (vectorizable_conversion): Likewise.
15480 (vectorizable_assignment): Likewise.
15481 (vectorizable_shift): Likewise.
15482 (vectorizable_operation): Likewise.
15483 (vectorizable_store): Likewise.
15484 (vectorizable_load): Likewise.
15485 (vectorizable_condition): Likewise.
15486 (vectorizable_comparison): Likewise.
15487 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15489 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15490 Alan Hayward <alan.hayward@arm.com>
15491 David Sherwood <david.sherwood@arm.com>
15493 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15494 of vect_init_vector.
15496 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15497 Alan Hayward <alan.hayward@arm.com>
15498 David Sherwood <david.sherwood@arm.com>
15500 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15501 an inline wrapper that provides a location.
15502 (gimple_build_vector): Likewise.
15503 * gimple-fold.c (gimple_build_vector_from_val): New function.
15504 (gimple_build_vector): Likewise.
15505 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15506 functions to build the initial value. Always return a gimple value.
15507 (get_initial_defs_for_reduction): Likewise. Only compute
15509 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15510 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15511 (vectorizable_induction): Use gimple_build_vector rather than
15514 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15515 Alan Hayward <alan.hayward@arm.com>
15516 David Sherwood <david.sherwood@arm.com>
15518 * target.h (vec_perm_indices): New typedef.
15519 (auto_vec_perm_indices): Likewise.
15520 * optabs-query.h: Include target.h
15521 (can_vec_perm_p): Take a vec_perm_indices *.
15522 * optabs-query.c (can_vec_perm_p): Likewise.
15523 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15524 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15525 * tree-vect-generic.c (lower_vec_perm): Likewise.
15526 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15527 (vect_grouped_load_supported): Likewise.
15528 (vect_shift_permute_load_chain): Likewise.
15529 (vect_permute_store_chain): Use auto_vec_perm_indices.
15530 (vect_permute_load_chain): Likewise.
15531 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15532 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15533 Update uses of can_vec_perm_p.
15534 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15535 mode with a number of elements. Take a vec_perm_indices *.
15536 (vect_create_epilog_for_reduction): Update accordingly.
15537 Use auto_vec_perm_indices.
15538 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15539 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15540 (vect_transform_slp_perm_load): Likewise.
15541 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15542 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15543 (vect_gen_perm_mask_checked): Likewise.
15544 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15545 (vect_gen_perm_mask_checked): Likewise.
15546 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15547 (vectorizable_store): Likewise.
15548 (vectorizable_load): Likewise.
15549 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15550 (vectorizable_bswap): Likewise.
15552 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15553 Alan Hayward <alan.hayward@arm.com>
15554 David Sherwood <david.sherwood@arm.com>
15556 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15557 * tree.c (build_vector): Likewise.
15558 (build_vector_from_ctor): Update accordingly.
15559 (build_vector_from_val): Likewise.
15560 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15561 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15562 * tree-vect-generic.c (add_rshift): Likewise.
15563 (expand_vector_divmod): Likewise.
15564 (optimize_vector_constructor): Likewise.
15565 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15566 (vect_transform_slp_perm_load): Likewise.
15567 (vect_schedule_slp_instance): Likewise.
15568 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15569 (vectorizable_call): Likewise.
15570 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15571 * expmed.c (make_tree): Likewise.
15572 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15573 a vector passed to build_vector.
15574 (fold_convert_const): Likewise.
15575 (exact_inverse): Likewise.
15576 (fold_ternary_loc): Likewise.
15577 (fold_relational_const): Likewise.
15578 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15579 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15580 (const_unop): Likewise. Store the reduction accumulator in a
15581 variable rather than an array.
15582 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15583 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15584 the new vector, rather than constructing it after the input arrays.
15585 (native_interpret_vector): Use auto_vec<tree> when building
15586 a vector passed to build_vector. Add elements in order.
15587 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15588 auto_vec<tree> when building a vector passed to build_vector.
15589 (vect_create_epilog_for_reduction): Likewise.
15590 (vectorizable_induction): Likewise.
15591 (get_initial_def_for_reduction): Likewise. Fix indentation of
15593 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15595 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15596 passed to build_vector.
15598 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15599 Alan Hayward <alan.hayward@arm.com>
15600 David Sherwood <david.sherwood@arm.com>
15602 * tree-core.h (tree_base::u): Add an "nelts" field.
15603 (tree_vector): Use VECTOR_CST_NELTS as the length.
15604 * tree.c (tree_size): Likewise.
15605 (make_vector): Initialize VECTOR_CST_NELTS.
15606 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15607 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15608 TYPE_VECTOR_SUBPARTS.
15609 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15610 as the number of units, setting it from VECTOR_CST_NELTS.
15611 (const_vector_from_tree): Likewise.
15612 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15613 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15614 (fold_negate_expr_1): Likewise.
15615 (fold_convert_const): Likewise.
15616 (const_binop): Likewise. Differentiate the number of output and
15618 (const_unop): Likewise.
15619 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15620 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15621 in cases that did the opposite.
15623 2017-09-14 Richard Biener <rguenther@suse.de>
15625 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15628 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15630 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15632 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15635 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15636 if and where to split a bb, except for splitting before debug insn
15637 sequences followed by non-label real insn. Delete debug insns
15638 in between basic blocks.
15640 * combine.c (make_compound_operation_int): Formatting fixes.
15642 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15643 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15644 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15645 * config/sol2.h (LINK_EH_SPEC): Likewise.
15646 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15647 * config/s390/linux.h (LINK_SPEC): Likewise.
15648 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15649 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15650 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15651 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15652 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15653 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15654 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15655 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15657 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15659 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15661 (ENDFILE_LINUX_SPEC): Likewise.
15662 (LINK_EH_SPEC): Likewise.
15663 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15664 (LINK_OS_LINUX_SPEC32): Likewise.
15665 (LINK_OS_LINUX_SPEC64): Likewise.
15666 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15667 (LINK_OS_LINUX_SPEC): Likewise.
15669 2017-09-13 Martin Liska <mliska@suse.cz>
15671 PR middle-end/82154
15672 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15673 CASE_HIGH is NULL_TREE.
15675 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15676 Alan Hayward <alan.hayward@arm.com>
15677 David Sherwood <david.sherwood@arm.com>
15679 * target.def (secondary_memory_needed): New hook.
15680 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15681 instead of SECONDARY_MEMORY_NEEDED.
15682 (secondary_memory_needed_mode): Likewise.
15683 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15684 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15685 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15686 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15687 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15688 * doc/tm.texi: Regenerate.
15689 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15690 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15691 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15692 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15693 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15694 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15695 mode argument first and change the reg_class arguments to reg_class_t.
15696 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15697 Make static. Update the call to inline_secondary_memory_needed.
15698 (ix86_register_move_cost): Update the call to
15699 inline_secondary_memory_needed.
15700 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15701 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15703 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15704 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15706 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15707 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15708 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15709 and match hook interface. Add comment from mips.h.
15710 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15711 * config/mmix/mmix.md (truncdfsf2): Refer to
15712 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15714 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15715 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15716 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15717 (pa_secondary_memory_needed): New function.
15718 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15719 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15720 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15721 (pdp11_secondary_memory_needed): Make static and match hook interface.
15722 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15723 * config/powerpcspe/powerpcspe-protos.h
15724 (rs6000_secondary_memory_needed_ptr): Delete.
15725 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15727 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15728 (rs6000_option_override_internal): Assign to
15729 targetm.secondary_memory_needed rather than
15730 rs6000_secondary_memory_needed_ptr.
15731 (rs6000_secondary_memory_needed): Match hook interface.
15732 (rs6000_debug_secondary_memory_needed): Likewise.
15733 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15734 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15735 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15736 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15737 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15738 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15740 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15741 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15742 (rs6000_option_override_internal): Assign to
15743 targetm.secondary_memory_needed rather than
15744 rs6000_secondary_memory_needed_ptr.
15745 (rs6000_secondary_memory_needed): Match hook interface.
15746 (rs6000_debug_secondary_memory_needed): Likewise.
15747 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15748 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15749 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15750 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15751 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15752 (sparc_secondary_memory_needed): New function.
15753 * lra-constraints.c (check_and_process_move): Refer to
15754 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15756 (curr_insn_transform): Likewise.
15757 (process_alt_operands): Use targetm.secondary_memory_needed
15758 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15759 (check_secondary_memory_needed_p): Likewise.
15760 (choose_split_class): Likewise.
15761 * reload.c: Unconditionally include code that was previously
15762 conditional on SECONDARY_MEMORY_NEEDED.
15763 (push_secondary_reload): Use targetm.secondary_memory_needed
15764 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15765 (push_reload): Likewise.
15766 * reload1.c: Unconditionally include code that was previously
15767 conditional on SECONDARY_MEMORY_NEEDED.
15768 (choose_reload_regs): Use targetm.secondary_memory_needed
15769 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15770 (gen_reload): Likewise.
15771 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15773 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15774 Alan Hayward <alan.hayward@arm.com>
15775 David Sherwood <david.sherwood@arm.com>
15777 * target.def (secondary_memory_needed_mode): New hook:
15778 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15779 * targhooks.h (default_secondary_memory_needed_mode): New function.
15780 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15781 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15782 * doc/tm.texi: Regenerate.
15783 * lra-constraints.c (check_and_process_move): Use
15784 targetm.secondary_memory_needed_mode instead of
15785 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15786 (curr_insn_transform): Likewise.
15787 * reload.c (get_secondary_mem): Likewise.
15788 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15789 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
15791 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15792 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15793 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
15794 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15795 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
15797 * config/powerpcspe/powerpcspe-protos.h
15798 (rs6000_secondary_memory_needed_mode): Delete.
15799 * config/powerpcspe/powerpcspe.c
15800 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15801 (rs6000_secondary_memory_needed_mode): Make static.
15802 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15803 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
15805 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15807 (rs6000_secondary_memory_needed_mode): Make static.
15808 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15809 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
15810 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15811 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15812 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15814 (sparc_secondary_memory_needed_mode): New function.
15815 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
15817 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15819 * config/aarch64/constraints.md (Umq): New constraint.
15820 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15822 (mov<mode>): Update condition.
15824 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15826 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15827 when bitposition is the same.
15829 2017-09-13 Richard Biener <rguenther@suse.de>
15831 * dwarf2out.c (output_die_symbol): Remove.
15832 (output_die): Do not output a DIEs symbol.
15834 2017-09-13 Richard Biener <rguenther@suse.de>
15836 PR middle-end/82128
15837 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15838 default-def to avoid breaking iterator update with the weird
15839 interaction with cgraph_update_edges_for_call_stmt_node.
15841 2017-09-13 Richard Biener <rguenther@suse.de>
15843 * tree-cfg.c (verify_gimple_assign_binary): Add verification
15844 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
15845 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
15846 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
15848 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15850 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15851 Disable pc relative literal load irrespective of
15852 TARGET_FIX_ERR_A53_84341 for default.
15854 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
15856 * config/sparc/sparc.c (output_return): Output the source location of
15857 the insn in the delay slot, if any.
15858 (output_sibcall): Likewise.
15860 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
15863 * common.opt (-static-pie): New alias.
15864 (shared): Negate static-pie.
15865 (-no-pie): Update help text.
15867 (static-pie): New option.
15868 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
15869 -static-pie support.
15870 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
15871 (LINK_EH_SPEC): Likewise.
15872 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15873 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15874 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15875 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15876 (init_gcc_specs): Likewise.
15877 (init_spec): Likewise.
15878 (display_help): Update help message for -pie.
15879 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15882 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15884 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15885 (movdi_aarch64): Likewise.
15886 (movti_aarch64): Likewise.
15888 2017-09-12 Simon Wright <simon@pushface.org>
15891 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15892 calculation of the minor version, always output as 0.
15894 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15897 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15898 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15899 on it early, rather than manual conversion late. For
15900 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15901 instead of performing manual conversion.
15903 2017-09-12 Carl Love <cel@us.ibm.com>
15905 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15906 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15908 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15909 VMULOSW): Add definitions.
15910 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15911 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15912 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15913 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15914 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15916 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15918 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15920 (movti_aarch64): Likewise.
15921 (movdf_aarch64): Likewise.
15922 (movtf_aarch64): Likewise.
15923 (load_pairdi): Likewise.
15924 (store_pairdi): Likewise.
15925 (load_pairdf): Likewise.
15926 (store_pairdf): Likewise.
15927 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15928 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15929 (ldr_got_small_<mode>): Likewise.
15930 (ldr_got_small_28k_<mode>): Likewise.
15931 (ldr_got_tiny): Likewise.
15932 * config/aarch64/iterators.md (ldst_sz): New.
15933 (ldpstp_sz): Likewise.
15934 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15936 (thunderx_load): Split load_8 to load_16.
15937 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15939 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15940 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15941 (xgene1_store_pair): Split store_8 to store_16.
15942 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15943 (falkor_st_0_st_sd): Split store_8 to store_16.
15945 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15947 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15948 and store1/2/3/4 to store_4/8/12/16.
15949 * config/aarch64/aarch64.md: Update for rename.
15950 * config/arm/arm.md: Likewise.: Likewise.
15951 * config/arm/arm.c: Likewise.
15952 * config/arm/thumb1.md: Likewise.
15953 * config/arm/thumb2.md: Likewise.
15954 * config/arm/vfp.md: Likewise.
15955 * config/arm/arm-generic.md: Likewise.
15956 * config/arm/arm1020e.md: Likewise.
15957 * config/arm/arm1026ejs.md: Likewise.
15958 * config/arm/arm1136jfs.md: Likewise.
15959 * config/arm/arm926ejs.md: Likewise.
15960 * config/arm/cortex-a15.md: Likewise.
15961 * config/arm/cortex-a17.md: Likewise.
15962 * config/arm/cortex-a5.md: Likewise.
15963 * config/arm/cortex-a53.md: Likewise.
15964 * config/arm/cortex-a57.md: Likewise.
15965 * config/arm/cortex-a7.md: Likewise.
15966 * config/arm/cortex-a8.md: Likewise.
15967 * config/arm/cortex-a9.md: Likewise.
15968 * config/arm/cortex-m4.md: Likewise.
15969 * config/arm/cortex-m7.md: Likewise.
15970 * config/arm/cortex-r4.md: Likewise.
15971 * config/arm/exynos-m1.md: Likewise.
15972 * config/arm/fa526.md: Likewise.
15973 * config/arm/fa606te.md: Likewise.
15974 * config/arm/fa626te.md: Likewise.
15975 * config/arm/fa726te.md: Likewise.
15976 * config/arm/fmp626.md: Likewise.
15977 * config/arm/iwmmxt.md: Likewise.
15978 * config/arm/ldmstm.md: Likewise.
15979 * config/arm/marvell-pj4.md: Likewise.
15980 * config/arm/xgene1.md: Likewise.
15981 * config/aarch64/thunderx.md: Likewise.
15982 * config/aarch64/thunderx2t99.md: Likewise.
15983 * config/aarch64/falkor.md: Likewise.
15985 2017-09-12 Martin Liska <mliska@suse.cz>
15987 * attribs.c (private_lookup_attribute): New function.
15988 * attribs.h (private_lookup_attribute): Declared here.
15989 (lookup_attribute): Called from this place.
15991 2017-09-12 Richard Biener <rguenther@suse.de>
15993 PR tree-optimization/82157
15994 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15995 stmts with side-effects.
15997 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15998 Alan Hayward <alan.hayward@arm.com>
15999 David Sherwood <david.sherwood@arm.com>
16001 * target.def (hard_regno_nregs): New hook.
16002 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16003 * targhooks.h (default_hard_regno_nregs): Declare.
16004 * targhooks.c (default_hard_regno_nregs): New function.
16005 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16006 (TARGET_HARD_REGNO_NREGS): ...this hook.
16007 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16008 (CLASS_MAX_NREGS): Likewise.
16009 * doc/tm.texi: Regenerate.
16010 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16011 instead of HARD_REGNO_NREGS.
16012 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16013 HARD_REGNO_NREGS in the comment.
16014 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16015 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16016 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16017 Return an unsigned int.
16018 (TARGET_HARD_REGNO_NREGS): Redefine.
16019 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16020 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16021 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16022 (arc_hard_regno_nregs): New function.
16023 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16024 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16025 (arm_hard_regno_nregs): New function.
16026 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16027 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16028 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16029 (TARGET_HARD_REGNO_NREGS): Redefine.
16030 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16031 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16032 (HARD_REGNO_NREGS): Delete.
16033 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16034 (cr16_hard_regno_nregs): New function.
16035 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16036 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16037 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16038 (cris_hard_regno_nregs): New function.
16039 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16040 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16041 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16042 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16043 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16044 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16045 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16046 (frv_hard_regno_nregs): Make static. Take and return an
16048 (frv_class_max_nregs): Remove outdated copy of documentation.
16049 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16050 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16051 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16052 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16053 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16054 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16055 (TARGET_HARD_REGNO_NREGS): Redefine.
16056 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16057 (CLASS_MAX_NREGS): Update comment.
16058 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16059 (ia64_hard_regno_nregs): New function.
16060 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16061 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16062 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16063 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16064 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16066 (m32c_hard_regno_nregs): Likewise. Make static.
16067 (TARGET_HARD_REGNO_NREGS): Redefine.
16068 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16069 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16070 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16071 (m68k_hard_regno_nregs): New function.
16072 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16073 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16074 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16075 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16076 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16077 Take and return an unsigned int.
16078 (TARGET_HARD_REGNO_NREGS): Redefine.
16079 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16080 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16081 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16082 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16083 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16084 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16085 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16086 (msp430_hard_regno_nregs): Make static. Take and return an
16088 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16089 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16090 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16091 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16092 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16093 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16094 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16095 (TARGET_HARD_REGNO_NREGS): Redefine.
16096 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16097 (PA_HARD_REGNO_NREGS): ...this.
16098 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16099 (PA_HARD_REGNO_NREGS): ...this.
16100 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16101 (pa_hard_regno_nregs): New function.
16102 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16103 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16104 (pdp11_hard_regno_nregs): New function.
16105 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16106 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16107 (rs6000_hard_regno_nregs_hook): New function.
16108 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16109 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16110 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16111 Take and return an unsigned int. Move earlier in file.
16112 (TARGET_HARD_REGNO_NREGS): Redefine.
16113 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16114 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16115 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16116 (rl78_hard_regno_nregs): Make static. Take and return an
16118 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16119 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16120 (rs6000_hard_regno_nregs_hook): New function.
16121 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16122 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16123 (TARGET_HARD_REGNO_NREGS): Redefine.
16124 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16125 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16126 instead of HARD_REGNO_NREGS.
16127 (s390_hard_regno_nregs): New function.
16128 (s390_hard_regno_mode_ok): Add comment from s390.h.
16129 (TARGET_HARD_REGNO_NREGS): Redefine.
16130 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16131 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16132 (sh_hard_regno_nregs): New function.
16133 (sh_pass_in_reg_p): Use it.
16134 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16135 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16136 (sparc_hard_regno_nregs): New function.
16137 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16138 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16139 (spu_function_arg_advance): Use it, supplying a valid register number.
16140 (TARGET_HARD_REGNO_NREGS): Redefine.
16141 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16142 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16143 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16144 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16145 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16146 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16147 (CLASS_MAX_NREGS): Remove copy of old documentation.
16148 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16149 (visium_hard_regno_nregs): New function.
16150 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16151 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16152 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16153 xtensa_hard_regno_nregs): New function.
16154 * system.h (HARD_REGNO_NREGS): Poison.
16156 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16158 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16159 hard_regno_nregs instead of HARD_REGNO_NREGS.
16160 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16161 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16162 (c6x_expand_epilogue): Likewise.
16163 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16164 (frv_read_iacc_argument): Likewise.
16165 * config/sh/sh.c: Include regs.h.
16166 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16167 (regs_used): Likewise.
16168 (output_stack_adjust): Likewise.
16169 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16170 * expmed.c: Include regs.h.
16171 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16172 * ree.c: Include regs.h.
16173 (combine_reaching_defs): Use hard_regno_nregs instead of
16175 (add_removable_extension): Likewise.
16177 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16179 * regs.h (hard_regno_nregs): Turn into a function.
16180 (end_hard_regno): Update accordingly.
16181 * caller-save.c (setup_save_areas): Likewise.
16182 (save_call_clobbered_regs): Likewise.
16183 (replace_reg_with_saved_mem): Likewise.
16184 (insert_restore): Likewise.
16185 (insert_save): Likewise.
16186 * combine.c (can_change_dest_mode): Likewise.
16187 (move_deaths): Likewise.
16188 (distribute_notes): Likewise.
16189 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16190 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16191 (rs6000_split_multireg_move): Likewise.
16192 (rs6000_register_move_cost): Likewise.
16193 (rs6000_memory_move_cost): Likewise.
16194 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16195 (rs6000_split_multireg_move): Likewise.
16196 (rs6000_register_move_cost): Likewise.
16197 (rs6000_memory_move_cost): Likewise.
16198 * cselib.c (cselib_reset_table): Likewise.
16199 (cselib_lookup_1): Likewise.
16200 * emit-rtl.c (set_mode_and_regno): Likewise.
16201 * function.c (aggregate_value_p): Likewise.
16202 * ira-color.c (setup_profitable_hard_regs): Likewise.
16203 (check_hard_reg_p): Likewise.
16204 (calculate_saved_nregs): Likewise.
16205 (assign_hard_reg): Likewise.
16206 (improve_allocation): Likewise.
16207 (calculate_spill_cost): Likewise.
16208 * ira-emit.c (modify_move_list): Likewise.
16209 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16210 (ira_hard_reg_in_set_p): Likewise.
16211 * ira.c (setup_reg_mode_hard_regset): Likewise.
16212 (clarify_prohibited_class_mode_regs): Likewise.
16213 (check_allocation): Likewise.
16214 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16215 (lra_setup_reg_renumber): Likewise.
16216 (setup_try_hard_regno_pseudos): Likewise.
16217 (spill_for): Likewise.
16218 (assign_hard_regno): Likewise.
16219 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16220 * lra-constraints.c (in_class_p): Likewise.
16221 (lra_constraint_offset): Likewise.
16222 (simplify_operand_subreg): Likewise.
16223 (lra_constraints): Likewise.
16224 (split_reg): Likewise.
16225 (split_if_necessary): Likewise.
16226 (invariant_p): Likewise.
16227 (inherit_in_ebb): Likewise.
16228 * lra-lives.c (process_bb_lives): Likewise.
16229 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16230 (get_hard_regs): Likewise.
16231 (do_remat): Likewise.
16232 * lra-spills.c (assign_spill_hard_regs): Likewise.
16233 * mode-switching.c (create_pre_exit): Likewise.
16234 * postreload.c (reload_combine_recognize_pattern): Likewise.
16235 * recog.c (peep2_find_free_register): Likewise.
16236 * regcprop.c (kill_value_regno): Likewise.
16237 (set_value_regno): Likewise.
16238 (copy_value): Likewise.
16239 (maybe_mode_change): Likewise.
16240 (find_oldest_value_reg): Likewise.
16241 (copyprop_hardreg_forward_1): Likewise.
16242 * regrename.c (check_new_reg_p): Likewise.
16243 (regrename_do_replace): Likewise.
16244 * reload.c (push_reload): Likewise.
16245 (combine_reloads): Likewise.
16246 (find_dummy_reload): Likewise.
16247 (operands_match_p): Likewise.
16248 (find_reloads): Likewise.
16249 (find_equiv_reg): Likewise.
16250 (reload_adjust_reg_for_mode): Likewise.
16251 * reload1.c (count_pseudo): Likewise.
16252 (count_spilled_pseudo): Likewise.
16253 (find_reg): Likewise.
16254 (clear_reload_reg_in_use): Likewise.
16255 (free_for_value_p): Likewise.
16256 (allocate_reload_reg): Likewise.
16257 (choose_reload_regs): Likewise.
16258 (reload_adjust_reg_for_temp): Likewise.
16259 (emit_reload_insns): Likewise.
16260 (delete_output_reload): Likewise.
16261 * rtlanal.c (subreg_get_info): Likewise.
16262 * sched-deps.c (sched_analyze_reg): Likewise.
16263 * sel-sched.c (init_regs_for_mode): Likewise.
16264 (mark_unavailable_hard_regs): Likewise.
16265 (choose_best_reg_1): Likewise.
16266 (verify_target_availability): Likewise.
16267 * valtrack.c (dead_debug_insert_temp): Likewise.
16268 * var-tracking.c (track_loc_p): Likewise.
16269 (emit_note_insn_var_location): Likewise.
16270 * varasm.c (make_decl_rtl): Likewise.
16271 * reginfo.c (choose_hard_reg_mode): Likewise.
16272 (init_reg_modes_target): Refer directly to
16273 this_target_regs->x_hard_regno_nregs.
16275 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16277 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16278 instead of hard_regno_nregs.
16280 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16282 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16283 end_hard_regno instead of hard_regno_nregs.
16284 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16285 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16286 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16287 * ira-color.c (improve_allocation): Likewise.
16288 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16289 * lra-lives.c (mark_regno_live): Likewise.
16290 (mark_regno_dead): Likewise.
16291 * lra-remat.c (operand_to_remat): Likewise.
16292 * lra.c (collect_non_operand_hard_regs): Likewise.
16293 * postreload.c (reload_combine_note_store): Likewise.
16294 (move2add_valid_value_p): Likewise.
16295 * reload.c (regno_clobbered_p): Likewise.
16297 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16299 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16301 * config/v850/v850.c (v850_reorg): Likewise.
16302 * reload.c (refers_to_regno_for_reload_p): Likewise.
16303 (find_equiv_reg): Likewise.
16304 * reload1.c (reload_reg_reaches_end_p): Likewise.
16306 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16308 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16310 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16311 * config/arm/arm.c (output_move_neon): Likewise.
16312 (arm_attr_length_move_neon): Likewise.
16313 (neon_split_vcombine): Likewise.
16314 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16315 (c6x_mark_reg_written): Likewise.
16316 (c6x_dwarf_register_span): Likewise.
16317 * config/i386/i386.c (ix86_save_reg): Likewise.
16318 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16319 (rws_access_reg): Likewise.
16320 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16321 * mode-switching.c (create_pre_exit): Likewise.
16322 * ree.c (combine_reaching_defs): Likewise.
16323 (add_removable_extension): Likewise.
16324 * regcprop.c (find_oldest_value_reg): Likewise.
16325 (copyprop_hardreg_forward_1): Likewise.
16326 * reload.c (reload_inner_reg_of_subreg): Likewise.
16327 (push_reload): Likewise.
16328 (combine_reloads): Likewise.
16329 (find_dummy_reload): Likewise.
16330 (reload_adjust_reg_for_mode): Likewise.
16331 * reload1.c (find_reload_regs): Likewise.
16332 (forget_old_reloads_1): Likewise.
16333 (reload_reg_free_for_value_p): Likewise.
16334 (reload_adjust_reg_for_temp): Likewise.
16335 (emit_reload_insns): Likewise.
16336 (delete_output_reload): Likewise.
16337 * sel-sched.c (choose_best_reg_1): Likewise.
16338 (choose_best_pseudo_reg): Likewise.
16340 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16341 Alan Hayward <alan.hayward@arm.com>
16342 David Sherwood <david.sherwood@arm.com>
16344 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16345 * target.def (slow_unaligned_access): New hook.
16346 * targhooks.h (default_slow_unaligned_access): Declare.
16347 * targhooks.c (default_slow_unaligned_access): New function.
16348 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16349 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16350 * doc/tm.texi: Regenerate.
16351 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16352 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16353 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16355 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16356 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16358 (rs6000_slow_unaligned_access): New function.
16359 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16360 (expand_block_compare): Likewise.
16361 (expand_strn_compare): Likewise.
16362 (rs6000_rtx_costs): Likewise.
16363 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16364 (riscv_slow_unaligned_access): Likewise.
16365 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16366 (riscv_slow_unaligned_access_p): ...this and make static.
16367 (riscv_option_override): Update accordingly.
16368 (riscv_slow_unaligned_access): New function.
16369 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16370 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16371 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16372 (rs6000_slow_unaligned_access): New function.
16373 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16374 (rs6000_rtx_costs): Likewise.
16375 * config/rs6000/rs6000-string.c (expand_block_compare)
16376 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16377 of SLOW_UNALIGNED_ACCESS.
16378 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16379 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16380 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16381 of SLOW_UNALIGNED_ACCESS.
16382 * expmed.c (simple_mem_bitfield_p): Likewise.
16383 * expr.c (alignment_for_piecewise_move): Likewise.
16384 (emit_group_load_1): Likewise.
16385 (emit_group_store): Likewise.
16386 (copy_blkmode_from_reg): Likewise.
16387 (emit_push_insn): Likewise.
16388 (expand_assignment): Likewise.
16389 (store_field): Likewise.
16390 (expand_expr_real_1): Likewise.
16391 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16392 * lra-constraints.c (simplify_operand_subreg): Likewise.
16393 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16394 * gimple-ssa-store-merging.c: Likewise in block comment at start
16396 * tree-ssa-strlen.c: Include target.h.
16397 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16398 of SLOW_UNALIGNED_ACCESS.
16399 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16401 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16403 PR rtl-optimization/82185
16404 * expmed.c (emit_store_flag_int): Only test tem if it has been
16407 2017-09-12 Richard Biener <rguenther@suse.de>
16409 PR middle-end/82149
16410 * match.pd ((FTYPE) N CMP CST): Fix typo.
16412 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16414 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16416 (mips_near_type_p): Add 'short_call' attribute as a synonym
16418 * doc/extend.texi (short_call): Document new function attribute.
16420 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16423 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16424 assertion check that in the condition.
16425 (get_atomic_generic_size): Likewise. Before testing if parameter
16426 has pointer type, if it has array type, call for C++
16427 default_conversion to perform array-to-pointer conversion.
16429 2017-09-12 Richard Biener <rguenther@suse.de>
16431 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16432 for operations we cannot scalarize.
16434 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16436 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16437 vectors heap vectors. Clean up comments.
16438 Make visited_bbs a reference.
16439 (profitable_jump_thread_path): Make GC
16440 vectors heap vectors. Clean up comments.
16442 (convert_and_register_jump_thread_path): Make GC vectors heap
16444 (check_subpath_and_update_thread_path): Same. Clean up comments.
16445 Make visited_bbs a reference.
16446 (handle_phi): Abstract common code to to
16447 register_jump_thread_path_if_profitable.
16448 Rename VAR_BB to DEF_BB.
16450 Make GC vectors heap vectors.
16451 Make visited_bbs a reference.
16452 (handle_assignment): Same.
16453 (register_jump_thread_path_if_profitable): New.
16454 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16456 Make GC vectors heap vectors. Clean up comments.
16457 Make visited_bbs a reference.
16458 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16459 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16462 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16465 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16466 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16468 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16470 Revert r251800 and r251799.
16472 2017-09-11 Martin Jambor <mjambor@suse.cz>
16475 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16476 arguments in advance.
16477 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16478 use it to find predecessor edges.
16479 (naive_outof_ssa): Collect vector of predecessors.
16481 2017-09-08 Jason Merrill <jason@redhat.com>
16483 PR c++/70029 - ICE with ref-qualifier and -flto
16484 * langhooks.h (struct lang_hooks_for_types): Add
16485 copy_lang_qualifiers.
16486 * attribs.c (build_type_attribute_qual_variant): Use it.
16487 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16489 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16490 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16492 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16495 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16496 (*mulsi3_sp64): New instruction.
16497 (mulsi3): New expander.
16499 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16501 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16503 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16505 * sancov.c: Include memmodel.h.
16507 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16510 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16513 2017-09-07 Carl Love <cel@us.ibm.com>
16515 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16516 the sldi instruction.
16518 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16520 * sancov.c: Include tm_p.h.
16522 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16525 * output.h (switch_to_other_text_partition): New declaration.
16526 * varasm.c (switch_to_other_text_partition): New function.
16527 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16528 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16529 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16530 to the other text partition before emitting LCL label and switch back
16531 after emitting the word after it.
16533 2017-09-07 Richard Biener <rguenther@suse.de>
16535 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16536 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16537 critical edges here, after loop init.
16538 (pass_data_pre): Remove PROP_no_crit_edges flags.
16539 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16540 for valueization of call args to avoid leaking VN_TOP.
16541 (visit_use): Assert we do not visit default defs.
16542 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16543 Use error_mark_node to more easily detect leaking VN_TOP.
16544 All default-defs are varying, not VN_TOP. Mark them visited.
16545 (run_scc_vn): Make code match comment.
16547 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16549 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16550 OPTION_MASK_FLOAT128_KEYWORD.
16551 (POWERPC_MASKS): Likewise.
16552 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16553 support for the -mfloat128-type option, and make -mfloat128
16554 default on PowerPC Linux systems. Define or undefine
16555 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16556 Define __float128 to be __ieee128 if IEEE 128-bit support is
16557 enabled, or undefine it.
16558 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16559 Delete defining __FLOAT128_TYPE__.
16560 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16561 -mfloat128-type option and make -mfloat128 default on PowerPC
16563 (TARGET_FLOAT128_TYPE): Likewise.
16564 (-mfloat128-type): Likewise.
16565 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16566 Delete the -mfloat128-type option and make -mfloat128 default on
16567 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16568 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16569 128-bit floating point is enabled. Change tests from using
16570 -mfloat128-type to -mfloat128.
16571 (rs6000_mangle_type): Use the correct mangling for the __float128
16572 type even if normal long double is restricted to 64-bits.
16573 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16575 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16576 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16577 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16579 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16580 documentation for -mfloat128.
16582 2017-09-06 Olivier Hainque <hainque@adacore.com>
16584 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16586 2017-09-06 Wish Wu <wishwu007@gmail.com>
16587 Jakub Jelinek <jakub@redhat.com>
16589 * asan.c (initialize_sanitizer_builtins): Add
16590 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16591 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16592 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16593 BT_FN_VOID_UINT64_PTR variables.
16594 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16595 (BT_FN_VOID_UINT16_UINT16): Likewise.
16596 (BT_FN_VOID_UINT32_UINT32): Likewise.
16597 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16598 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16599 (BT_FN_VOID_UINT64_PTR): Likewise.
16600 * common.opt (flag_sanitize_coverage): New variable.
16601 (fsanitize-coverage=trace-pc): Remove.
16602 (fsanitize-coverage=): Add.
16603 * flag-types.h (enum sanitize_coverage_code): New enum.
16604 * fold-const.c (fold_range_test): Disable non-short-circuit
16605 optimization if flag_sanitize_coverage.
16606 (fold_truth_andor): Likewise.
16607 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16608 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16609 (coverage_sanitizer_opts): New array.
16610 (get_closest_sanitizer_option): Add OPTS argument, handle also
16611 OPT_fsanitize_coverage_.
16612 (parse_sanitizer_options): Adjusted to also handle
16613 OPT_fsanitize_coverage_.
16614 (common_handle_option): Add OPT_fsanitize_coverage_.
16615 * sancov.c (instrument_comparison, instrument_switch): New function.
16616 (sancov_pass): Add trace-cmp support.
16617 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16618 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16619 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16620 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16621 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16622 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16623 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16624 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16625 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16626 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16628 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16630 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16631 error. Only quit immediately if parsing is complete.
16632 (BEGIN): Initialize fatal_err and parse_done.
16633 (begin fpu, end fpu): Check number of arguments.
16634 (begin arch, end arch): Likewise.
16635 (begin cpu, end cpu): Likewise.
16636 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16637 (optalias): Likewise.
16639 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16641 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16642 * config/arm/arm-isa.h: Delete. Move definitions to ...
16643 * arm-cpus.in: ... here. Use new feature and fgroup values.
16644 * config/arm/arm.c (arm_option_override): Use lower case for feature
16646 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16647 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16648 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16650 (print_isa_bits_for): New function.
16651 (gen_isa): New function.
16652 (gen_comm_data): Use print_isa_bits_for.
16653 (define feature): New keyword.
16654 (define fgroup): New keyword.
16655 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16656 (arm-isa.h): Add rule to generate file.
16657 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16658 case for feature bit names.
16660 2017-09-06 Richard Biener <rguenther@suse.de>
16662 * tree-ssa-pre.c (NECESSARY): Remove.
16663 (create_expression_by_pieces): Do not touch pass-local flags.
16664 (insert_into_preds_of_block): Likewise.
16665 (do_pre_regular_insertion): Likewise.
16666 (eliminate_insert): Likewise.
16667 (eliminate_dom_walker::before_dom_children): Likewise.
16668 (fini_eliminate): Do not look at inserted_exprs.
16669 (mark_operand_necessary): Remove.
16670 (remove_dead_inserted_code): Replace with simple work-list
16671 algorithm based on inserted_exprs and SSA uses.
16672 (pass_pre::execute): Re-order fini_eliminate and
16673 remove_dead_inserted_code.
16675 2017-09-06 Olivier Hainque <hainque@adacore.com>
16677 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16678 for VxWorks 7. Adjust surrounding comments.
16680 2017-09-06 Richard Biener <rguenther@suse.de>
16682 * gimple-ssa-strength-reduction.c
16683 (find_candidates_dom_walker::before_dom_children): Also allow
16686 2017-09-06 Richard Biener <rguenther@suse.de>
16688 PR tree-optimization/82108
16689 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16690 for gap in the non-permutation SLP case.
16692 2017-09-06 Martin Jambor <mjambor@suse.cz>
16694 PR tree-optimization/82078
16695 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16696 add_access_to_work_queue...
16697 (build_accesses_from_assign): ...here.
16698 (propagate_all_subaccesses): Make sure racc is the group
16699 representative, if there is one.
16701 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16703 PR middle-end/82095
16704 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16707 2017-09-06 Richard Biener <rguenther@suse.de>
16709 * gimple-ssa-strength-reduction.c
16710 (find_candidates_dom_walker::before_doom_children): Use a
16711 type and not a mode check.
16713 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16716 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16718 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16720 2017-09-05 Jeff Law <law@redhat.com>
16722 PR tree-optimization/64910
16723 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16724 cases where we have 3 or more operands.
16726 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16728 PR middle-end/81768
16729 * omp-low.c (lower_omp_for): Recompute tree invariant if
16730 gimple_omp_for_initial/final is ADDR_EXPR.
16732 PR middle-end/81768
16733 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16734 into gimple val before gimplification fo the COND_EXPR.
16736 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16738 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16739 REGION_COPY argument.
16740 (thread_through_all_blocks): Remove unused argument to
16741 duplicate_thread_path.
16743 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16744 Alan Hayward <alan.hayward@arm.com>
16745 David Sherwood <david.sherwood@arm.com>
16747 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16748 Take a scalar_mode rather than a machine_mode.
16749 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16750 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16751 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16752 (aarch64_gen_adjusted_ldpstp): Likewise.
16753 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16755 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16756 Alan Hayward <alan.hayward@arm.com>
16757 David Sherwood <david.sherwood@arm.com>
16759 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16760 Take a scalar_int_mode instead of a machine_mode.
16761 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16762 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16763 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16764 (aarch64_simd_attr_length_rglist): Delete.
16765 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16766 a scalar_int_mode instead of a machine_mode.
16767 (aarch64_add_offset): Likewise.
16768 (aarch64_internal_mov_immediate): Likewise
16769 (aarch64_add_constant_internal): Likewise.
16770 (aarch64_add_constant): Likewise.
16771 (aarch64_movw_imm): Likewise.
16772 (aarch64_rtx_arith_op_extract_p): Likewise.
16773 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16774 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16775 Remove assert that the mode isn't a vector.
16776 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16777 (aarch64_expand_mov_immediate): Update calls after above changes.
16778 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16779 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16780 (aarch64_move_imm): Likewise.
16781 (aarch64_can_const_movi_rtx_p): Likewise.
16782 (aarch64_strip_extend): Likewise.
16783 (aarch64_extr_rtx_p): Likewise.
16784 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16785 a CONST_INT when the mode parameter is VOIDmode.
16786 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16788 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16790 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
16791 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16792 (bitwise_type_for_mode): Update accordingly.
16794 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16796 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
16797 * stor-layout.c (mode_for_size_tree): Likewise.
16798 (mode_for_array): Update accordingly.
16799 (layout_decl): Likewise.
16800 (compute_record_mode): Likewise. Only set the mode once.
16802 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16804 * target.def (get_mask_mode): Change return type to opt_mode.
16806 * doc/tm.texi: Regenerate.
16807 * targhooks.h (default_get_mask_mode): Return an opt_mode.
16808 * targhooks.c (default_get_mask_mode): Likewise.
16809 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16810 * optabs-query.c (can_vec_mask_load_store_p): Update use of
16811 targetm.get_mask_mode.
16812 * tree.c (build_truth_vector_type): Likewise.
16814 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16816 * machmode.h (mode_for_vector): Return an opt_mode.
16817 * stor-layout.c (mode_for_vector): Likewise.
16818 (mode_for_int_vector): Update accordingly.
16819 (layout_type): Likewise.
16820 * config/i386/i386.c (emit_memmov): Likewise.
16821 (ix86_expand_set_or_movmem): Likewise.
16822 (ix86_expand_vector_init): Likewise.
16823 (ix86_get_mask_mode): Likewise.
16824 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16826 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16827 * expmed.c (extract_bit_field_1): Likewise.
16828 * expr.c (expand_expr_real_2): Likewise.
16829 * optabs-query.c (can_vec_perm_p): Likewise.
16830 (can_vec_mask_load_store_p): Likewise.
16831 * optabs.c (expand_vec_perm): Likewise.
16832 * targhooks.c (default_get_mask_mode): Likewise.
16833 * tree-vect-stmts.c (vectorizable_store): Likewise.
16834 (vectorizable_load): Likewise.
16835 (get_vectype_for_scalar_type_and_size): Likewise.
16837 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16839 * machmode.h (mode_for_int_vector): New function.
16840 * stor-layout.c (mode_for_int_vector): Likewise.
16841 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16842 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
16843 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16844 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16845 (s390_expand_vcond): Likewise.
16847 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16849 * machmode.h (opt_machine_mode): New type.
16850 (opt_mode<T>): Allow construction from anything that can be
16852 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16853 (mode_for_size): Return an opt_machine_mode.
16854 * stor-layout.c (mode_for_size): Likewise.
16855 (mode_for_size_tree): Update call accordingly.
16856 (bitwise_mode_for_mode): Likewise.
16857 (make_fract_type): Likewise.
16858 (make_accum_type): Likewise.
16859 * caller-save.c (replace_reg_with_saved_mem): Update call
16861 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16862 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16863 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16864 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16865 * expmed.c (extract_bit_field_1): Likewise.
16866 * reload.c (get_secondary_mem): Likewise.
16867 * varasm.c (assemble_integer): Likewise.
16868 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
16871 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16873 * machmode.h (decimal_float_mode_for_size): New function.
16874 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16875 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16876 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16877 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16878 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16879 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16881 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16883 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16884 (get_builtin_sync_mode): Likewise.
16885 (expand_ifn_atomic_compare_exchange): Likewise.
16886 (expand_builtin_atomic_clear): Likewise.
16887 (expand_builtin_atomic_test_and_set): Likewise.
16888 (fold_builtin_atomic_always_lock_free): Likewise.
16889 * calls.c (compute_argument_addresses): Likewise.
16890 (emit_library_call_value_1): Likewise.
16891 (store_one_arg): Likewise.
16892 * combine.c (combine_instructions): Likewise.
16893 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16894 * config/arm/arm.c (arm_function_value): Likewise.
16895 (aapcs_allocate_return_reg): Likewise.
16896 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16897 * config/i386/i386.c (construct_container): Likewise.
16898 (ix86_gimplify_va_arg): Likewise.
16899 (ix86_expand_sse_cmp): Likewise.
16900 (emit_memmov): Likewise.
16901 (emit_memset): Likewise.
16902 (expand_small_movmem_or_setmem): Likewise.
16903 (ix86_expand_pextr): Likewise.
16904 (ix86_expand_pinsr): Likewise.
16905 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16906 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16908 * config/mips/mips.c (mips_function_value_1) Likewise.
16909 (mips_block_move_straight): Likewise.
16910 (mips_expand_ins_as_unaligned_store): Likewise.
16911 * config/powerpcspe/powerpcspe.c
16912 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16913 (rs6000_darwin64_record_arg_flush): Likewise.
16914 * config/rs6000/rs6000.c
16915 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16916 (rs6000_darwin64_record_arg_flush): Likewise.
16917 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16918 (sparc_function_value_1): Likewise.
16919 * config/spu/spu.c (adjust_operand): Likewise.
16920 (spu_emit_branch_or_set): Likewise.
16921 (arith_immediate_p): Likewise.
16922 * emit-rtl.c (gen_lowpart_common): Likewise.
16923 * expr.c (expand_expr_real_1): Likewise.
16924 * function.c (assign_parm_setup_block): Likewise.
16925 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16926 * reload1.c (alter_reg): Likewise.
16927 * stor-layout.c (mode_for_vector): Likewise.
16928 (layout_type): Likewise.
16930 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16932 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16933 (spu_convert_move): Likewise.
16934 * lower-subreg.c (resolve_simple_move): Likewise.
16936 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16939 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16940 define_insn to a define_expand.
16941 (altivec_vsum2sws_direct): New define_insn.
16942 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16944 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16946 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16949 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16954 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16955 realign stack if stack alignment needed is less than incoming
16958 2017-09-05 Marek Polacek <polacek@redhat.com>
16961 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16964 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16966 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16968 2017-09-05 Richard Biener <rguenther@suse.de>
16970 PR tree-optimization/82084
16971 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16973 2017-09-05 Richard Biener <rguenther@suse.de>
16975 PR tree-optimization/82102
16976 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16978 2017-09-05 Martin Liska <mliska@suse.cz>
16980 PR tree-optimization/82032
16981 * tree-cfg.c (generate_range_test): New function.
16982 * tree-cfg.h (generate_range_test): Declared here.
16983 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16984 (cleanup_control_expr_graph): Use it.
16985 * tree-switch-conversion.c (try_switch_expansion): Remove
16987 (emit_case_nodes): Use generate_range_test.
16989 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16992 * config/i386/i386.md (*<btsc><mode>_mask): Add
16993 TARGET_USE_BT to insn constraint.
16994 (*btr<mode>_mask): Ditto.
16996 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16998 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16999 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17001 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17004 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17005 TARGET_NEON and TARGET_IWMMXT.
17006 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17007 TARGET_NEON and TARGET_IWMMXT.
17008 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17010 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17012 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17013 (ix86_rewrite_tls_address): Ditto.
17014 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17015 (ix86_rewrite_tls_address_1): Ditto.
17016 (ix86_rewrite_tls_address): Ditto.
17017 * config/i386/predicates.md (tls_address_pattern): New predicate.
17018 * config/i386/i386.md (TLS address splitter): New splitter.
17020 2017-09-04 Richard Biener <rguenther@suse.de>
17022 PR tree-optimization/82084
17023 * fold-const.h (can_native_encode_string_p): Declare.
17024 * fold-const.c (can_native_encode_string_p): Factor out from ...
17025 (native_encode_string): ... here.
17026 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17027 vectorizing stores from constants we later cannot handle.
17029 2017-09-04 Marek Polacek <polacek@redhat.com>
17032 * doc/invoke.texi: Update -Wtautological-compare documentation.
17034 2017-09-04 Jeff Law <law@redhat.com>
17036 PR tree-optimization/64910
17037 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17038 swap the first and last operand if the last is a constant.
17040 2017-09-04 Marek Polacek <polacek@redhat.com>
17043 * convert.c (do_narrow): When sanitizing signed integer overflows,
17044 bail out for signed types.
17045 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17047 2017-09-04 Richard Biener <rguenther@suse.de>
17049 PR tree-optimization/82060
17050 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17051 Move devirtualization after stmt folding and before EH/AB/noreturn
17052 cleanup to get the stmt refs canonicalized. Use a bool instead
17053 of gimple_modified_p since that doesn't work for NOPs. Schedule
17054 NOPs generated by folding for removal.
17056 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17057 Alan Hayward <alan.hayward@arm.com>
17058 David Sherwood <david.sherwood@arm.com>
17060 * coretypes.h (pad_direction): New enum.
17061 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17062 (FUNCTION_ARG_PADDING): Likewise.
17063 * target.def (function_arg_padding): New hook.
17064 * targhooks.h (default_function_arg_padding): Declare.
17065 * targhooks.c (default_function_arg_padding): New function.
17066 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17067 (TARGET_FUNCTION_ARG_PADDING): ...this.
17068 * doc/tm.texi: Regenerate.
17069 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17070 instead of direction.
17071 (compute_argument_addresses): Likewise.
17072 (load_register_parameters): Likewise.
17073 (emit_library_call_value_1): Likewise.
17074 (store_one_arg): Use targetm.calls.function_arg_padding instead
17075 of FUNCTION_ARG_PADDING.
17076 (must_pass_in_stack_var_size_or_pad): Likewise.
17077 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17078 (emit_group_store): Likewise.
17079 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17080 instead of FUNCTION_ARG_PADDING.
17081 (emit_push_insn): Likewise, and propagate enum change throughout
17083 * function.h (direction): Delete.
17084 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17086 * function.c (assign_parm_find_stack_rtl): Likewise.
17087 (assign_parm_setup_block_p): Likewise.
17088 (assign_parm_setup_block): Likewise.
17089 (gimplify_parameters): Likewise.
17090 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17091 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17093 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17094 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17095 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17096 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17097 (aarch64_function_arg_padding): ...this new function.
17098 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17099 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17100 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17101 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17102 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17103 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17104 (arm_pad_arg_upward): Replace with...
17105 (arm_function_arg_padding): ...this new function.
17106 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17108 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17109 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17110 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17111 (ia64_hpux_function_arg_padding): Replace with...
17112 (ia64_function_arg_padding): ...this new function. Use pad_direction
17113 instead of direction. Check for TARGET_HPUX.
17114 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17115 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17116 (iq2000_function_arg_padding): New function.
17117 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17118 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17119 (mips_function_arg_padding): ...this new function.
17120 (mips_pad_reg_upward): Update accordingly.
17121 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17122 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17123 targetm.calls.function_arg_padding.
17124 (FUNCTION_ARG_PADDING): Delete.
17125 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17126 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17127 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17128 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17129 (nios2_block_reg_padding): Return pad_direction instead of direction.
17130 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17131 instead of direction.
17132 (nios2_function_arg_padding): Likewise. Make static.
17133 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17134 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17135 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17136 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17137 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17138 (pa_function_arg_padding): Make static. Return pad_direction instead
17140 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17141 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17142 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17143 instead of direction. Use targetm.calls.function_arg_padding.
17144 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17145 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17146 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17147 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17148 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17150 (function_arg_padding): Rename to...
17151 (rs6000_function_arg_padding): ...this. Make static. Return
17152 pad_direction instead of direction.
17153 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17154 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17155 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17156 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17157 instead of direction. Use targetm.calls.function_arg_padding.
17158 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17159 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17160 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17161 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17162 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17163 (function_arg_padding): Rename to...
17164 (rs6000_function_arg_padding): ...this. Make static. Return
17165 pad_direction instead of direction.
17166 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17167 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17168 * config/s390/s390.c (s390_function_arg_padding): New function.
17169 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17170 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17171 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17172 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17173 (function_arg_padding): Rename to...
17174 (sparc_function_arg_padding): ...this. Make static. Return
17175 pad_direction instead of direction.
17176 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17177 * config/spu/spu.c (spu_function_arg_padding): New function.
17178 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17179 * system.h (FUNCTION_ARG_PADDING): Poison.
17181 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17182 Alan Hayward <alan.hayward@arm.com>
17183 David Sherwood <david.sherwood@arm.com>
17185 * target.def (modes_tieable_p): New hook.
17186 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17187 (TARGET_MODES_TIEABLE_P): ...this.
17188 * doc/tm.texi.in: Regenerate.
17189 * hooks.h (hook_bool_mode_mode_true): Declare.
17190 * hooks.c (hook_bool_mode_mode_true): New function.
17191 * combine.c (subst): Use targetm.modes_tieable_p instead of
17193 * dse.c (find_shift_sequence): Likewise.
17194 * expmed.c (extract_low_bits): Likewise.
17195 * lower-subreg.c: Include target.h.
17196 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17198 * rtlanal.c (rtx_cost): Likewise.
17199 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17200 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17201 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17202 (TARGET_MODES_TIEABLE_P): Redefine.
17203 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17204 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17205 (TARGET_MODES_TIEABLE_P): Redefine.
17206 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17207 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17208 (arc_modes_tieable_p): New function.
17209 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17210 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17211 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17212 (arm_modes_tieable_p): Make static.
17213 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17214 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17215 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17216 (TARGET_MODES_TIEABLE_P): Redefine.
17217 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17218 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17219 (TARGET_MODES_TIEABLE_P): Redefine.
17220 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17221 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17222 (cr16_modes_tieable_p): New function.
17223 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17224 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17225 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17226 (TRULY_NOOP_TRUNCATION): Update comment.
17227 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17228 (TRULY_NOOP_TRUNCATION): Update comment.
17229 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17230 (frv_modes_tieable_p): New function.
17231 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17232 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17233 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17234 (TARGET_MODES_TIEABLE_P): Redefine.
17235 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17236 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17237 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17238 (TARGET_MODES_TIEABLE_P): Redefine.
17239 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17240 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17241 (ia64_modes_tieable_p): New function.
17242 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17243 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17244 (iq2000_modes_tieable_p): New function.
17245 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17246 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17247 (lm32_modes_tieable_p): New function.
17248 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17249 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17250 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17251 (TARGET_MODES_TIEABLE_P): Redefine.
17252 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17253 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17254 (m32r_modes_tieable_p): New function.
17255 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17256 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17257 (m68k_modes_tieable_p): New function.
17258 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17259 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17260 (mcore_modes_tieable_p): New function.
17261 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17262 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17264 (TARGET_MODES_TIEABLE_P): Redefine.
17265 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17266 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17267 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17268 (TARGET_MODES_TIEABLE_P): Redefine.
17269 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17270 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17271 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17272 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17273 (mn10300_modes_tieable_p): ...this and make static.
17274 (TARGET_MODES_TIEABLE_P): Redefine.
17275 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17276 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17277 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17278 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17279 (msp430_modes_tieable_p): Make static.
17280 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17281 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17282 (TARGET_MODES_TIEABLE_P): Redefine.
17283 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17284 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17285 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17286 (TARGET_MODES_TIEABLE_P): Redefine.
17287 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17288 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17289 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17290 (TARGET_MODES_TIEABLE_P): Redefine.
17291 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17292 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17293 (pdp11_modes_tieable_p): New function.
17294 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17295 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17296 (rs6000_modes_tieable_p): New function.
17297 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17298 * config/powerpcspe/powerpcspe.md: Update comment.
17299 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17300 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17301 (TARGET_MODES_TIEABLE_P): Redefine.
17302 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17303 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17304 (rl78_modes_tieable_p): New function.
17305 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17306 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17307 (rs6000_modes_tieable_p): New function.
17308 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17309 * config/rs6000/rs6000.md: Update comment.
17310 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17311 * config/rx/rx.c (rx_modes_tieable_p): New function.
17312 (TARGET_MODES_TIEABLE_P): Redefine.
17313 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17314 * config/s390/s390.c (s390_modes_tieable_p): New function.
17315 (TARGET_MODES_TIEABLE_P): Redefine.
17316 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17317 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17318 (sh_modes_tieable_p): New function.
17319 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17320 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17321 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17322 (sparc_modes_tieable_p): Make static.
17323 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17324 * config/spu/spu.c (spu_modes_tieable_p): New function.
17325 (TARGET_MODES_TIEABLE_P): Redefine.
17326 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17327 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17328 (TARGET_MODES_TIEABLE_P): Redefine.
17329 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17330 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17331 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17332 * config/v850/v850.c (v850_modes_tieable_p): New function.
17333 (TARGET_MODES_TIEABLE_P): Redefine.
17334 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17335 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17336 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17337 (visium_modes_tieable_p): New function.
17338 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17339 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17340 (xtensa_modes_tieable_p): New function.
17341 * system.h (MODES_TIEABLE_P): Poison.
17343 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17344 Alan Hayward <alan.hayward@arm.com>
17345 David Sherwood <david.sherwood@arm.com>
17347 * target.def (hard_regno_mode_ok): New hook.
17348 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17349 (TARGET_HARD_REGNO_MODE_OK): ...this.
17350 * doc/tm.texi.in: Regenerate.
17351 * hooks.h (hook_bool_uint_mode_true): Declare.
17352 * hooks.c (hook_bool_uint_mode_true): New function.
17353 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17354 HARD_REGNO_MODE_OK.
17355 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17356 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17357 instead of HARD_REGNO_MODE_OK.
17358 * caller-save.c: Include target.h.
17359 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17360 HARD_REGNO_MODE_OK.
17361 * combine.c (can_combine_p): Likewise.
17362 (combinable_i3pat): Likewise.
17363 (can_change_dest_mode): Likewise.
17364 * expr.c (init_expr_target): Likewise.
17365 (convert_move): Likewise.
17366 (convert_modes): Likewise.
17367 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17368 (setup_prohibited_mode_move_regs): Likewise.
17369 * ira.h (target_ira): Likewise.
17370 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17371 * lra-constraints.c (process_alt_operands): Likewise.
17372 (split_reg): Likewise.
17373 * recog.c (peep2_find_free_register): Likewise.
17374 * ree.c (combine_reaching_defs): Likewise.
17375 * regcprop.c (maybe_mode_change): Likewise.
17376 * reginfo.c (init_reg_sets_1): Likewise.
17377 (choose_hard_reg_mode): Likewise.
17378 (simplifiable_subregs): Likewise.
17379 * regrename.c (check_new_reg_p): Likewise.
17380 * reload.c (find_valid_class): Likewise.
17381 (find_valid_class_1): Likewise.
17382 (reload_inner_reg_of_subreg): Likewise.
17383 (push_reload): Likewise.
17384 (combine_reloads): Likewise.
17385 (find_dummy_reload): Likewise.
17386 (find_reloads): Likewise.
17387 * reload1.c (find_reg): Likewise.
17388 (set_reload_reg): Likewise.
17389 (allocate_reload_reg): Likewise.
17390 (choose_reload_regs): Likewise.
17391 (reload_adjust_reg_for_temp): Likewise.
17392 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17393 (simplify_subreg_regno): Likewise.
17394 * sel-sched.c (init_regs_for_mode): Likewise.
17395 * varasm.c (make_decl_rtl): Likewise.
17396 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17397 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17398 HARD_REGNO_MODE_OK.
17399 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17400 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17401 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17402 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17403 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17404 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17405 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17406 (arc_mode_class): Delete.
17407 (HARD_REGNO_MODE_OK): Delete.
17408 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17409 (arc_hard_regno_mode_ok): Rename old array to...
17410 (arc_hard_regno_mode_ok_modes): ...this.
17411 (arc_conditional_register_usage): Update accordingly.
17412 (arc_mode_class): Make static.
17413 (arc_hard_regno_mode_ok): New function.
17414 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17415 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17416 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17417 (arm_hard_regno_mode_ok): Make static.
17418 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17419 HARD_REGNO_MODE_OK.
17420 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17421 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17422 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17424 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17425 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17426 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17427 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17428 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17429 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17430 * config/bfin/predicates.md (valid_reg_operand): Use
17431 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17432 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17433 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17434 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17435 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17436 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17437 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17438 (cr16_hard_regno_mode_ok): Make static and return a bool.
17439 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17440 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17441 (cris_hard_regno_mode_ok): New function.
17442 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17443 (epiphany_mode_class): Delete.
17444 (HARD_REGNO_MODE_OK): Delete.
17445 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17446 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17447 (hard_regno_mode_ok): Rename to...
17448 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17449 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17450 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17451 HARD_REGNO_MODE_OK.
17452 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17453 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17454 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17455 (frv_hard_regno_mode_ok): Make static and return a bool.
17456 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17457 HARD_REGNO_MODE_OK.
17458 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17459 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17460 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17461 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17463 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17464 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17465 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17466 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17468 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17469 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17470 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17471 (ia64_hard_regno_mode_ok): New function.
17472 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17473 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17474 (iq2000_hard_regno_mode_ok): New function.
17475 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17476 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17477 (lm32_hard_regno_mode_ok): New function.
17478 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17479 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17480 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17481 instead of HARD_REGNO_MODE_OK.
17482 (m32c_hard_regno_ok): Rename to...
17483 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17484 (m32c_cannot_change_mode_class): Update accordingly.
17485 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17486 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17487 (m32r_mode_class): Delete.
17488 (HARD_REGNO_MODE_OK): Delete.
17489 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17490 (m32r_hard_regno_mode_ok): Rename to...
17491 (m32r_hard_regno_modes): ...this.
17492 (m32r_mode_class): Make static.
17493 (m32r_hard_regno_mode_ok): New function.
17494 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17495 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17496 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17497 (m68k_hard_regno_mode_ok): Make static.
17498 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17499 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17500 (mcore_hard_regno_mode_ok): New function.
17501 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17502 (HARD_REGNO_MODE_OK): Delete.
17503 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17505 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17506 (microblaze_hard_regno_mode_ok): New function.
17507 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17508 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17509 (mips_hard_regno_mode_ok): Delete.
17510 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17511 (mips_hard_regno_mode_ok_p): ...this and make static.
17512 (mips_hard_regno_mode_ok_p): Rename to...
17513 (mips_hard_regno_mode_ok_uncached): ...this.
17514 (mips_hard_regno_mode_ok): New function.
17515 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17516 of HARD_REGNO_MODE_OK.
17517 (mips_option_override): Update after above name changes.
17518 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17519 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17520 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17521 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17522 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17523 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17524 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17525 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17526 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17527 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17528 (msp430_hard_regno_mode_ok): Make static and return a bool.
17529 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17530 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17531 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17533 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17534 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17535 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17536 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17537 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17538 (PA_HARD_REGNO_MODE_OK): ...this
17539 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17540 (PA_HARD_REGNO_MODE_OK): ...this.
17541 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17542 (pa_hard_regno_mode_ok): New function.
17543 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17544 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17545 (pdp11_hard_regno_mode_ok): New function.
17546 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17547 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17549 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17551 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17552 (rs6000_hard_regno_mode_ok): Rename to...
17553 (rs6000_hard_regno_mode_ok_uncached): ...this.
17554 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17555 (rs6000_hard_regno_mode_ok): New function.
17556 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17557 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17558 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17559 (riscv_hard_regno_mode_ok): ...this and make static.
17560 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17561 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17562 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17563 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17564 (rl78_hard_regno_mode_ok): Make static and return bool.
17565 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17566 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17568 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17569 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17570 (rs6000_hard_regno_mode_ok): Rename to...
17571 (rs6000_hard_regno_mode_ok_uncached): ...this.
17572 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17573 (rs6000_hard_regno_mode_ok): New function.
17574 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17575 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17576 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17577 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17578 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17579 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17580 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17581 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17582 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17583 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17584 (sh_hard_regno_mode_ok): Make static.
17585 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17586 instead of HARD_REGNO_MODE_OK.
17587 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17588 (sparc_mode_class): Delete.
17589 (HARD_REGNO_MODE_OK): Delete.
17590 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17591 (hard_regno_mode_classes): Make static.
17592 (sparc_mode_class): Likewise.
17593 (sparc_hard_regno_mode_ok): New function.
17594 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17595 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17596 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17598 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17599 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17600 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17601 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17602 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17603 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17604 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17605 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17606 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17607 (visium_hard_regno_mode_ok): New function.
17608 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17609 instead of HARD_REGNO_MODE_OK.
17610 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17611 (HARD_REGNO_MODE_OK): Delete.
17612 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17613 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17614 (xtensa_option_override): Update accordingly.
17615 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17616 (xtensa_hard_regno_mode_ok): New function.
17617 * system.h (HARD_REGNO_MODE_OK): Poison.
17619 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17620 Alan Hayward <alan.hayward@arm.com>
17621 David Sherwood <david.sherwood@arm.com>
17623 * target.def (hard_regno_call_part_clobbered): New hook.
17624 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17625 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17626 * doc/tm.texi: Regenerate.
17627 * hooks.h (hook_bool_uint_mode_false): Declare.
17628 * hooks.c (hook_bool_uint_mode_false): New function.
17629 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17630 * cselib.c (cselib_process_insn): Use
17631 targetm.hard_regno_call_part_clobbered instead of
17632 HARD_REGNO_CALL_PART_CLOBBERED.
17633 * ira-conflicts.c (ira_build_conflicts): Likewise.
17634 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17635 * lra-constraints.c (need_for_call_save_p): Likewise.
17636 * lra-lives.c: Include target.h.
17637 (check_pseudos_live_through_calls): Use
17638 targetm.hard_regno_call_part_clobbered instead of
17639 HARD_REGNO_CALL_PART_CLOBBERED.
17640 * regcprop.c: Include target.h.
17641 (copyprop_hardreg_forward_1): Use
17642 targetm.hard_regno_call_part_clobbered instead of
17643 HARD_REGNO_CALL_PART_CLOBBERED.
17644 * reginfo.c (choose_hard_reg_mode): Likewise.
17645 * regrename.c (check_new_reg_p): Likewise.
17646 * reload.c (find_equiv_reg): Likewise.
17647 * reload1.c (emit_reload_insns): Likewise.
17648 * sched-deps.c (deps_analyze_insn): Likewise.
17649 * sel-sched.c (init_regs_for_mode): Likewise.
17650 (mark_unavailable_hard_regs): Likewise.
17651 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17652 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17653 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17655 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17656 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17657 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17659 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17661 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17662 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17663 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17665 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17666 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17667 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17669 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17670 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17672 * config/powerpcspe/powerpcspe.c
17673 (rs6000_hard_regno_call_part_clobbered): New function.
17674 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17675 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17676 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17678 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17679 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17680 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17682 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17683 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17684 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17686 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17687 Alan Hayward <alan.hayward@arm.com>
17688 David Sherwood <david.sherwood@arm.com>
17690 * rtl.h (subreg_memory_offset): Declare.
17691 * emit-rtl.c (subreg_memory_offset): New function.
17692 * expmed.c (store_bit_field_1): Use it.
17693 * expr.c (undefined_operand_subword_p): Likewise.
17694 * simplify-rtx.c (simplify_subreg): Likewise.
17696 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17698 PR rtl-optimization/57448
17701 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17702 using atomic_load pattern.
17703 (expand_atomic_store): Likewise.
17705 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17708 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17709 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17710 handling. Use replace_call_with_value with NULL instead of
17711 gsi_replace, unlink_stmt_vdef and release_defs.
17713 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17716 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17718 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17721 * rtl.h (emit_library_call_value_1): Declare.
17722 (emit_library_call): Replace declaration with a series of overloads.
17723 Remove the parameter count argument.
17724 (emit_library_call_value): Likewise.
17725 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17726 with an "rtx_mode_t *".
17727 (emit_library_call_value): Delete.
17728 (emit_library_call): Likewise.
17729 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17730 (asan_emit_allocas_unpoison): Likewise.
17731 * builtins.c (expand_builtin_powi): Likewise.
17732 (expand_asan_emit_allocas_unpoison): Likewise.
17733 * cfgexpand.c (expand_main_function): Likewise.
17734 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17735 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17736 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17737 * config/arm/arm.c (arm_trampoline_init): Likewise.
17738 (arm_call_tls_get_addr): Likewise.
17739 (arm_expand_divmod_libfunc): Likewise.
17740 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17741 (smulsi3_highpart): Likewise.
17742 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17743 (c6x_expand_compare): Likewise.
17744 (c6x_expand_movmem): Likewise.
17745 * config/frv/frv.c (frv_trampoline_init): Likewise.
17746 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17747 (ix86_expand_divmod_libfunc): Likewise.
17748 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17749 (ia64_expand_compare): Likewise.
17750 (ia64_profile_hook): Likewise.
17751 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17752 (nonlocal_goto): Likewise.
17753 (restore_stack_nonlocal): Likewise.
17754 * config/m32r/m32r.c (block_move_call): Likewise.
17755 (m32r_trampoline_init): Likewise.
17756 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17757 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17758 (m68k_call_m68k_read_tp): Likewise.
17759 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17760 (microblaze_expand_divide): Likewise.
17761 * config/mips/mips.h (mips_args): Likewise.
17762 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17763 (MIPS_ICACHE_SYNC): Likewise.
17764 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17765 (nios2_trampoline_init): Likewise.
17766 * config/pa/pa.c (hppa_tls_call): Likewise.
17767 (pa_trampoline_init): Likewise.
17768 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17769 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17770 (expand_strn_compare): Likewise.
17771 (rs6000_generate_compare): Likewise.
17772 (rs6000_expand_float128_convert): Likewise.
17773 (output_profile_hook): Likewise.
17774 (rs6000_trampoline_init): Likewise.
17775 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17776 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17777 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17778 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17779 (rs6000_generate_compare): Likewise.
17780 (rs6000_expand_float128_convert): Likewise.
17781 (output_profile_hook): Likewise.
17782 (rs6000_trampoline_init): Likewise.
17783 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17784 * config/sh/sh.c (sh_trampoline_init): Likewise.
17785 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17786 (sparc_emit_float_lib_cmp): Likewise.
17787 (sparc32_initialize_trampoline): Likewise.
17788 (sparc64_initialize_trampoline): Likewise.
17789 (sparc_profile_hook): Likewise.
17790 * config/spu/spu.c (ea_load_store): Likewise.
17791 * config/spu/spu.md (floatunssidf2): Likewise.
17792 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
17793 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
17794 * config/visium/visium.c (expand_block_move_4): Likewise.
17795 (expand_block_move_2): Likewise.
17796 (expand_block_move_1): Likewise.
17797 (expand_block_set_4): Likewise.
17798 (expand_block_set_2): Likewise.
17799 (expand_block_set_1): Likewise.
17800 (visium_trampoline_init): Likewise.
17801 (visium_profile_hook): Likewise.
17802 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
17803 (xtensa_setup_frame_addresses): Likewise.
17804 (xtensa_trampoline_init): Likewise.
17805 * except.c (sjlj_emit_function_enter): Likewise.
17806 (sjlj_emit_function_exit): Likewise.
17807 * explow.c (allocate_dynamic_stack_space): Likewise.
17808 (probe_stack_range): Likewise.
17809 * expr.c (convert_mode_scalar): Likewise.
17810 * optabs.c (expand_binop): Likewise.
17811 (expand_twoval_binop_libfunc): Likewise.
17812 (expand_unop): Likewise.
17813 (prepare_cmp_insn): Likewise.
17814 (prepare_float_lib_cmp): Likewise.
17815 (expand_float): Likewise.
17816 (expand_fix): Likewise.
17817 (expand_fixed_convert): Likewise.
17818 (maybe_emit_sync_lock_test_and_set): Likewise.
17819 (expand_atomic_compare_and_swap): Likewise.
17820 (expand_mem_thread_fence): Likewise.
17821 (expand_atomic_fetch_op): Likewise.
17823 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17825 * doc/generic.texi (OpenACC): Adjust URL.
17826 * doc/invoke.texi (C Dialect Options): Ditto.
17828 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17830 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17831 predicate for operand 1. Add (m,<S>) constraint.
17832 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17833 Prevent memory operand 1 with register operand 2.
17835 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17837 PR rtl-optimization/82024
17838 * combine.c (try_combine): If the combination result is a PARALLEL,
17839 and we only need to retain the SET in there that would be placed
17840 at I2, check that we can place that at I3 instead, before doing so.
17842 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17845 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
17847 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
17848 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17851 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17852 Jeff Law <law@redhat.com>
17854 * varasm.c (bss_initializer_p): Do not put constants into .bss
17855 (categorize_decl_for_section): Handle bss_initializer_p returning
17856 false when DECL_INITIAL is NULL.
17858 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17861 * config/s390/s390.c (s390_can_inline_p): New function.
17863 2017-09-01 Jeff Law <law@redhat.com>
17865 PR tree-optimization/82052
17866 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17867 Always initialize the returned slot after a hash table miss
17868 when INSERT is true.
17870 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
17872 * config/s390/s390.md (mem_signal_fence): Remove.
17873 * doc/md.texi (mem_signal_fence): Remove.
17874 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17876 * target-insns.def (mem_signal_fence): Remove.
17878 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17881 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17884 * asan.c (create_odr_indicator): Strip name encoding from assembler
17885 name before appending it after __odr_asan_.
17887 2017-09-01 Martin Liska <mliska@suse.cz>
17889 PR tree-optimization/82059
17890 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17891 frequency only when an edge is redirected.
17893 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17895 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17896 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17897 (arc_conditional_register_usage): Remove ARC600 lp_count
17899 (arc_file_start): Emit Tag_ARC_CPU_variation.
17900 (arc_can_use_doloop_p): New conditions to use ZOLs.
17901 (hwloop_fail): New function.
17902 (hwloop_optimize): Likewise.
17903 (hwloop_pattern_reg): Likewise.
17904 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17905 (arc_reorg_loops): New function, calls reorg_loops.
17906 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17907 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17909 (arc_loop_hazard): Remove function, functionality moved into
17911 (arc_hazard): Remove arc_loop_hazard call.
17912 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17913 into hwloop_optimize.
17914 (arc_label_align): Remove ZOL handling.
17915 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17916 * config/arc/arc.md (doloop_begin): Remove pattern.
17917 (doloop_begin_i): Likewise.
17918 (doloop_end_i): Likewise.
17919 (doloop_fallback): Likewise.
17920 (doloop_fallback_m): Likewise.
17921 (doloop_end): Reimplement expand.
17922 (arc_lp): New pattern for LP instruction.
17923 (loop_end): New pattern.
17924 (loop_fail): Likewise.
17925 (decrement_and_branch_until_zero): Likewise.
17926 * config/arc/arc.opt (mlpc-width): New option.
17927 * doc/invoke.texi (mlpc-width): Document option.
17929 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17931 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17932 (arc_ccfsm_advance): Fix checking for delay slots.
17933 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17935 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17937 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17939 (movhi_insn): Update store instruction constraint which are saving
17940 6-bit short immediates.
17941 (movsi_insn): Consider also short scaled load operations.
17942 (zero_extendhisi2_i): Use Usd constraint instead of T.
17943 (extendhisi2_i): Add q constraint.
17944 (arc_clzsi2): Add type and length attributes.
17945 (arc_ctzsi2): Likewise.
17946 * config/arc/constraints.md (Usc): Update constraint, the
17947 assembler can parse two relocations for a single instruction.
17949 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17951 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17952 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17954 2017-08-31 Olivier Hainque <hainque@adacore.com>
17956 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17957 match as powerpc-wrs-vxworks*.
17959 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17961 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17962 register constraint for by-element operand.
17963 (aarch64_mls_elt_merge<mode>): Likewise.
17965 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17967 * config/arc/arc.c (arc_can_follow_jump): Check for short
17970 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17972 * config.gcc: Use g.opt for arc.
17973 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17974 functionality moved to ...
17975 (legitimate_scaled_address_p): New function, ...here.
17976 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17977 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17978 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17980 (arc_override_options): Handle G option.
17981 (arc_output_pic_addr_const): Correct function definition.
17982 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17983 (arc_decl_anon_ns_mem_p): Delete.
17984 (arc_in_small_data_p): Overhaul this function to take into
17985 consideration the value given via G option.
17986 (arc_rewrite_small_data_1): Renamed and corrected old
17987 arc_rewrite_small_data function.
17988 (arc_rewrite_small_data): New function.
17989 (small_data_pattern): Don't use pic_offset_table_rtx.
17990 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17991 * config/arc/simdext.md (movmisalignv2hi): Use
17992 prepare_move_operands function.
17994 (movmisalign*): Likewise.
17995 * doc/invoke.texi (ARC options): Document -G option.
17997 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17999 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18001 * config/arc/arc.c (arc_print_operand): Output scalled address for
18002 sdata whenever is possible.
18003 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18004 load/stores are available.
18005 (compact_sda_memory_operand): Check for the alignment required by
18006 code density instructions.
18007 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18009 * config/arc/constraints.md (Usd): Update constraint.
18010 (Us0): New constraint.
18011 (Usc): Update constraint.
18013 2017-08-31 Richard Biener <rguenther@suse.de>
18015 PR middle-end/82054
18016 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18017 function only once.
18019 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18021 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18022 Resize type_signature.
18024 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18025 Alan Hayward <alan.hayward@arm.com>
18026 David Sherwood <david.sherwood@arm.com>
18028 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18029 subregs whose inner modes can be stored in GPRs.
18030 (aarch64_classify_index): Likewise.
18032 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18033 Alan Hayward <alan.hayward@arm.com>
18034 David Sherwood <david.sherwood@arm.com>
18036 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18037 (V_INT_EQUIV): ...this.
18038 (v_cmp_result): Rename to...
18039 (v_int_equiv): ...this.
18040 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18041 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18042 (copysign<mode>3): Likewise.
18043 (aarch64_simd_bsl<mode>_internal): Likewise.
18044 (aarch64_simd_bsl<mode>): Likewise.
18045 (vec_cmp<mode><mode>): Likewise.
18046 (vcond<mode><mode>): Likewise.
18047 (vcond<v_cmp_mixed><mode>): Likewise.
18048 (vcondu<mode><v_cmp_mixed>): Likewise.
18049 (aarch64_cm<optab><mode>): Likewise.
18050 (aarch64_cmtst<mode>): Likewise.
18051 (aarch64_fac<optab><mode>): Likewise.
18052 (vec_perm_const<mode>): Likewise.
18053 (vcond_mask_<mode><v_cmp_result>): Rename to...
18054 (vcond_mask_<mode><v_int_equiv>): ...this.
18055 (vec_cmp<mode><v_cmp_result>): Rename to...
18056 (vec_cmp<mode><v_int_equiv>): ...this.
18058 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18059 Alan Hayward <alan.hayward@arm.com>
18060 David Sherwood <david.sherwood@arm.com>
18062 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18064 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18065 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18066 (UNSPEC_LD4_DREG): New unspecs.
18067 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18068 (aarch64_ld2<mode>_dreg_be): Replace with...
18069 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18071 (aarch64_ld3<mode>_dreg_le)
18072 (aarch64_ld3<mode>_dreg_be): Replace with...
18073 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18075 (aarch64_ld4<mode>_dreg_le)
18076 (aarch64_ld4<mode>_dreg_be): Replace with...
18077 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18080 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18082 PR tree-optimization/81987
18083 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18084 insert an initializer in a location not dominated by the stride
18087 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18089 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18090 on the entire header of the finally block in the fallthru case.
18092 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18094 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18096 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18098 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18099 rs6000_emit_move_from_cr and call renamed function.
18100 (rs6000_emit_prologue): Call renamed functions.
18101 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18102 movesi_from_cr, remove volatile CRs.
18104 2017-08-30 Jon Beniston <jon@beniston.com>
18105 Richard Biener <rguenther@suse.de>
18107 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18108 of VECTOR_MODE_P check.
18109 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18110 element vector types.
18112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18114 * df.h (df_read_modify_subreg_p): Remove in favor of...
18115 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18116 const_rtx instead of an rtx.
18117 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18118 * df-problems.c (df_word_lr_mark_ref): Likewise.
18119 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18120 (mark_pseudo_reg_dead): Likewise.
18121 (mark_ref_dead): Likewise.
18122 * reginfo.c (init_subregs_of_mode): Likewise.
18123 * sched-deps.c (sched_analyze_1): Likewise.
18124 * df-scan.c (df_def_record_1): Likewise.
18125 (df_uses_record): Likewise.
18126 (df_read_modify_subreg_p): Remove in favor of...
18127 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18128 const_rtx instead of an rtx.
18130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18131 Alan Hayward <alan.hayward@arm.com>
18132 David Sherwood <david.sherwood@arm.com>
18134 * rtl.h (partial_subreg_p): New function.
18135 * caller-save.c (save_call_clobbered_regs): Use it.
18136 * calls.c (expand_call): Likewise.
18137 * combine.c (combinable_i3pat): Likewise.
18138 (simplify_set): Likewise.
18139 (make_extraction): Likewise.
18140 (make_compound_operation_int): Likewise.
18141 (gen_lowpart_or_truncate): Likewise.
18142 (force_to_mode): Likewise.
18143 (make_field_assignment): Likewise.
18144 (reg_truncated_to_mode): Likewise.
18145 (record_truncated_value): Likewise.
18146 (move_deaths): Likewise.
18147 * cse.c (record_jump_cond): Likewise.
18148 (cse_insn): Likewise.
18149 * cselib.c (cselib_lookup_1): Likewise.
18150 * expmed.c (extract_bit_field_using_extv): Likewise.
18151 * function.c (assign_parm_setup_reg): Likewise.
18152 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18153 * ira-build.c (create_insn_allocnos): Likewise.
18154 * lra-coalesce.c (merge_pseudos): Likewise.
18155 * lra-constraints.c (match_reload): Likewise.
18156 (simplify_operand_subreg): Likewise.
18157 (curr_insn_transform): Likewise.
18158 * lra-lives.c (process_bb_lives): Likewise.
18159 * lra.c (new_insn_reg): Likewise.
18160 (lra_substitute_pseudo): Likewise.
18161 * regcprop.c (mode_change_ok): Likewise.
18162 (maybe_mode_change): Likewise.
18163 (copyprop_hardreg_forward_1): Likewise.
18164 * reload.c (push_reload): Likewise.
18165 (find_reloads): Likewise.
18166 (find_reloads_subreg_address): Likewise.
18167 * reload1.c (alter_reg): Likewise.
18168 (eliminate_regs_1): Likewise.
18169 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18171 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18173 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18174 back to if statements, including unpack.
18176 2017-08-30 Martin Liska <mliska@suse.cz>
18178 PR inline-asm/82001
18179 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18181 (func_checker::compare_asm_inputs_outputs): ... this function.
18182 (func_checker::compare_gimple_asm): Use the function to compare
18183 also ASM constrains.
18184 * ipa-icf-gimple.h: Rename the function.
18186 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18187 Alan Hayward <alan.hayward@arm.com>
18188 David Sherwood <david.sherwood@arm.com>
18190 * coretypes.h (complex_mode): New type.
18191 * gdbhooks.py (build_pretty_printer): Handle it.
18192 * machmode.h (complex_mode): New class.
18193 (complex_mode::includes_p): New function.
18194 (is_complex_int_mode): Likewise.
18195 (is_complex_float_mode): Likewise.
18196 * genmodes.c (get_mode_class): Handle complex mode classes.
18197 * function.c (expand_function_end): Use is_complex_int_mode.
18199 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18200 Alan Hayward <alan.hayward@arm.com>
18201 David Sherwood <david.sherwood@arm.com>
18203 * coretypes.h (scalar_mode_pod): New typedef.
18204 * gdbhooks.py (build_pretty_printer): Handle it.
18205 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18206 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18207 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18208 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18209 as_a <scalar_mode>.
18211 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18212 Alan Hayward <alan.hayward@arm.com>
18213 David Sherwood <david.sherwood@arm.com>
18215 * machmode.h (mode_for_vector): Take a scalar_mode instead
18217 * stor-layout.c (mode_for_vector): Likewise.
18218 * explow.c (promote_mode): Use as_a <scalar_mode>.
18219 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18222 Alan Hayward <alan.hayward@arm.com>
18223 David Sherwood <david.sherwood@arm.com>
18225 * target.def (preferred_simd_mode): Take a scalar_mode
18226 instead of a machine_mode.
18227 * targhooks.h (default_preferred_simd_mode): Likewise.
18228 * targhooks.c (default_preferred_simd_mode): Likewise.
18229 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18230 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18231 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18232 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18233 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18234 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18235 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18236 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18238 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18239 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18240 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18241 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18242 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18243 * doc/tm.texi: Regenerate.
18244 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18248 Alan Hayward <alan.hayward@arm.com>
18249 David Sherwood <david.sherwood@arm.com>
18251 * target.def (scalar_mode_supported_p): Take a scalar_mode
18252 instead of a machine_mode.
18253 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18254 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18255 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18256 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18257 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18258 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18259 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18260 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18261 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18262 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18263 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18264 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18265 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18266 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18268 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18269 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18270 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18271 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18272 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18274 * doc/tm.texi: Regenerate.
18276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18277 Alan Hayward <alan.hayward@arm.com>
18278 David Sherwood <david.sherwood@arm.com>
18280 * coretypes.h (opt_scalar_mode): New typedef.
18281 * gdbhooks.py (build_pretty_printers): Handle it.
18282 * machmode.h (mode_iterator::get_2xwider): Add overload for
18284 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18286 * expr.c (convert_mode_scalar): Likewise.
18287 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18288 * optabs.c (expand_float): Likewise.
18289 (expand_fix): Likewise.
18290 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18293 Alan Hayward <alan.hayward@arm.com>
18294 David Sherwood <david.sherwood@arm.com>
18296 * optabs.c (expand_float): Explicitly check for scalars before
18297 using a branching expansion.
18298 (expand_fix): Likewise.
18300 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18301 Alan Hayward <alan.hayward@arm.com>
18302 David Sherwood <david.sherwood@arm.com>
18304 * expr.c (convert_mode): Split scalar handling out into...
18305 (convert_mode_scalar): ...this new function. Treat the modes
18308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18309 Alan Hayward <alan.hayward@arm.com>
18310 David Sherwood <david.sherwood@arm.com>
18312 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18314 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18317 Alan Hayward <alan.hayward@arm.com>
18318 David Sherwood <david.sherwood@arm.com>
18320 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18321 rather than a machine_mode.
18322 (fixed_from_string): Likewise.
18323 (fixed_convert): Likewise.
18324 (fixed_convert_from_int): Likewise.
18325 (fixed_convert_from_real): Likewise.
18326 (real_convert_from_fixed): Likewise.
18327 * fixed-value.c (fixed_from_double_int): Likewise.
18328 (fixed_from_string): Likewise.
18329 (fixed_convert): Likewise.
18330 (fixed_convert_from_int): Likewise.
18331 (fixed_convert_from_real): Likewise.
18332 (real_convert_from_fixed): Likewise.
18333 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18336 Alan Hayward <alan.hayward@arm.com>
18337 David Sherwood <david.sherwood@arm.com>
18339 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18340 of separate mode class checks. Do not allow vector modes here.
18341 (immed_wide_int_const): Use as_a <scalar_mode>.
18342 * explow.c (trunc_int_for_mode): Likewise.
18343 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18344 (wi::shwi): Likewise.
18345 (wi::min_value): Likewise.
18346 (wi::max_value): Likewise.
18347 * dwarf2out.c (loc_descriptor): Likewise.
18348 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18349 for CONST_WIDE_INT.
18351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18352 Alan Hayward <alan.hayward@arm.com>
18353 David Sherwood <david.sherwood@arm.com>
18355 * tree.h (SCALAR_TYPE_MODE): New macro.
18356 * expr.c (expand_expr_addr_expr_1): Use it.
18357 (expand_expr_real_2): Likewise.
18358 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18359 (fold_convert_const_fixed_from_int): Likewise.
18360 (fold_convert_const_fixed_from_real): Likewise.
18361 (native_encode_fixed): Likewise
18362 (native_encode_complex): Likewise
18363 (native_encode_vector): Likewise.
18364 (native_interpret_fixed): Likewise.
18365 (native_interpret_real): Likewise.
18366 (native_interpret_complex): Likewise.
18367 (native_interpret_vector): Likewise.
18368 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18369 (simd_clone_adjust_argument_types): Likewise.
18370 (simd_clone_init_simd_arrays): Likewise.
18371 (simd_clone_adjust): Likewise.
18372 * stor-layout.c (layout_type): Likewise.
18373 * tree.c (build_minus_one_cst): Likewise.
18374 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18375 * tree-inline.c (estimate_move_cost): Likewise.
18376 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18377 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18378 (vectorizable_reduction): Likewise.
18379 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18380 (vect_recog_mixed_size_cond_pattern): Likewise.
18381 (check_bool_pattern): Likewise.
18382 (adjust_bool_pattern): Likewise.
18383 (search_type_for_mask_1): Likewise.
18384 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18385 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18386 (vectorizable_load): Likewise.
18387 (vectorizable_store): Likewise.
18388 * ubsan.c (ubsan_encode_value): Likewise.
18389 * varasm.c (output_constant): Likewise.
18391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18392 Alan Hayward <alan.hayward@arm.com>
18393 David Sherwood <david.sherwood@arm.com>
18395 * coretypes.h (scalar_mode): New class.
18396 * machmode.h (scalar_mode): Likewise.
18397 (scalar_mode::includes_p): New function.
18398 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18399 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18400 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18401 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18402 * expmed.c (store_bit_field_1): Likewise.
18403 (extract_bit_field_1): Likewise.
18404 * expr.c (write_complex_part): Likewise.
18405 (read_complex_part): Likewise.
18406 (emit_move_complex_push): Likewise.
18407 (expand_expr_real_2): Likewise.
18408 * function.c (assign_parm_setup_reg): Likewise.
18409 (assign_parms_unsplit_complex): Likewise.
18410 * optabs.c (expand_binop): Likewise.
18411 * rtlanal.c (subreg_get_info): Likewise.
18412 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18413 * varasm.c (output_constant_pool_2): Likewise.
18415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18416 Alan Hayward <alan.hayward@arm.com>
18417 David Sherwood <david.sherwood@arm.com>
18419 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18421 (expmed_mult_highpart_optab): Likewise.
18422 (expmed_mult_highpart): 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 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18429 instead of a machine_mode.
18430 (builtin_memset_read_str): Likewise.
18431 * builtins.c (c_readstr): Likewise.
18432 (builtin_memcpy_read_str): Likewise.
18433 (builtin_strncpy_read_str): Likewise.
18434 (builtin_memset_read_str): Likewise.
18435 (builtin_memset_gen_str): Likewise.
18436 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18437 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18438 instead of a machine_mode.
18439 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18441 (make_extraction): Likewise.
18442 (try_widen_shift_mode): Take and return scalar_int_modes instead
18444 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18445 a scalar_int_mode instead of a machine_mode.
18446 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18447 (avr_addr_space_pointer_mode): Likewise.
18448 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18449 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18450 (msp430_unwind_word_mode): Likewise.
18451 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18452 (spu_addr_space_pointer_mode): Likewise.
18453 (spu_addr_space_address_mode): Likewise.
18454 (spu_libgcc_cmp_return_mode): Likewise.
18455 (spu_libgcc_shift_count_mode): Likewise.
18456 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18457 (rl78_addr_space_pointer_mode): Likewise.
18458 (fl78_unwind_word_mode): Likewise.
18459 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18461 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18462 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18463 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18464 (mips_valid_pointer_mode): Likewise.
18465 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18466 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18467 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18469 (ft32_addr_space_address_mode): Likewise.
18470 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18471 scalar_int_mode instead of a machine_mode.
18472 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18474 (m32c_addr_space_address_mode): Likewise.
18475 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18476 (rs6000_eh_return_filter_mode): Likewise.
18477 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18478 (rs6000_eh_return_filter_mode): Likewise.
18479 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18480 (s390_libgcc_shift_count_mode): Likewise.
18481 (s390_unwind_word_mode): Likewise.
18482 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18484 * target.def (mode_rep_extended): Likewise.
18485 (valid_pointer_mode): Likewise.
18486 (addr_space.valid_pointer_mode): Likewise.
18487 (eh_return_filter_mode): Return a scalar_int_mode rather than
18489 (libgcc_cmp_return_mode): Likewise.
18490 (libgcc_shift_count_mode): Likewise.
18491 (unwind_word_mode): Likewise.
18492 (addr_space.pointer_mode): Likewise.
18493 (addr_space.address_mode): Likewise.
18494 * doc/tm.texi: Regenerate.
18495 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18497 (do_jump): Use scalar_int_mode for local variables.
18498 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18499 rather than a machine_mode.
18500 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18501 (scompare_loc_descriptor_wide): Likewise.
18502 (scompare_loc_descriptor_narrow): Likewise.
18503 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18505 * except.c (sjlj_emit_dispatch_table): Likewise.
18506 (expand_builtin_eh_copy_values): Likewise.
18507 * explow.c (convert_memory_address_addr_space_1): Likewise.
18508 Take a scalar_int_mode rather than a machine_mode.
18509 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18510 than a machine_mode.
18511 (memory_address_addr_space): Use scalar_int_mode for local variables.
18512 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18513 rather than a machine_mode.
18514 * expmed.c (mask_rtx): Likewise.
18515 (init_expmed_one_conv): Likewise.
18516 (expand_mult_highpart_adjust): Likewise.
18517 (extract_high_half): Likewise.
18518 (expmed_mult_highpart_optab): Likewise.
18519 (expmed_mult_highpart): Likewise.
18520 (expand_smod_pow2): Likewise.
18521 (expand_sdiv_pow2): Likewise.
18522 (emit_store_flag_int): Likewise.
18523 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18525 (extract_low_bits): Likewise.
18526 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18528 * expr.c (pieces_addr::adjust): Likewise.
18529 (can_store_by_pieces): Likewise.
18530 (store_by_pieces): Likewise.
18531 (clear_by_pieces_1): Likewise.
18532 (expand_expr_addr_expr_1): Likewise.
18533 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18534 (expand_expr_real_1): Likewise.
18535 (try_casesi): Likewise.
18536 * final.c (shorten_branches): Likewise.
18537 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18538 type of "mode" to machine_mode.
18539 * internal-fn.c (expand_arith_overflow_result_store): Take a
18540 scalar_int_mode rather than a machine_mode.
18541 (expand_mul_overflow): Use scalar_int_mode for local variables.
18542 * loop-doloop.c (doloop_modify): Likewise.
18543 (doloop_optimize): Likewise.
18544 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18545 than a machine_mode.
18546 (expand_doubleword_shift_condmove): Likewise.
18547 (expand_doubleword_shift): Likewise.
18548 (expand_doubleword_clz): Likewise.
18549 (expand_doubleword_popcount): Likewise.
18550 (expand_doubleword_parity): Likewise.
18551 (expand_absneg_bit): Use scalar_int_mode for local variables.
18552 (prepare_float_lib_cmp): Likewise.
18553 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18554 rather than a machine_mode.
18555 (convert_memory_address_addr_space): Likewise.
18556 (get_mode_bounds): Likewise.
18557 (get_address_mode): Return a scalar_int_mode rather than a
18559 * rtlanal.c (get_address_mode): Likewise.
18560 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18561 than a machine_mode.
18562 * targhooks.c (default_mode_rep_extended): Likewise.
18563 (default_valid_pointer_mode): Likewise.
18564 (default_addr_space_valid_pointer_mode): Likewise.
18565 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18566 than a machine_mode.
18567 (default_libgcc_cmp_return_mode): Likewise.
18568 (default_libgcc_shift_count_mode): Likewise.
18569 (default_unwind_word_mode): Likewise.
18570 (default_addr_space_pointer_mode): Likewise.
18571 (default_addr_space_address_mode): Likewise.
18572 * targhooks.h (default_eh_return_filter_mode): Likewise.
18573 (default_libgcc_cmp_return_mode): Likewise.
18574 (default_libgcc_shift_count_mode): Likewise.
18575 (default_unwind_word_mode): Likewise.
18576 (default_addr_space_pointer_mode): Likewise.
18577 (default_addr_space_address_mode): Likewise.
18578 (default_mode_rep_extended): Take a scalar_int_mode rather than
18580 (default_valid_pointer_mode): Likewise.
18581 (default_addr_space_valid_pointer_mode): Likewise.
18582 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18584 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18585 rather than a machine_mode.
18586 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18587 for local variables.
18588 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18589 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18590 than a machine_mode.
18592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18593 Alan Hayward <alan.hayward@arm.com>
18594 David Sherwood <david.sherwood@arm.com>
18596 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18597 the mode argument to scalar_int_mode.
18598 (do_jump_by_parts_zero_rtx): Likewise.
18599 (do_jump_by_parts_equality_rtx): Likewise.
18600 (do_jump_by_parts_greater): Take a mode argument.
18601 (do_jump_by_parts_equality): Likewise.
18602 (do_jump_1): Update calls accordingly.
18604 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18605 Alan Hayward <alan.hayward@arm.com>
18606 David Sherwood <david.sherwood@arm.com>
18608 * is-a.h (safe_dyn_cast): New function.
18609 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18610 (jump_table_for_label): Likewise.
18611 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18612 instead of an rtx_insn *.
18613 (shorten_branches): Use dyn_cast instead of LABEL_P and
18614 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18615 rtx_jump_table_data::get_data_mode.
18616 (final_scan_insn): Likewise.
18618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18619 Alan Hayward <alan.hayward@arm.com>
18620 David Sherwood <david.sherwood@arm.com>
18622 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18623 trying to combine a full-register integer set with a subreg
18626 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18627 Alan Hayward <alan.hayward@arm.com>
18628 David Sherwood <david.sherwood@arm.com>
18630 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18631 that is always either address_mode or pointer_mode.
18633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18634 Alan Hayward <alan.hayward@arm.com>
18635 David Sherwood <david.sherwood@arm.com>
18637 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18638 when the two are known to be equal.
18640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18641 Alan Hayward <alan.hayward@arm.com>
18642 David Sherwood <david.sherwood@arm.com>
18644 * simplify-rtx.c (simplify_const_unary_operation): Use
18645 is_a <scalar_int_mode> instead of checking for a nonzero
18646 precision. Forcibly convert op_mode to a scalar_int_mode
18647 in that case. More clearly differentiate the operand and
18648 result modes and use the former when deciding what the value
18649 of a count-bits operation should be. Use is_int_mode instead
18650 of checking for a MODE_INT. Remove redundant check for whether
18651 this mode has a zero precision.
18653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18654 Alan Hayward <alan.hayward@arm.com>
18655 David Sherwood <david.sherwood@arm.com>
18657 * optabs.c (widen_leading): Change the type of the mode argument
18658 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18659 (widen_bswap): Likewise.
18660 (expand_parity): Likewise.
18661 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18662 (expand_ffs): Likewise.
18663 (epand_unop): Check for scalar integer modes before calling the
18666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18667 Alan Hayward <alan.hayward@arm.com>
18668 David Sherwood <david.sherwood@arm.com>
18670 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18672 (expand_expr_real_1): Update call accordingly.
18674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18675 Alan Hayward <alan.hayward@arm.com>
18676 David Sherwood <david.sherwood@arm.com>
18678 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18679 value_mode arguments. Use scalar_int_mode internally.
18680 (store_bit_field_1): Rename the new integer mode from imode
18681 to op0_mode and use it instead of GET_MODE (op0). Update calls
18682 to store_split_bit_field, store_bit_field_using_insv and
18683 store_fixed_bit_field.
18684 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18685 Use scalar_int_mode internally. Use a bit count rather than a mode
18686 when calculating the largest bit size for get_best_mode.
18687 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18688 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18689 Remove assertion that OP0 has a scalar integer mode.
18690 (store_split_bit_field): Add op0_mode and value_mode arguments.
18691 Update calls to extract_fixed_bit_field.
18692 (extract_bit_field_using_extv): Add an op0_mode argument.
18693 Use scalar_int_mode internally.
18694 (extract_bit_field_1): Rename the new integer mode from imode to
18695 op0_mode and use it instead of GET_MODE (op0). Update calls to
18696 extract_split_bit_field, extract_bit_field_using_extv and
18697 extract_fixed_bit_field.
18698 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18699 to extract_split_bit_field and extract_fixed_bit_field_1.
18700 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18701 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18702 on the target mode.
18703 (extract_split_bit_field): Add an op0_mode argument. Update call
18704 to extract_fixed_bit_field.
18706 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18707 Alan Hayward <alan.hayward@arm.com>
18708 David Sherwood <david.sherwood@arm.com>
18710 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18711 * explow.c (hard_function_value): Likewise.
18712 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18713 convert_to_mode call outside the loop.
18714 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18715 for the mode iterator. Require the mode specified by max_pieces
18717 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18719 (copy_blkmode_to_reg): Likewise.
18720 (set_storage_via_setmem): Likewise.
18721 * optabs.c (prepare_cmp_insn): Likewise.
18722 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18723 * stor-layout.c (finish_bitfield_representative): Likewise.
18725 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18726 Alan Hayward <alan.hayward@arm.com>
18727 David Sherwood <david.sherwood@arm.com>
18729 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18730 * expr.c (convert_move): Use them.
18731 (convert_modes): Likewise.
18732 (store_expr_with_bounds): 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 (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18739 parameter for the mode of "x". Remove the "known_x", "known_mode"
18740 and "known_ret" arguments. Change the type of the mode argument
18741 to scalar_int_mode.
18742 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18743 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18744 (reg_num_sign_bit_copies_for_combine): Likewise.
18745 * rtlanal.c (nonzero_bits1): Likewise.
18746 (num_sign_bit_copies1): Likewise.
18747 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18748 (reg_num_sign_bit_copies_general): Likewise.
18749 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18750 (reg_nonzero_bits_general): Likewise.
18752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18753 Alan Hayward <alan.hayward@arm.com>
18754 David Sherwood <david.sherwood@arm.com>
18756 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18757 than in subroutines. Return 1 for non-integer modes.
18758 (cached_num_sign_bit_copies): Change the type of the mode parameter
18759 to scalar_int_mode.
18760 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18761 classes. Handle CONST_INT_P first and then check whether X also
18762 has a scalar integer mode. Check the same thing for inner registers
18763 of a SUBREG and for values that are being extended or truncated.
18765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18766 Alan Hayward <alan.hayward@arm.com>
18767 David Sherwood <david.sherwood@arm.com>
18769 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18770 in subroutines. Return the mode mask for non-integer modes.
18771 (cached_nonzero_bits): Change the type of the mode parameter
18772 to scalar_int_mode.
18773 (nonzero_bits1): Likewise. Remove early exit for other mode
18774 classes. Handle CONST_INT_P first and then check whether X
18775 also has a scalar integer mode.
18777 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18778 Alan Hayward <alan.hayward@arm.com>
18779 David Sherwood <david.sherwood@arm.com>
18781 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18782 Return a scalar_int_mode and assert that the size is greater than
18784 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18785 (op_by_pieces_d::op_by_pieces_d): Likewise.
18786 (op_by_pieces_d::run): Likewise.
18787 (can_store_by_pieces): Likewise.
18789 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18790 Alan Hayward <alan.hayward@arm.com>
18791 David Sherwood <david.sherwood@arm.com>
18793 * combine.c (extract_left_shift): Add a mode argument and update
18795 (make_compound_operation_int): Change the type of the mode parameter
18796 to scalar_int_mode and update the call to extract_left_shift.
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 (simplify_and_const_int): Change the type of the mode
18803 parameter to scalar_int_mode.
18804 (simplify_and_const_int_1): Likewise. Update recursive call.
18806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18807 Alan Hayward <alan.hayward@arm.com>
18808 David Sherwood <david.sherwood@arm.com>
18810 * combine.c (simplify_compare_const): Check that the mode is a
18811 scalar_int_mode (rather than VOIDmode) before testing its
18813 (simplify_comparison): Move COMPARISON_P handling out of the
18814 loop and restrict the latter part of the loop to scalar_int_modes.
18815 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
18816 and when considering SUBREG_REGs. Use is_int_mode instead of
18817 checking GET_MODE_CLASS against MODE_INT.
18819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18820 Alan Hayward <alan.hayward@arm.com>
18821 David Sherwood <david.sherwood@arm.com>
18823 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18824 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18825 shift_unit_mode and for modes involved in scalar shifts.
18827 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18828 Alan Hayward <alan.hayward@arm.com>
18829 David Sherwood <david.sherwood@arm.com>
18831 * combine.c (force_int_to_mode): New function, split out from...
18832 (force_to_mode): ...here. Keep xmode up-to-date and use it
18833 instead of GET_MODE (x).
18835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18836 Alan Hayward <alan.hayward@arm.com>
18837 David Sherwood <david.sherwood@arm.com>
18839 * optabs-query.h (extraction_insn::struct_mode): Change type to
18840 opt_scalar_int_mode and update comment.
18841 (extraction_insn::field_mode): Change type to scalar_int_mode.
18842 (extraction_insn::pos_mode): Likewise.
18843 * combine.c (make_extraction): Update accordingly.
18844 * optabs-query.c (get_traditional_extraction_insn): Likewise.
18845 (get_optab_extraction_insn): Likewise.
18846 * recog.c (simplify_while_replacing): Likewise.
18847 * expmed.c (narrow_bit_field_mem): Change the type of the mode
18848 parameter to opt_scalar_int_mode.
18850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18851 Alan Hayward <alan.hayward@arm.com>
18852 David Sherwood <david.sherwood@arm.com>
18854 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18855 to a scalar_int_mode instead of a machine_mode.
18856 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
18857 (get_best_mode): Return a boolean and use a pointer argument to store
18858 the selected mode. Replace the limit mode parameter with a bit limit.
18859 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
18860 for the values returned by bit_field_mode_iterator::next_mode.
18861 (store_bit_field): Update call to get_best_mode.
18862 (store_fixed_bit_field): Likewise.
18863 (extract_fixed_bit_field): Likewise.
18864 * expr.c (optimize_bitfield_assignment_op): Likewise.
18865 * fold-const.c (optimize_bit_field_compare): Likewise.
18866 (fold_truth_andor_1): Likewise.
18867 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
18868 Update for new type of m_mode.
18869 (get_best_mode): As above.
18871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18872 Alan Hayward <alan.hayward@arm.com>
18873 David Sherwood <david.sherwood@arm.com>
18875 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18876 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18877 (store_bit_field): Check is_a <scalar_int_mode> before calling
18878 strict_volatile_bitfield_p.
18879 (extract_bit_field): Likewise.
18881 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18882 Alan Hayward <alan.hayward@arm.com>
18883 David Sherwood <david.sherwood@arm.com>
18885 * target.def (cstore_mode): Return a scalar_int_mode.
18886 * doc/tm.texi: Regenerate.
18887 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18888 * targhooks.h (default_cstore_mode): Likewise.
18889 * targhooks.c (default_cstore_mode): Likewise, using a forced
18891 * expmed.c (emit_cstore): Expect the target of the cstore to be
18894 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18895 Alan Hayward <alan.hayward@arm.com>
18896 David Sherwood <david.sherwood@arm.com>
18898 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18900 (niter_desc): Likewise mode.
18901 (iv_analyze): Add a mode parameter.
18903 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18904 and change its type to scalar_int_mode.
18905 * loop-iv.c: Update commentary at head of file.
18906 (iv_constant): Pass the mode paraeter before the rtx it describes
18907 and change its type to scalar_int_mode. Remove VOIDmode handling.
18908 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18909 (iv_extend): Likewise.
18910 (shorten_into_mode): Likewise.
18911 (iv_add): Use scalar_int_mode.
18912 (iv_mult): Likewise.
18913 (iv_shift): Likewise.
18914 (canonicalize_iv_subregs): Likewise.
18915 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18916 it describes and change its mode to scalar_int_mode. Also change
18917 the type of the returned inner_mode to scalar_int_mode.
18918 (get_biv_step): Likewise, turning outer_mode from a pointer
18919 into a direct parameter. Update call to get_biv_step_1.
18920 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18921 iv_constant and get_biv_step.
18922 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18923 and change its type to scalar_int_mode. Don't initialise iv->mode
18924 to VOIDmode and remove later checks for its still being VOIDmode.
18925 Update calls to iv_analyze_op and iv_analyze_expr. Check
18926 is_a <scalar_int_mode> when changing the mode under consideration.
18927 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18928 Update call to iv_analyze_expr.
18929 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18930 inner register is not also a scalar_int_mode. Update call to
18932 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18933 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18934 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18935 separate mode class checks. Update calls to iv_analyze. Remove
18936 fix-up of VOIDmodes after iv_analyze_biv.
18937 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18938 don't have a scalar_int_mode. Update call to biv_p.
18940 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18941 Alan Hayward <alan.hayward@arm.com>
18942 David Sherwood <david.sherwood@arm.com>
18944 * cfgexpand.c (convert_debug_memory_address): Use
18945 as_a <scalar_int_mode>.
18946 * combine.c (expand_compound_operation): Likewise.
18947 (make_extraction): Likewise.
18948 (change_zero_ext): Likewise.
18949 (simplify_comparison): Likewise.
18950 * cse.c (cse_insn): Likewise.
18951 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18952 (mem_loc_descriptor): Likewise.
18953 (loc_descriptor): Likewise.
18954 * expmed.c (init_expmed_one_mode): Likewise.
18955 (synth_mult): Likewise.
18956 (emit_store_flag_1): Likewise.
18957 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18958 of a comparison with size.
18959 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18960 (reduce_to_bit_field_precision): Likewise.
18961 * function.c (expand_function_end): Likewise.
18962 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18963 * loop-doloop.c (doloop_modify): Likewise.
18964 * optabs.c (expand_binop): Likewise.
18965 (expand_unop): Likewise.
18966 (expand_copysign_absneg): Likewise.
18967 (prepare_cmp_insn): Likewise.
18968 (maybe_legitimize_operand): Likewise.
18969 * recog.c (const_scalar_int_operand): Likewise.
18970 * rtlanal.c (get_address_mode): Likewise.
18971 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18972 (simplify_cond_clz_ctz): Likewise.
18973 * tree-nested.c (get_nl_goto_field): Likewise.
18974 * tree.c (build_vector_type_for_mode): Likewise.
18975 * var-tracking.c (use_narrower_mode): Likewise.
18977 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18978 Alan Hayward <alan.hayward@arm.com>
18979 David Sherwood <david.sherwood@arm.com>
18981 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18982 * builtins.c (expand_builtin_signbit): Use it.
18983 * cfgexpand.c (expand_debug_expr): Likewise.
18984 * dojump.c (do_jump): Likewise.
18985 (do_compare_and_jump): Likewise.
18986 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18987 * expmed.c (make_tree): Likewise.
18988 * expr.c (expand_expr_real_2): Likewise.
18989 (expand_expr_real_1): Likewise.
18990 (try_casesi): Likewise.
18991 * fold-const-call.c (fold_const_call_ss): Likewise.
18992 * fold-const.c (unextend): Likewise.
18993 (extract_muldiv_1): Likewise.
18994 (fold_single_bit_test): Likewise.
18995 (native_encode_int): Likewise.
18996 (native_encode_string): Likewise.
18997 (native_interpret_int): Likewise.
18998 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18999 * internal-fn.c (expand_addsub_overflow): Likewise.
19000 (expand_neg_overflow): Likewise.
19001 (expand_mul_overflow): Likewise.
19002 (expand_arith_overflow): Likewise.
19003 * match.pd: Likewise.
19004 * stor-layout.c (layout_type): Likewise.
19005 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19006 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19007 * tree-ssanames.c (get_range_info): Likewise.
19008 * tree-switch-conversion.c (array_value_type) Likewise.
19009 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19010 (vect_recog_divmod_pattern): Likewise.
19011 (vect_recog_mixed_size_cond_pattern): Likewise.
19012 * tree-vrp.c (extract_range_basic): Likewise.
19013 (simplify_float_conversion_using_ranges): Likewise.
19014 * tree.c (int_fits_type_p): Likewise.
19015 * ubsan.c (instrument_bool_enum_load): Likewise.
19016 * varasm.c (mergeable_string_section): Likewise.
19017 (narrowing_initializer_constant_valid_p): Likewise.
19018 (output_constant): Likewise.
19020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19021 Alan Hayward <alan.hayward@arm.com>
19022 David Sherwood <david.sherwood@arm.com>
19024 * machmode.h (NARROWEST_INT_MODE): New macro.
19025 * expr.c (alignment_for_piecewise_move): Use it instead of
19026 GET_CLASS_NARROWEST_MODE (MODE_INT).
19027 (push_block): Likewise.
19028 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19030 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19033 Alan Hayward <alan.hayward@arm.com>
19034 David Sherwood <david.sherwood@arm.com>
19036 * postreload.c (move2add_valid_value_p): Change the type of the
19037 mode parameter to scalar_int_mode.
19038 (move2add_use_add2_insn): Add a mode parameter and use it instead
19040 (move2add_use_add3_insn): Likewise.
19041 (reload_cse_move2add): Update accordingly.
19043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19044 Alan Hayward <alan.hayward@arm.com>
19045 David Sherwood <david.sherwood@arm.com>
19047 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19049 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19050 * optabs.c (expand_unop): Likewise.
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 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19057 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19058 (popcount_loc_descriptor): Likewise.
19059 (bswap_loc_descriptor): Likewise.
19060 (rotate_loc_descriptor): Likewise.
19061 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19062 calling the functions above.
19064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19065 Alan Hayward <alan.hayward@arm.com>
19066 David Sherwood <david.sherwood@arm.com>
19068 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19070 (try_combine): Likewise.
19071 (simplify_if_then_else): Likewise.
19072 * cse.c (cse_insn): Likewise.
19073 * dwarf2out.c (mem_loc_descriptor): Likewise.
19074 * emit-rtl.c (gen_lowpart_common): Likewise.
19075 * simplify-rtx.c (simplify_truncation): Likewise.
19076 (simplify_binary_operation_1): Likewise.
19077 (simplify_const_relational_operation): Likewise.
19078 (simplify_ternary_operation): Likewise.
19079 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19082 Alan Hayward <alan.hayward@arm.com>
19083 David Sherwood <david.sherwood@arm.com>
19085 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19086 * reload.c (push_reload): Likewise.
19087 (find_reloads): Likewise.
19089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19090 Alan Hayward <alan.hayward@arm.com>
19091 David Sherwood <david.sherwood@arm.com>
19093 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19094 (make_compound_operation_int): Likewise.
19095 (change_zero_ext): Likewise.
19096 * expr.c (convert_move): Likewise.
19097 (convert_modes): Likewise.
19098 * fwprop.c (forward_propagate_subreg): Likewise.
19099 * loop-iv.c (get_biv_step_1): Likewise.
19100 * optabs.c (widen_operand): Likewise.
19101 * postreload.c (move2add_valid_value_p): Likewise.
19102 * recog.c (simplify_while_replacing): Likewise.
19103 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19104 (simplify_binary_operation_1): Likewise. Remove redundant
19105 mode equality check.
19107 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19108 Alan Hayward <alan.hayward@arm.com>
19109 David Sherwood <david.sherwood@arm.com>
19111 * combine.c (combine_simplify_rtx): Add checks for
19112 is_a <scalar_int_mode>.
19113 (simplify_if_then_else): Likewise.
19114 (make_field_assignment): Likewise.
19115 (simplify_comparison): Likewise.
19116 * ifcvt.c (noce_try_bitop): Likewise.
19117 * loop-invariant.c (canonicalize_address_mult): Likewise.
19118 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19121 Alan Hayward <alan.hayward@arm.com>
19122 David Sherwood <david.sherwood@arm.com>
19124 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19125 is_a <scalar_int_mode> instead of != BLKmode.
19127 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19128 Alan Hayward <alan.hayward@arm.com>
19129 David Sherwood <david.sherwood@arm.com>
19131 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19132 instead of != VOIDmode.
19133 * combine.c (if_then_else_cond): Likewise.
19134 (change_zero_ext): Likewise.
19135 * dwarf2out.c (mem_loc_descriptor): Likewise.
19136 (loc_descriptor): Likewise.
19137 * rtlanal.c (canonicalize_condition): Likewise.
19138 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19141 Alan Hayward <alan.hayward@arm.com>
19142 David Sherwood <david.sherwood@arm.com>
19144 * simplify-rtx.c (simplify_binary_operation_1): Use
19145 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19148 Alan Hayward <alan.hayward@arm.com>
19149 David Sherwood <david.sherwood@arm.com>
19151 * wide-int.h (int_traits<unsigned char>) New class.
19152 (int_traits<unsigned short>) Likewise.
19153 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19154 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19156 * combine.c (set_nonzero_bits_and_sign_copies): Use
19157 is_a <scalar_int_mode>.
19158 (find_split_point): Likewise.
19159 (combine_simplify_rtx): Likewise.
19160 (simplify_logical): Likewise.
19161 (expand_compound_operation): Likewise.
19162 (expand_field_assignment): Likewise.
19163 (make_compound_operation): Likewise.
19164 (extended_count): Likewise.
19165 (change_zero_ext): Likewise.
19166 (simplify_comparison): Likewise.
19167 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19168 (ucompare_loc_descriptor): Likewise.
19169 (minmax_loc_descriptor): Likewise.
19170 (mem_loc_descriptor): Likewise.
19171 (loc_descriptor): Likewise.
19172 * expmed.c (init_expmed_one_mode): Likewise.
19173 * lra-constraints.c (lra_constraint_offset): Likewise.
19174 * optabs.c (prepare_libcall_arg): Likewise.
19175 * postreload.c (move2add_note_store): Likewise.
19176 * reload.c (operands_match_p): Likewise.
19177 * rtl.h (load_extend_op): Likewise.
19178 * rtlhooks.c (gen_lowpart_general): Likewise.
19179 * simplify-rtx.c (simplify_truncation): Likewise.
19180 (simplify_unary_operation_1): Likewise.
19181 (simplify_binary_operation_1): Likewise.
19182 (simplify_const_binary_operation): Likewise.
19183 (simplify_const_relational_operation): Likewise.
19184 (simplify_subreg): Likewise.
19185 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19186 * var-tracking.c (adjust_mems): Likewise.
19187 (prepare_call_arguments): Likewise.
19189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19190 Alan Hayward <alan.hayward@arm.com>
19191 David Sherwood <david.sherwood@arm.com>
19193 * machmode.h (is_int_mode): New fuction.
19194 * combine.c (find_split_point): Use it.
19195 (combine_simplify_rtx): Likewise.
19196 (simplify_if_then_else): Likewise.
19197 (simplify_set): Likewise.
19198 (simplify_shift_const_1): Likewise.
19199 (simplify_comparison): Likewise.
19200 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19201 * cse.c (notreg_cost): Likewise.
19202 (cse_insn): Likewise.
19203 * cselib.c (cselib_lookup_1): Likewise.
19204 * dojump.c (do_jump_1): Likewise.
19205 (do_compare_rtx_and_jump): Likewise.
19206 * dse.c (get_call_args): Likewise.
19207 * dwarf2out.c (rtl_for_decl_init): Likewise.
19208 (native_encode_initializer): Likewise.
19209 * expmed.c (emit_store_flag_1): Likewise.
19210 (emit_store_flag): Likewise.
19211 * expr.c (convert_modes): Likewise.
19212 (store_field): Likewise.
19213 (expand_expr_real_1): Likewise.
19214 * fold-const.c (fold_read_from_constant_string): Likewise.
19215 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19216 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19217 * optabs.c (expand_binop): Likewise.
19218 (expand_unop): Likewise.
19219 (expand_abs_nojump): Likewise.
19220 (expand_one_cmpl_abs_nojump): Likewise.
19221 * simplify-rtx.c (mode_signbit_p): Likewise.
19222 (val_signbit_p): Likewise.
19223 (val_signbit_known_set_p): Likewise.
19224 (val_signbit_known_clear_p): Likewise.
19225 (simplify_relational_operation_1): Likewise.
19226 * tree.c (vector_type_mode): Likewise.
19228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19229 Alan Hayward <alan.hayward@arm.com>
19230 David Sherwood <david.sherwood@arm.com>
19232 * machmode.h (smallest_mode_for_size): Fix formatting.
19233 (smallest_int_mode_for_size): New function.
19234 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19235 instead of smallest_mode_for_size.
19236 * combine.c (make_extraction): Likewise.
19237 * config/arc/arc.c (arc_expand_movmem): Likewise.
19238 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19239 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19240 * config/s390/s390.c (s390_expand_insv): Likewise.
19241 * config/sparc/sparc.c (assign_int_registers): Likewise.
19242 * config/spu/spu.c (spu_function_value): Likewise.
19243 (spu_function_arg): Likewise.
19244 * coverage.c (get_gcov_type): Likewise.
19245 (get_gcov_unsigned_t): Likewise.
19246 * dse.c (find_shift_sequence): Likewise.
19247 * expmed.c (store_bit_field_1): Likewise.
19248 * expr.c (convert_move): Likewise.
19249 (store_field): Likewise.
19250 * internal-fn.c (expand_arith_overflow): Likewise.
19251 * optabs-query.c (get_best_extraction_insn): Likewise.
19252 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19253 * stor-layout.c (layout_type): Likewise.
19254 (initialize_sizetypes): Likewise.
19255 * targhooks.c (default_get_mask_mode): Likewise.
19256 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19259 Alan Hayward <alan.hayward@arm.com>
19260 David Sherwood <david.sherwood@arm.com>
19262 * machmode.h (opt_mode::else_blk): New function.
19263 (int_mode_for_mode): Declare.
19264 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19265 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19267 * cfgexpand.c (expand_debug_expr): Likewise.
19268 * combine.c (gen_lowpart_or_truncate): Likewise.
19269 (gen_lowpart_for_combine): Likewise.
19270 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19271 * config/avr/avr.c (avr_to_int_mode): Likewise.
19272 (avr_out_plus_1): Likewise.
19273 (avr_out_plus): Likewise.
19274 (avr_out_round): Likewise.
19275 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19276 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19277 (s390_expand_vcond): Likewise.
19278 * config/spu/spu.c (spu_split_immediate): Likewise.
19279 (spu_expand_mov): Likewise.
19280 * dse.c (get_stored_val): Likewise.
19281 * expmed.c (store_bit_field_1): Likewise.
19282 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19284 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19285 (extract_low_bits): Likewise.
19286 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19287 handling rather than repeating the check.
19288 (emit_group_store): Likewise.
19289 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19290 * optabs.c (expand_absneg_bit): Likewise.
19291 (expand_copysign_absneg): Likewise.
19292 (expand_copysign_bit): Likewise.
19293 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19294 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19295 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19296 * var-tracking.c (prepare_call_arguments): Likewise.
19297 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19298 int_mode_for_mode instead of mode_for_size.
19299 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19302 Alan Hayward <alan.hayward@arm.com>
19303 David Sherwood <david.sherwood@arm.com>
19305 * machmode.h (int_mode_for_size): New function.
19306 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19307 instead of mode_for_size.
19308 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19310 * combine.c (expand_field_assignment): Use int_mode_for_size
19311 instead of mode_for_size.
19312 (make_extraction): Likewise.
19313 (simplify_shift_const_1): Likewise.
19314 (simplify_comparison): Likewise.
19315 * dojump.c (do_jump): Likewise.
19316 * dwarf2out.c (mem_loc_descriptor): Likewise.
19317 * emit-rtl.c (init_derived_machine_modes): Likewise.
19318 * expmed.c (flip_storage_order): Likewise.
19319 (convert_extracted_bit_field): Likewise.
19320 * expr.c (copy_blkmode_from_reg): Likewise.
19321 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19322 * internal-fn.c (expand_mul_overflow): Likewise.
19323 * lower-subreg.c (simple_move): Likewise.
19324 * optabs-libfuncs.c (init_optabs): Likewise.
19325 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19326 * tree.c (vector_type_mode): Likewise.
19327 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19328 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19329 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19330 * tree-vect-stmts.c (vectorizable_load): Likewise.
19331 (vectorizable_store): Likewise.
19333 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19334 Alan Hayward <alan.hayward@arm.com>
19335 David Sherwood <david.sherwood@arm.com>
19337 * coretypes.h (pod_mode): New type.
19338 (scalar_int_mode_pod): New typedef.
19339 * machmode.h (pod_mode): New class.
19340 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19341 * genmodes.c (emit_mode_int_n): Update accordingly.
19342 * lower-subreg.h (target_lower_subreg): Change type to
19343 scalar_int_mode_pod.
19344 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19345 scalar_int_mode_pod.
19347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19348 Alan Hayward <alan.hayward@arm.com>
19349 David Sherwood <david.sherwood@arm.com>
19351 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19352 machine_mode to scalar_int_mode.
19353 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19354 (rs6000_option_override_internal): Remove cast to int.
19355 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19356 machine_mode to scalar_int_mode.
19357 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19358 (rs6000_option_override_internal): Remove cast to int.
19359 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19360 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19362 * config/s390/s390.c (s390_expand_builtin): Likewise.
19363 * coretypes.h (scalar_int_mode): New type.
19364 (opt_scalar_int_mode): New typedef.
19365 * machmode.h (scalar_int_mode): New class.
19366 (scalar_int_mode::includes_p): New function.
19367 (byte_mode): Change type to scalar_int_mode.
19368 (word_mode): Likewise.
19369 (ptr_mode): Likewise.
19370 * emit-rtl.c (byte_mode): Likewise.
19371 (word_mode): Likewise.
19372 (ptr_mode): Likewise.
19373 (init_derived_machine_modes): Update accordingly.
19374 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19375 and MODE_PARTIAL_INT.
19376 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19377 opt_scalar_int_mode.
19379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19380 Alan Hayward <alan.hayward@arm.com>
19381 David Sherwood <david.sherwood@arm.com>
19383 * target.def (libgcc_floating_mode_supported_p): Take a
19385 * doc/tm.texi: Regenerate.
19386 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19388 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19389 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19393 Alan Hayward <alan.hayward@arm.com>
19394 David Sherwood <david.sherwood@arm.com>
19396 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19397 * doc/tm.texi: Regenerate.
19398 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19399 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19400 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19401 * targhooks.h (default_floatn_mode): Likewise.
19402 * targhooks.c (default_floatn_mode): Likewise.
19403 * tree.c (build_common_tree_nodes): Update accordingly.
19405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19406 Alan Hayward <alan.hayward@arm.com>
19407 David Sherwood <david.sherwood@arm.com>
19409 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19410 (mode_iterator::iterate_p): Likewise.
19411 (mode_iterator::get_wider): Likewise.
19412 * expr.c (init_expr_target): Use opt_scalar_float_mode.
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 * coretypes.h (opt_scalar_float_mode): New typedef.
19419 * machmode.h (float_mode_for_size): New function.
19420 * emit-rtl.c (double_mode): Delete.
19421 (init_emit_once): Use float_mode_for_size.
19422 * stor-layout.c (layout_type): Likewise.
19423 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19426 Alan Hayward <alan.hayward@arm.com>
19427 David Sherwood <david.sherwood@arm.com>
19429 * output.h (assemble_real): Take a scalar_float_mode.
19430 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19431 * config/arm/arm.md (consttable_4): Likewise.
19432 (consttable_8): Likewise.
19433 (consttable_16): Likewise.
19434 * config/mips/mips.md (consttable_float): Likewise.
19435 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19436 * varasm.c (assemble_real): Take a scalar_float_mode.
19437 (output_constant_pool_2): Update accordingly.
19438 (output_constant): Likewise.
19440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19441 Alan Hayward <alan.hayward@arm.com>
19442 David Sherwood <david.sherwood@arm.com>
19444 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19445 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19446 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19447 (native_encode_real): Likewise.
19448 (native_interpret_real): Likewise.
19449 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19450 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19453 Alan Hayward <alan.hayward@arm.com>
19454 David Sherwood <david.sherwood@arm.com>
19456 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19457 <scalar_float_mode>. Simplify.
19458 (gen_extend_conv_libfunc): Likewise.
19460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19461 Alan Hayward <alan.hayward@arm.com>
19462 David Sherwood <david.sherwood@arm.com>
19464 * coretypes.h (scalar_float_mode): New type.
19465 * machmode.h (mode_traits::from_int): Use machine_mode if
19466 USE_ENUM_MODES is defined.
19467 (is_a): New function.
19469 (dyn_cast): Likewise.
19470 (scalar_float_mode): New class.
19471 (scalar_float_mode::includes_p): New function.
19472 (is_float_mode): Likewise.
19473 * gdbhooks.py (MachineModePrinter): New class.
19474 (build_pretty_printer): Use it for scalar_float_mode.
19475 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19476 (format_helper::format_helper): Turn into a template.
19477 * genmodes.c (get_mode_class): New function.
19478 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19479 or machine_mode if none.
19480 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19481 as_a <scalar_float_mode>.
19482 * dwarf2out.c (mem_loc_descriptor): Likewise.
19483 (insert_float): Likewise.
19484 (add_const_value_attribute): Likewise.
19485 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19486 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19487 (expand_unop): Update accordingly.
19488 (expand_abs_nojump): Likewise.
19489 (expand_copysign_absneg): Take a scalar_float_mode.
19490 (expand_copysign_bit): Likewise.
19491 (expand_copysign): Update accordingly.
19493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19494 Alan Hayward <alan.hayward@arm.com>
19495 David Sherwood <david.sherwood@arm.com>
19497 * coretypes.h (opt_mode): New class.
19498 * machmode.h (opt_mode): Likewise.
19499 (opt_mode::else_void): New function.
19500 (opt_mode::require): Likewise.
19501 (opt_mode::exists): Likewise.
19502 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19503 (GET_MODE_2XWIDER_MODE): Likewise.
19504 (mode_iterator::get_wider): Update accordingly.
19505 (mode_iterator::get_2xwider): Likewise.
19506 (mode_iterator::get_known_wider): Likewise, turning into a template.
19507 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19508 forcing a wider mode to exist.
19509 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19510 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19511 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19512 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19513 * lower-subreg.c (init_lower_subreg): Likewise.
19514 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19515 on the final iteration.
19516 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19517 a wider mode exists before asking for a move pattern.
19518 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19519 forcing a wider mode to exist.
19520 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19521 returning false if no such mode exists.
19522 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19523 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19524 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19525 Avoid checking for a MODE_INT if we already know the mode is not a
19527 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19528 forcing a wider mode to exist.
19529 (expmed_mult_highpart_optab): Likewise.
19530 (expmed_mult_highpart): Likewise.
19531 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19533 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19534 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19535 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19536 * internal-fn.c (expand_mul_overflow): Update use of
19537 GET_MODE_2XWIDER_MODE.
19538 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19539 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19540 GET_MODE_WIDER_MODE.
19541 (convert_plusminus_to_widen): Likewise.
19542 * tree-switch-conversion.c (array_value_type): Likewise.
19543 * var-tracking.c (emit_note_insn_var_location): Likewise.
19544 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19545 Return false inside rather than outside the loop if no wider mode
19547 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19548 and GET_MODE_2XWIDER_MODE
19549 (can_compare_p): Use else_void.
19550 * gdbhooks.py (OptMachineModePrinter): New class.
19551 (build_pretty_printer): Use it for opt_mode.
19553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19554 Alan Hayward <alan.hayward@arm.com>
19555 David Sherwood <david.sherwood@arm.com>
19557 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19558 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19561 Alan Hayward <alan.hayward@arm.com>
19562 David Sherwood <david.sherwood@arm.com>
19564 * machmode.h (mode_traits): New structure.
19565 (get_narrowest_mode): New function.
19566 (mode_iterator::start): Likewise.
19567 (mode_iterator::iterate_p): Likewise.
19568 (mode_iterator::get_wider): Likewise.
19569 (mode_iterator::get_known_wider): Likewise.
19570 (mode_iterator::get_2xwider): Likewise.
19571 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19572 (FOR_EACH_MODE): Likewise.
19573 (FOR_EACH_MODE_FROM): Likewise.
19574 (FOR_EACH_MODE_UNTIL): Likewise.
19575 (FOR_EACH_WIDER_MODE): Likewise.
19576 (FOR_EACH_2XWIDER_MODE): Likewise.
19577 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19578 * combine.c (simplify_comparison): Likewise
19579 * config/i386/i386.c (type_natural_mode): Likewise.
19580 * cse.c (cse_insn): Likewise.
19581 * dse.c (find_shift_sequence): Likewise.
19582 * emit-rtl.c (init_derived_machine_modes): Likewise.
19583 (init_emit_once): Likewise.
19584 * explow.c (hard_function_value): Likewise.
19585 * expmed.c (extract_fixed_bit_field_1): Likewise.
19586 (extract_bit_field_1): Likewise.
19587 (expand_divmod): Likewise.
19588 (emit_store_flag_1): Likewise.
19589 * expr.c (init_expr_target): Likewise.
19590 (convert_move): Likewise.
19591 (alignment_for_piecewise_move): Likewise.
19592 (widest_int_mode_for_size): Likewise.
19593 (emit_block_move_via_movmem): Likewise.
19594 (copy_blkmode_to_reg): Likewise.
19595 (set_storage_via_setmem): Likewise.
19596 (compress_float_constant): Likewise.
19597 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19598 * optabs-query.c (get_best_extraction_insn): Likewise.
19599 * optabs.c (expand_binop): Likewise.
19600 (expand_twoval_unop): Likewise.
19601 (expand_twoval_binop): Likewise.
19602 (widen_leading): Likewise.
19603 (widen_bswap): Likewise.
19604 (expand_parity): Likewise.
19605 (expand_unop): Likewise.
19606 (prepare_cmp_insn): Likewise.
19607 (prepare_float_lib_cmp): Likewise.
19608 (expand_float): Likewise.
19609 (expand_fix): Likewise.
19610 (expand_sfix_optab): Likewise.
19611 * postreload.c (move2add_use_add2_insn): Likewise.
19612 * reg-stack.c (reg_to_stack): Likewise.
19613 * reginfo.c (choose_hard_reg_mode): Likewise.
19614 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19615 * stor-layout.c (mode_for_size): Likewise.
19616 (smallest_mode_for_size): Likewise.
19617 (mode_for_vector): Likewise.
19618 (finish_bitfield_representative): Likewise.
19619 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19620 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19621 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19622 * var-tracking.c (prepare_call_arguments): Likewise.
19624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19625 Alan Hayward <alan.hayward@arm.com>
19626 David Sherwood <david.sherwood@arm.com>
19628 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19629 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19630 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19631 * machmode.h (mode_size): Move earlier in file.
19632 (mode_precision): Likewise.
19633 (mode_inner): Likewise.
19634 (mode_nunits): Likewise.
19635 (mode_unit_size): Likewise.
19636 (unit_unit_precision): Likewise.
19637 (mode_wider): Likewise.
19638 (mode_2xwider): Likewise.
19639 (machine_mode): New class.
19640 (mode_to_bytes): New function.
19641 (mode_to_bits): Likewise.
19642 (mode_to_precision): Likewise.
19643 (mode_to_inner): Likewise.
19644 (mode_to_unit_size): Likewise.
19645 (mode_to_unit_precision): Likewise.
19646 (mode_to_nunits): Likewise.
19647 (GET_MODE_SIZE): Use mode_to_bytes.
19648 (GET_MODE_BITSIZE): Use mode_to_bits.
19649 (GET_MODE_PRECISION): Use mode_to_precision.
19650 (GET_MODE_INNER): Use mode_to_inner.
19651 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19652 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19653 (GET_MODE_NUNITS): Use mode_to_nunits.
19654 * system.h (ALWAYS_INLINE): New macro.
19655 * config/powerpcspe/powerpcspe-c.c
19656 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19657 int for arg1_mode and arg2_mode.
19659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19660 Alan Hayward <alan.hayward@arm.com>
19661 David Sherwood <david.sherwood@arm.com>
19663 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19664 Prefix mode names with E_ in case statements.
19665 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19666 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19667 (aarch64_split_simd_move): Likewise.
19668 (aarch64_gen_storewb_pair): Likewise.
19669 (aarch64_gen_loadwb_pair): Likewise.
19670 (aarch64_gen_store_pair): Likewise.
19671 (aarch64_gen_load_pair): Likewise.
19672 (aarch64_get_condition_code_1): Likewise.
19673 (aarch64_constant_pool_reload_icode): Likewise.
19674 (get_rsqrte_type): Likewise.
19675 (get_rsqrts_type): Likewise.
19676 (get_recpe_type): Likewise.
19677 (get_recps_type): Likewise.
19678 (aarch64_gimplify_va_arg_expr): Likewise.
19679 (aarch64_simd_container_mode): Likewise.
19680 (aarch64_emit_load_exclusive): Likewise.
19681 (aarch64_emit_store_exclusive): Likewise.
19682 (aarch64_expand_compare_and_swap): Likewise.
19683 (aarch64_gen_atomic_cas): Likewise.
19684 (aarch64_emit_bic): Likewise.
19685 (aarch64_emit_atomic_swap): Likewise.
19686 (aarch64_emit_atomic_load_op): Likewise.
19687 (aarch64_evpc_trn): Likewise.
19688 (aarch64_evpc_uzp): Likewise.
19689 (aarch64_evpc_zip): Likewise.
19690 (aarch64_evpc_ext): Likewise.
19691 (aarch64_evpc_rev): Likewise.
19692 (aarch64_evpc_dup): Likewise.
19693 (aarch64_gen_ccmp_first): Likewise.
19694 (aarch64_gen_ccmp_next): Likewise.
19695 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19696 (alpha_emit_xfloating_libcall): Likewise.
19697 (emit_insxl): Likewise.
19698 (alpha_arg_type): Likewise.
19699 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19700 (arc_preferred_simd_mode): Likewise.
19701 (arc_secondary_reload): Likewise.
19702 (get_arc_condition_code): Likewise.
19703 (arc_print_operand): Likewise.
19704 (arc_legitimate_constant_p): Likewise.
19705 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19706 * config/arc/arc.md (casesi_load): Likewise.
19707 (casesi_compact_jump): Likewise.
19708 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19709 (cc_use_register): Likewise.
19710 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19711 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19712 (arm_init_iwmmxt_builtins): Likewise.
19713 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19714 (neon_expand_vector_init): Likewise.
19715 (arm_attr_length_move_neon): Likewise.
19716 (maybe_get_arm_condition_code): Likewise.
19717 (arm_emit_vector_const): Likewise.
19718 (arm_preferred_simd_mode): Likewise.
19719 (arm_output_iwmmxt_tinsr): Likewise.
19720 (thumb1_output_casesi): Likewise.
19721 (thumb2_output_casesi): Likewise.
19722 (arm_emit_load_exclusive): Likewise.
19723 (arm_emit_store_exclusive): Likewise.
19724 (arm_expand_compare_and_swap): Likewise.
19725 (arm_evpc_neon_vuzp): Likewise.
19726 (arm_evpc_neon_vzip): Likewise.
19727 (arm_evpc_neon_vrev): Likewise.
19728 (arm_evpc_neon_vtrn): Likewise.
19729 (arm_evpc_neon_vext): Likewise.
19730 (arm_validize_comparison): Likewise.
19731 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19732 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19733 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19734 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19735 (c6x_preferred_simd_mode): Likewise.
19736 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19737 (epiphany_rtx_costs): Likewise.
19738 * config/epiphany/predicates.md (proper_comparison_operator):
19740 * config/frv/frv.c (condexec_memory_operand): Likewise.
19741 (frv_emit_move): Likewise.
19742 (output_move_single): Likewise.
19743 (output_condmove_single): Likewise.
19744 (frv_hard_regno_mode_ok): Likewise.
19745 (frv_matching_accg_mode): Likewise.
19746 * config/h8300/h8300.c (split_adds_subs): Likewise.
19747 (h8300_rtx_costs): Likewise.
19748 (h8300_print_operand): Likewise.
19749 (compute_mov_length): Likewise.
19750 (output_logical_op): Likewise.
19751 (compute_logical_op_length): Likewise.
19752 (compute_logical_op_cc): Likewise.
19753 (h8300_shift_needs_scratch_p): Likewise.
19754 (output_a_shift): Likewise.
19755 (compute_a_shift_length): Likewise.
19756 (compute_a_shift_cc): Likewise.
19757 (expand_a_rotate): Likewise.
19758 (output_a_rotate): Likewise.
19759 * config/i386/i386.c (classify_argument): Likewise.
19760 (function_arg_advance_32): Likewise.
19761 (function_arg_32): Likewise.
19762 (function_arg_64): Likewise.
19763 (function_value_64): Likewise.
19764 (ix86_gimplify_va_arg): Likewise.
19765 (ix86_legitimate_constant_p): Likewise.
19766 (put_condition_code): Likewise.
19767 (split_double_mode): Likewise.
19768 (ix86_avx256_split_vector_move_misalign): Likewise.
19769 (ix86_expand_vector_logical_operator): Likewise.
19770 (ix86_split_idivmod): Likewise.
19771 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19772 (ix86_build_const_vector): Likewise.
19773 (ix86_build_signbit_mask): Likewise.
19774 (ix86_match_ccmode): Likewise.
19775 (ix86_cc_modes_compatible): Likewise.
19776 (ix86_expand_branch): Likewise.
19777 (ix86_expand_sse_cmp): Likewise.
19778 (ix86_expand_sse_movcc): Likewise.
19779 (ix86_expand_int_sse_cmp): Likewise.
19780 (ix86_expand_vec_perm_vpermi2): Likewise.
19781 (ix86_expand_vec_perm): Likewise.
19782 (ix86_expand_sse_unpack): Likewise.
19783 (ix86_expand_int_addcc): Likewise.
19784 (ix86_split_to_parts): Likewise.
19785 (ix86_vectorize_builtin_gather): Likewise.
19786 (ix86_vectorize_builtin_scatter): Likewise.
19787 (avx_vpermilp_parallel): Likewise.
19788 (inline_memory_move_cost): Likewise.
19789 (ix86_tieable_integer_mode_p): Likewise.
19790 (x86_maybe_negate_const_int): Likewise.
19791 (ix86_expand_vector_init_duplicate): Likewise.
19792 (ix86_expand_vector_init_one_nonzero): Likewise.
19793 (ix86_expand_vector_init_one_var): Likewise.
19794 (ix86_expand_vector_init_concat): Likewise.
19795 (ix86_expand_vector_init_interleave): Likewise.
19796 (ix86_expand_vector_init_general): Likewise.
19797 (ix86_expand_vector_set): Likewise.
19798 (ix86_expand_vector_extract): Likewise.
19799 (emit_reduc_half): Likewise.
19800 (ix86_emit_i387_round): Likewise.
19801 (ix86_mangle_type): Likewise.
19802 (ix86_expand_round_sse4): Likewise.
19803 (expand_vec_perm_blend): Likewise.
19804 (canonicalize_vector_int_perm): Likewise.
19805 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
19806 (expand_vec_perm_1): Likewise.
19807 (expand_vec_perm_interleave3): Likewise.
19808 (expand_vec_perm_even_odd_pack): Likewise.
19809 (expand_vec_perm_even_odd_1): Likewise.
19810 (expand_vec_perm_broadcast_1): Likewise.
19811 (ix86_vectorize_vec_perm_const_ok): Likewise.
19812 (ix86_expand_vecop_qihi): Likewise.
19813 (ix86_expand_mul_widen_hilo): Likewise.
19814 (ix86_expand_sse2_abs): Likewise.
19815 (ix86_expand_pextr): Likewise.
19816 (ix86_expand_pinsr): Likewise.
19817 (ix86_preferred_simd_mode): Likewise.
19818 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19819 * config/i386/sse.md (*andnot<mode>3): Likewise.
19820 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19821 (*<code><mode>3): Likewise.
19822 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19823 (ia64_expand_atomic_op): Likewise.
19824 (ia64_arg_type): Likewise.
19825 (ia64_mode_to_int): Likewise.
19826 (ia64_scalar_mode_supported_p): Likewise.
19827 (ia64_vector_mode_supported_p): Likewise.
19828 (expand_vec_perm_broadcast): Likewise.
19829 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19830 (iq2000_function_arg_advance): Likewise.
19831 (iq2000_function_arg): Likewise.
19832 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19833 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19834 (m68k_libcall_value): Likewise.
19835 (m68k_function_value): Likewise.
19836 (sched_attr_op_type): Likewise.
19837 * config/mcore/mcore.c (mcore_output_move): Likewise.
19838 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19840 (microblaze_function_arg): Likewise.
19841 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19842 (mips_print_operand): Likewise.
19843 (mips_mode_ok_for_mov_fmt_p): Likewise.
19844 (mips_vector_mode_supported_p): Likewise.
19845 (mips_preferred_simd_mode): Likewise.
19846 (mips_expand_vpc_loongson_even_odd): Likewise.
19847 (mips_expand_vec_unpack): Likewise.
19848 (mips_expand_vi_broadcast): Likewise.
19849 (mips_expand_vector_init): Likewise.
19850 (mips_expand_vec_reduc): Likewise.
19851 (mips_expand_msa_cmp): Likewise.
19852 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19853 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19854 (cc_flags_for_mode): Likewise.
19855 * config/msp430/msp430.c (msp430_print_operand): Likewise.
19856 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
19857 (nds32_output_casesi_pc_relative): Likewise.
19858 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19859 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
19860 (nvptx_gen_unpack): Likewise.
19861 (nvptx_gen_pack): Likewise.
19862 (nvptx_gen_shuffle): Likewise.
19863 (nvptx_gen_wcast): Likewise.
19864 (nvptx_preferred_simd_mode): Likewise.
19865 * config/pa/pa.c (pa_secondary_reload): Likewise.
19866 * config/pa/predicates.md (base14_operand): Likewise.
19867 * config/powerpcspe/powerpcspe-c.c
19868 (altivec_resolve_overloaded_builtin): Likewise.
19869 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
19871 (rs6000_preferred_simd_mode): Likewise.
19872 (output_vec_const_move): Likewise.
19873 (rs6000_expand_vector_extract): Likewise.
19874 (rs6000_split_vec_extract_var): Likewise.
19875 (reg_offset_addressing_ok_p): Likewise.
19876 (rs6000_legitimate_offset_address_p): Likewise.
19877 (rs6000_legitimize_address): Likewise.
19878 (rs6000_emit_set_const): Likewise.
19879 (rs6000_const_vec): Likewise.
19880 (rs6000_emit_move): Likewise.
19881 (spe_build_register_parallel): Likewise.
19882 (rs6000_darwin64_record_arg_recurse): Likewise.
19883 (swap_selector_for_mode): Likewise.
19884 (spe_init_builtins): Likewise.
19885 (paired_init_builtins): Likewise.
19886 (altivec_init_builtins): Likewise.
19887 (do_load_for_compare): Likewise.
19888 (rs6000_generate_compare): Likewise.
19889 (rs6000_expand_float128_convert): Likewise.
19890 (emit_load_locked): Likewise.
19891 (emit_store_conditional): Likewise.
19892 (rs6000_output_function_epilogue): Likewise.
19893 (rs6000_handle_altivec_attribute): Likewise.
19894 (rs6000_function_value): Likewise.
19895 (emit_fusion_gpr_load): Likewise.
19896 (emit_fusion_p9_load): Likewise.
19897 (emit_fusion_p9_store): Likewise.
19898 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19899 (fusion_gpr_mem_load): Likewise.
19900 (fusion_addis_mem_combo_load): Likewise.
19901 (fusion_addis_mem_combo_store): Likewise.
19902 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19903 (fusion_gpr_mem_load): Likewise.
19904 (fusion_addis_mem_combo_load): Likewise.
19905 (fusion_addis_mem_combo_store): Likewise.
19906 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19908 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19909 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19910 (rs6000_preferred_simd_mode): Likewise.
19911 (output_vec_const_move): Likewise.
19912 (rs6000_expand_vector_extract): Likewise.
19913 (rs6000_split_vec_extract_var): Likewise.
19914 (reg_offset_addressing_ok_p): Likewise.
19915 (rs6000_legitimate_offset_address_p): Likewise.
19916 (rs6000_legitimize_address): Likewise.
19917 (rs6000_emit_set_const): Likewise.
19918 (rs6000_const_vec): Likewise.
19919 (rs6000_emit_move): Likewise.
19920 (rs6000_darwin64_record_arg_recurse): Likewise.
19921 (swap_selector_for_mode): Likewise.
19922 (paired_init_builtins): Likewise.
19923 (altivec_init_builtins): Likewise.
19924 (rs6000_expand_float128_convert): Likewise.
19925 (emit_load_locked): Likewise.
19926 (emit_store_conditional): Likewise.
19927 (rs6000_output_function_epilogue): Likewise.
19928 (rs6000_handle_altivec_attribute): Likewise.
19929 (rs6000_function_value): Likewise.
19930 (emit_fusion_gpr_load): Likewise.
19931 (emit_fusion_p9_load): Likewise.
19932 (emit_fusion_p9_store): Likewise.
19933 * config/rx/rx.c (rx_gen_move_template): Likewise.
19934 (flags_from_mode): Likewise.
19935 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19936 (s390_slb_comparison): Likewise.
19937 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19938 (s390_vector_mode_supported_p): Likewise.
19939 (s390_cc_modes_compatible): Likewise.
19940 (s390_match_ccmode_set): Likewise.
19941 (s390_canonicalize_comparison): Likewise.
19942 (s390_emit_compare_and_swap): Likewise.
19943 (s390_branch_condition_mask): Likewise.
19944 (s390_rtx_costs): Likewise.
19945 (s390_secondary_reload): Likewise.
19946 (__SECONDARY_RELOAD_CASE): Likewise.
19947 (s390_expand_cs): Likewise.
19948 (s390_preferred_simd_mode): Likewise.
19949 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19950 * config/sh/sh.c (sh_print_operand): Likewise.
19951 (dump_table): Likewise.
19952 (sh_secondary_reload): Likewise.
19953 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19954 * config/sh/sh.md (casesi_worker_1): Likewise.
19955 (casesi_worker_2): Likewise.
19956 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19957 (fcc_comparison_operator): Likewise.
19958 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19959 (emit_soft_tfmode_cvt): Likewise.
19960 (sparc_preferred_simd_mode): Likewise.
19961 (output_cbranch): Likewise.
19962 (sparc_print_operand): Likewise.
19963 (sparc_expand_vec_perm_bmask): Likewise.
19964 (vector_init_bshuffle): Likewise.
19965 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19966 (spu_vector_mode_supported_p): Likewise.
19967 (spu_expand_insv): Likewise.
19968 (spu_emit_branch_or_set): Likewise.
19969 (spu_handle_vector_attribute): Likewise.
19970 (spu_builtin_splats): Likewise.
19971 (spu_builtin_extract): Likewise.
19972 (spu_builtin_promote): Likewise.
19973 (spu_expand_sign_extend): Likewise.
19974 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19975 (tilegx_simd_int): Likewise.
19976 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19977 (tilepro_simd_int): Likewise.
19978 * config/v850/v850.c (const_double_split): Likewise.
19979 (v850_print_operand): Likewise.
19980 (ep_memory_offset): Likewise.
19981 * config/vax/vax.c (vax_rtx_costs): Likewise.
19982 (vax_output_int_move): Likewise.
19983 (vax_output_int_add): Likewise.
19984 (vax_output_int_subtract): Likewise.
19985 * config/visium/predicates.md (visium_branch_operator): Likewise.
19986 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19987 (visium_print_operand_address): Likewise.
19988 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19989 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19990 (xtensa_expand_conditional_branch): Likewise.
19991 (xtensa_copy_incoming_a7): Likewise.
19992 (xtensa_output_literal): Likewise.
19993 * dfp.c (decimal_real_maxval): Likewise.
19994 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19996 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19997 Alan Hayward <alan.hayward@arm.com>
19998 David Sherwood <david.sherwood@arm.com>
20000 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20001 (mode_nunits_inline): Likewise.
20002 (mode_inner_inline): Likewise.
20003 (mode_unit_size_inline): Likewise.
20004 (mode_unit_precision_inline): Likewise.
20005 (emit_insn_modes_h): Likewise. Also emit a #define of the
20007 (emit_mode_wider): Add an E_ prefix to mode names.
20008 (emit_mode_complex): Likewise.
20009 (emit_mode_inner): Likewise.
20010 (emit_mode_adjustments): Likewise.
20011 (emit_mode_int_n): Likewise.
20012 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20013 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20014 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20015 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20016 (CRC32_BUILTIN, ENTRY): Likewise.
20017 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20018 (aarch64_pop_regs): Likewise.
20019 (aarch64_process_components): Likewise.
20020 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20021 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20022 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20023 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20024 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20025 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20027 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20028 (pcmpistr): Likewise.
20029 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20030 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20031 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20033 * config/rl78/rl78.c (mduc_regs): Likewise.
20034 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20035 (htm_expand_builtin): Likewise.
20036 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20037 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20038 * config/xtensa/xtensa.c (print_operand): Likewise.
20039 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20040 (NUM_MODE_VECTOR_INT): Likewise.
20041 * genoutput.c (null_operand): Likewise.
20042 (output_operand_data): Likewise.
20043 * genrecog.c (print_parameter_value): Likewise.
20044 * lra.c (debug_operand_data): Likewise.
20046 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20048 * dwarf2out.c (scompare_loc_descriptor_wide)
20049 (scompare_loc_descriptor_narrow): New functions, split out from...
20050 (scompare_loc_descriptor): ...here.
20051 * expmed.c (emit_store_flag_int): New function, split out from...
20052 (emit_store_flag): ...here.
20054 2017-08-30 Richard Biener <rguenther@suse.de>
20056 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20057 (dwarf2out_early_finish): Move setting of AT_pubnames from
20058 early debug output to early finish.
20060 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20062 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20063 and -mdata-region to the assembler.
20065 2017-08-30 Richard Biener <rguenther@suse.de>
20067 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20069 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20070 (dwarf2out_early_global_decl): Only generate a DIE for the
20071 abstract origin if it doesn't already exist or is a declaration DIE.
20072 (resolve_addr): Do not add the linkage name twice when
20073 generating a stub DIE for the DW_TAG_GNU_call_site target.
20075 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20077 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20078 Use machine_mode rather than int for arg1_mode.
20080 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20083 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20084 that the second argument of the built-in functions to unpack
20085 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20086 switch statement instead a lot of if statements.
20087 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20088 Allow 64-bit values to be in Altivec registers as well as
20089 traditional floating point registers.
20090 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20092 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20094 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20095 MAX_REGS_PER_ADDRESS == 1.
20097 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20099 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20100 * config/i386/i386.c (ix86_option_override_internal): Simplify
20101 setting of opts->x_flag_entry.
20103 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20104 Jakub Jelinek <jakub@redhat.com>
20105 Richard Biener <rguenther@suse.de>
20107 PR tree-optimization/81503
20108 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20109 folded constant fits in the target type; reorder tests for clarity.
20111 2017-08-29 Martin Liska <mliska@suse.cz>
20113 * passes.def: Include pass_lower_switch.
20114 * stmt.c (dump_case_nodes): Remove and move to
20115 tree-switch-conversion.
20116 (case_values_threshold): Likewise.
20117 (expand_switch_as_decision_tree_p): Likewise.
20118 (emit_case_decision_tree): Likewise.
20119 (expand_case): Likewise.
20120 (balance_case_nodes): Likewise.
20121 (node_has_low_bound): Likewise.
20122 (node_has_high_bound): Likewise.
20123 (node_is_bounded): Likewise.
20124 (emit_case_nodes): Likewise.
20125 (struct simple_case_node): New struct.
20126 (add_case_node): Remove.
20127 (emit_case_dispatch_table): Use vector instead of case_list.
20128 (reset_out_edges_aux): Remove.
20129 (compute_cases_per_edge): Likewise.
20130 (expand_case): Build list of simple_case_node.
20131 (expand_sjlj_dispatch_table): Use it.
20132 * tree-switch-conversion.c (struct case_node): Moved from
20133 stmt.c and adjusted.
20134 (emit_case_nodes): Likewise.
20135 (node_has_low_bound): Likewise.
20136 (node_has_high_bound): Likewise.
20137 (node_is_bounded): Likewise.
20138 (case_values_threshold): Likewise.
20139 (reset_out_edges_aux): Likewise.
20140 (compute_cases_per_edge): Likewise.
20141 (add_case_node): Likewise.
20142 (dump_case_nodes): Likewise.
20143 (balance_case_nodes): Likewise.
20144 (expand_switch_as_decision_tree_p): Likewise.
20145 (emit_jump): Likewise.
20146 (emit_case_decision_tree): Likewise.
20147 (try_switch_expansion): Likewise.
20148 (do_jump_if_equal): Likewise.
20149 (emit_cmp_and_jump_insns): Likewise.
20150 (fix_phi_operands_for_edge): New function.
20151 (record_phi_operand_mapping): Likewise.
20152 (class pass_lower_switch): New pass.
20153 (pass_lower_switch::execute): New function.
20154 (make_pass_lower_switch): Likewise.
20155 (conditional_probability):
20156 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20157 * tree-pass.h: Add make_pass_lower_switch.
20159 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20162 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20165 2017-08-29 Richard Biener <rguenther@suse.de>
20167 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20168 we do not add a DW_AT_inline attribute twice.
20169 (gen_subprogram_die): Remove code setting DW_AT_inline on
20170 DECL_ABSTRACT_P nodes.
20172 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20174 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20175 calls to internal functions.
20176 (gimplify_modify_expr): Likewise.
20177 * tree-call-cdce.c (use_internal_fn): Likewise.
20178 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20179 (convert_to_divmod): Set the nothrow flag.
20180 * tree-if-conv.c (predicate_mem_writes): Likewise.
20181 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20182 (vectorizable_call): Likewise.
20183 (vectorizable_store): Likewise.
20184 (vectorizable_load): Likewise.
20185 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20186 (vect_recog_mask_conversion_pattern): Likewise.
20188 2017-08-29 Martin Liska <mliska@suse.cz>
20191 * gcc.c (driver_handle_option): Add new argument.
20192 * opts-common.c (handle_option): Pass
20193 target_option_override_hook.
20194 * opts-global.c (lang_handle_option): Add new option.
20195 (set_default_handlers): Add new argument.
20196 (decode_options): Likewise.
20197 * opts.c (target_handle_option): Likewise.
20198 (common_handle_option): Call target_option_override_hook.
20199 * opts.h (struct cl_option_handler_func): Add hook for
20200 target option override.
20201 (struct cl_option_handlers): Likewise.
20202 (set_default_handlers): Add new argument.
20203 (decode_options): Likewise.
20204 (common_handle_option): Likewise.
20205 (target_handle_option): Likewise.
20206 * toplev.c (toplev::main): Pass targetm.target_option.override
20209 2017-08-29 Richard Biener <rguenther@suse.de>
20210 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20212 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20213 life to the active subtree.
20215 2017-08-28 Jeff Law <law@redhat.com>
20217 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20218 derive_equivalences.
20219 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20220 Code moved into....
20221 (edge_info::derive_equivalences): New private member function
20223 * tree-ssa-dom.c (class edge_info): Changed from a struct
20224 to a class. Add ctor/dtor, methods and data members.
20225 (edge_info::edge_info): Renamed from allocate_edge_info.
20226 Initialize additional members.
20227 (edge_info::~edge_info): New.
20228 (free_dom_edge_info): Delete the edge info.
20229 (record_edge_info): Use new class & associated member functions.
20230 Tighten forms for testing for edge equivalences.
20231 (record_temporary_equivalences): Iterate over the simple
20232 equivalences rather than assuming there's only one per edge.
20233 (cprop_into_successor_phis): Iterate over the simple
20234 equivalences rather than assuming there's only one per edge.
20235 (optimize_stmt): Use operand_equal_p rather than pointer
20236 equality for mini-DSE code.
20238 2017-08-28 Nathan Sidwell <nathan@acm.org>
20240 * gcc.c (execute): Fold SIGPIPE handling into switch
20241 statement. Adjust internal error message.
20243 2017-08-28 Richard Biener <rguenther@suse.de>
20246 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20247 Do nothing for removed DIEs.
20249 2017-08-28 Richard Biener <rguenther@suse.de>
20251 PR tree-optimization/81977
20252 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20255 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20258 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20259 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20260 using targetm.gen_mem_thread_fence.
20262 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20265 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20266 predicate to register_operand. Reorder operands.
20267 (*btr<mode>): Ditto.
20268 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20269 (*btr<mode>_mask): Ditto.
20271 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20273 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20274 * config/rs6000/xmmintrin.h: New file.
20275 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20277 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20280 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20281 parameter and_insn and return it.
20282 (recombine_lvx_pattern): Insert a copy to ensure availability of
20283 the base register of the copied masking operation at the point of
20284 the instruction replacement.
20285 (recombine_stvx_pattern): Likewise.
20287 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20289 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20290 undocumented switches.
20291 (-mpower9-dform-vector): Likewise.
20292 (-mpower9-dform): Likewise.
20293 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20294 comments to delete references to -mpower9-dform* switches.
20295 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20296 Delete reference to -mpower9-dform* switches, test for
20297 -mpower9-vector instead.
20298 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20299 (OTHER_P9_VECTOR_MASKS): Likewise.
20300 (POWERPC_MASKS): Likewise.
20301 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20302 tests against -mpower9-dform* to -mpower9-vector. Delete code
20303 that checked for -mpower9-dform* consistancy with other options.
20304 Add test for -mpower9-misc to enable other power9 switches.
20305 (rs6000_init_hard_regno_mode_ok): Likewise.
20306 (rs6000_option_override_internal): Likewise.
20307 (rs6000_emit_prologue): Likewise.
20308 (rs6000_emit_epilogue): Likewise.
20309 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20310 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20311 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20313 (emit_fusion_p9_store): Likewise.
20314 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20315 resetting these macros if the assembler does not support ISA 3.0
20317 (TARGET_P9_DFORM_VECTOR): Likewise.
20318 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20319 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20321 2017-08-25 Alan Modra <amodra@gmail.com>
20323 PR rtl-optimization/81747
20324 * cse.c (cse_extended_basic_block): Don't attempt to record
20325 equivalences for degenerate conditional branches that branch
20326 to their fall-through.
20328 2017-08-24 Martin Sebor <msebor@redhat.com>
20330 PR middle-end/81908
20331 * gimple-fold.c (size_must_be_zero_p): New function.
20332 (gimple_fold_builtin_memory_op): Call it.
20334 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20336 * config/rs6000/mm_malloc.h: New file.
20338 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20340 PR tree-optimization/81913
20341 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20342 analysis when either IVs in condition can wrap.
20344 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20346 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20347 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20349 2017-08-24 Richard Biener <rguenther@suse.de>
20352 * targhooks.c (default_target_can_inline_p): Properly
20353 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20354 is present and always compare.
20355 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20356 infer -mfpmath=sse from TARGET_SSE_P.
20357 (ix86_can_inline_p): Properly use target_option_default_node when
20358 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20360 2017-08-24 Richard Biener <rguenther@suse.de>
20363 * dwarf2out.c (output_die): Handle flag_generate_offload like
20365 (output_comp_unit): Likewise.
20366 (gen_array_type_die): Likewise.
20367 (dwarf2out_early_finish): Likewise.
20368 (note_variable_value_in_expr): Likewise.
20369 (dwarf2out_finish): Likewise. Adjust assert.
20370 * cgraphunit.c (symbol_table::compile): Move setting of
20371 flag_generate_offload earlier ...
20372 (symbol_table::finalize_compilation_unit): ... here, before
20373 early debug finalization.
20375 2017-08-24 Richard Biener <rguenther@suse.de>
20377 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20378 and ipa-fnsummary.h.
20379 (ix86_can_inline_p): When ix86_fpmath flags do not match
20380 check whether the callee uses FP math at all.
20382 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20384 PR middle-end/81931
20385 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20386 instead of TYPE_PRECISION.
20388 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20389 Alan Hayward <alan.hayward@arm.com>
20390 David Sherwood <david.sherwood@arm.com>
20392 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20394 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20396 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20398 (ix86_frame::stack_realign_allocate): New field.
20399 (struct machine_frame_state): Modify comments.
20400 (machine_frame_state::sp_realigned_fp_end): New field.
20401 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20402 layout calculation.
20403 (sp_valid_at): Add assertion to assure no attempt to access invalid
20404 offset of a realigned stack.
20405 (fp_valid_at): Likewise.
20406 (choose_baseaddr): Modify comments.
20407 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20408 ix86_expand_prologue.
20409 (ix86_expand_prologue): Modify stack realignment and allocation.
20410 (ix86_expand_epilogue): Modify comments.
20411 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20412 avx2_runtime, avx512f, and avx512f_runtime.
20414 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20416 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20417 (mstackrealign): Do not init to -1.
20418 * config/i386/i386.c (ix86_option_override_internal):
20419 Check opts_set, not opts when setting default value of
20420 opts->x_ix86_force_align_arg_pointer.
20422 2017-08-23 Richard Biener <rguenther@suse.de>
20424 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20425 lang_hooks.decl_printable_name.
20426 * print-rtl-function.c (print_rtx_function): Likewise.
20427 * tree-pretty-print.c (dump_function_header): Likewise.
20429 2017-08-23 Richard Biener <rguenther@suse.de>
20432 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20433 -g0 at compile-time.
20435 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20437 PR middle-end/19706
20438 * doc/sourcebuild.texi (Other hardware attributes):
20441 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20443 PR middle-end/19706
20444 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20445 Fix single-use check.
20447 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20449 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20451 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20453 * doc/install.texi: Modify to add more details on running selected
20456 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20458 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20459 is combined with -mabi=ms.
20460 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20463 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20465 PR tree-optimization/81488
20466 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20467 and cached_basis fields.
20468 (MAX_SPREAD): New constant.
20469 (alloc_cand_and_find_basis): Initialize new fields.
20470 (clear_visited): New function.
20471 (create_phi_basis_1): Rename from create_phi_basis, set visited
20472 and cached_basis fields.
20473 (create_phi_basis): New wrapper function.
20474 (phi_add_costs_1): Rename from phi_add_costs, add spread
20475 parameter, set visited field, short-circuit when limits reached.
20476 (phi_add_costs): New wrapper function.
20477 (record_phi_increments_1): Rename from record_phi_increments, set
20479 (record_phi_increments): New wrapper function.
20480 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20481 (phi_incr_cost): New wrapper function.
20482 (all_phi_incrs_profitable_1): Rename from
20483 all_phi_incrs_profitable, set visited field.
20484 (all_phi_incrs_profitable): New wrapper function.
20486 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20487 Alan Hayward <alan.hayward@arm.com>
20488 David Sherwood <david.sherwood@arm.com>
20490 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20491 that takes the outer and inner modes.
20492 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20493 comparison as the canonical test for a paradoxical subreg.
20494 * combine.c (simplify_set): Use paradoxical_subreg_p.
20495 (make_extraction): Likewise.
20496 (force_to_mode): Likewise.
20497 (rtx_equal_for_field_assignment_p): Likewise.
20498 (gen_lowpart_for_combine): Likewise.
20499 (simplify_comparison): Likewise.
20500 * cse.c (equiv_constant): Likewise.
20501 * expmed.c (store_bit_field_1): Likewise.
20502 * final.c (alter_subreg): Likewise.
20503 * fwprop.c (propagate_rtx): Likewise.
20504 (forward_propagate_subreg): Likewise.
20505 * ira-conflicts.c (ira_build_conflicts): Likewise.
20506 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20507 * lra-constraints.c (curr_insn_transform): Likewise.
20508 (split_reg): Likewise.
20509 * lra-eliminations.c (move_plus_up): Likewise.
20510 (lra_eliminate_regs_1): Likewise.
20511 * recog.c (general_operand): Likewise.
20512 * ree.c (combine_reaching_defs): Likewise.
20513 * reload.c (push_reload): Likewise.
20514 (find_reloads): Likewise.
20515 * reload1.c (elimination_effects): Likewise.
20516 (compute_reload_subreg_offset): Likewise.
20517 (choose_reload_regs): Likewise.
20518 * rtlanal.c (subreg_lsb_1): Likewise.
20519 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20520 (simplify_subreg): Likewise.
20521 * var-tracking.c (track_loc_p): Likewise.
20522 * emit-rtl.c (byte_lowpart_offset): Likewise.
20523 (paradoxical_subreg_p): Delete out-of-line definition.
20525 2017-08-22 Jeff Law <law@redhat.com>
20527 PR tree-optimization/81741
20528 PR tree-optimization/71947
20529 * tree-ssa-dom.c: Include tree-inline.h.
20530 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20531 equivalences if one is more expensive to compute than the other.
20532 * tree-ssa-scopedtables.h (class const_or_copies): Make
20533 record_const_or_copy_raw method private.
20534 (class avail_exprs_stack): New method simplify_binary_operation.
20535 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20536 avail_exprs_stack::simplify_binary_operation as needed.
20537 (avail_exprs_stack::simplify_binary_operation): New function.
20539 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20541 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20542 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20543 (DOT_SYMBOLS): Likewise.
20544 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20545 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20546 (RS6000_ABI_NAME): Likewise.
20547 (TARGET_CMODEL): Likewise.
20548 (TOC_SECTION_ASM_OP): Likewise.
20549 (SET_CMODEL): New macro.
20550 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20552 2017-08-22 Richard Biener <rguenther@suse.de>
20554 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20555 to allow for free-lang-data replacements similar to verify_type_variant.
20557 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20558 Alan Hayward <alan.hayward@arm.com>
20559 David Sherwood <david.sherwood@arm.com>
20561 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20562 VOIDmode for the LABEL_REF.
20564 2017-08-22 Richard Biener <rguenther@suse.de>
20566 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20568 2017-08-22 Richard Biener <rguenther@suse.de>
20570 * common.opt (feliminate-dwarf2-dups): Ignore.
20571 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20572 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20573 same_die_p_wrap, compute_section_prefix,
20574 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20575 (comdat_symbol_id, comdat_symbol_number): Likewise.
20576 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20578 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20579 (output_die): Mark unreachable path unreachable.
20580 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20581 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20582 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20583 (dwarf2out_early_finish): Likewise.
20585 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20587 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20589 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20592 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20593 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20594 (avr_attribute_table) <io, io_low, address>: Initialize
20595 .decl_required with true.
20597 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20599 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20600 undocumented debugging options.
20601 (-mvsx-scalar-double): Likewise.
20602 (-mallow-df-permute): Likewise.
20603 (-mvectorize-builtins): Likewise.
20604 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20605 (rs6000_builtin_vectorized_function): Likewise.
20606 (rs6000_builtin_md_vectorized_function): Likewise.
20607 (rs6000_opt_vars): Likewise.
20609 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20612 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20613 (*btrq_imm): Rename from *btrq.
20614 (*btcq_imm): Rename from *btcq.
20615 (btsc): New code attribute.
20616 (*<btsc><mode>): New insn pattern.
20617 (*btr<mode>): Ditto.
20618 (*<btsc><mode>_mask): New insn_and_split pattern.
20619 (*btr<mode>_mask): Ditto.
20621 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20622 Alan Hayward <alan.hayward@arm.com>
20623 David Sherwood <david.sherwood@arm.com>
20625 * function.c (pad_below): Simplify padding calculation.
20627 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20628 Alan Hayward <alan.hayward@arm.com>
20629 David Sherwood <david.sherwood@arm.com>
20631 * target.def (function_prologue): Remove frame size argument.
20632 (function_epilogue): Likewise.
20633 * doc/tm.texi: Regenerate.
20634 * final.c (final_start_function): Update call to function_prologue.
20635 (final_end_function): Update call to function_epilogue.
20636 (default_function_pro_epilogue): Remove frame size argument.
20637 * output.h (default_function_pro_epilogue): Likewise.
20638 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20639 (arm_output_function_prologue): Likewise.
20640 * config/frv/frv.c (frv_function_prologue): Likewise.
20641 (frv_function_epilogue): Likewise.
20642 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20643 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20644 (ia64_output_function_epilogue): Likewise.
20645 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20646 (m32r_output_function_epilogue): Likewise.
20647 * config/microblaze/microblaze.c (microblaze_function_prologue)
20648 (microblaze_function_epilogue): Likewise.
20649 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20650 (mips_output_function_epilogue): Likewise.
20651 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20652 (mmix_target_asm_function_epilogue): Likewise.
20653 * config/msp430/msp430.c (msp430_start_function): Likewise.
20654 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20655 (nds32_asm_function_epilogue): Likewise.
20656 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20657 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20658 (pa_output_function_epilogue): Likewise.
20659 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20660 (rs6000_output_function_epilogue): Likewise.
20661 * config/rl78/rl78.c (rl78_start_function): Likewise.
20662 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20663 (rs6000_output_function_epilogue): Likewise.
20664 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20665 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20666 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20667 (sparc_asm_function_epilogue): Likewise.
20669 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20671 * tree.h (type_has_mode_precision_p): New function.
20672 * convert.c (convert_to_integer_1): Use it.
20673 * expr.c (expand_expr_real_2): Likewise.
20674 (expand_expr_real_1): Likewise.
20675 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20676 * match.pd: Likewise.
20677 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20678 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20679 * tree-tailcall.c (process_assignment): Likewise.
20680 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20681 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20682 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20683 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20684 (vectorizable_assignment): Likewise.
20685 (vectorizable_shift): Likewise.
20686 (vectorizable_operation): Likewise.
20687 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20689 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20691 * match.pd: Add pow (C, x) simplification.
20693 2017-08-21 Richard Biener <rguenther@suse.de>
20695 PR tree-optimization/81900
20696 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20697 for blocks with abnormal predecessors.
20698 (compute_antic): Do not set visited flag prematurely.
20700 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20703 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20705 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20707 * stor-layout.h (vector_type_mode): Move to...
20708 * tree.h (vector_type_mode): ...here.
20709 * stor-layout.c (vector_type_mode): Move to...
20710 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20712 2017-08-21 Richard Biener <rguenther@suse.de>
20714 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20715 register_external_die hooks.
20716 (debug_false_tree_charstarstar_uhwistar): Declare.
20717 (debug_nothing_tree_charstar_uhwi): Likewise.
20718 * debug.c (do_nothing_debug_hooks): Adjust.
20719 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20720 (debug_nothing_tree_charstar_uhwi): Likewise.
20721 * dbxout.c (dbx_debug_hooks): Adjust.
20722 (xcoff_debug_hooks): Likewise.
20723 * sdbout.c (sdb_debug_hooks): Likewise.
20724 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20725 * dwarf2out.c (macinfo_label_base): New global.
20726 (dwarf2out_register_external_die): New function for the
20727 register_external_die hook.
20728 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20729 (dwarf2_debug_hooks): Use them.
20730 (dwarf2_lineno_debug_hooks): Adjust.
20731 (struct die_struct): Add with_offset flag.
20732 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20733 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20734 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20735 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20736 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20737 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20738 defining section names for the early LTO debug variants.
20739 (reset_indirect_string): New helper.
20740 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20741 (print_dw_val): Add support for offsetted symbol references.
20742 (get_ultimate_context): Split out from is_cxx.
20743 (is_cxx): Use get_ultimate_context.
20744 (is_fortran): Add decl overload.
20745 (compute_comp_unit_symbol): Split out worker from
20746 compute_section_prefix.
20747 (compute_section_prefix): Call compute_comp_unit_symbol and
20748 set comdat_type_p here.
20749 (output_die): Skip DIE symbol output for the LTO added one.
20750 Handle DIE symbol references with offset.
20751 (output_comp_unit): Guard section name mangling properly.
20752 For LTO debug sections emit a symbol at the section beginning
20753 which we use to refer to its DIEs.
20754 (add_abstract_origin_attribute): For DIEs registered via
20755 dwarf2out_register_external_die directly refer to the early
20756 DIE rather than indirectly through the shadow one we created.
20757 Remove obsolete call to dwarf2out_abstract_function for
20758 non-function/block origins.
20759 (gen_array_type_die): When generating early LTO debug do
20760 not emit DW_AT_string_length.
20761 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20762 late when in LTO. As suggested place a gcc_unreachable for
20763 the DECL_ABSTRACT_P case.
20764 (gen_subprogram_die): Avoid another specification DIE
20765 for early built declarations/definitions for the late LTO case.
20766 (gen_variable_die): Add type references for late duplicated VLA dies
20768 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20769 we have the abstract instance already.
20770 (process_scope_var): Adjust decl DIE contexts in LTO which
20771 first puts them in limbo.
20772 (gen_decl_die): Do not generate type DIEs late apart from
20773 types for VLAs or for decls we do not yet have a DIE. Do not
20774 call dwarf2out_abstract_function late.
20775 (dwarf2out_early_global_decl): Make sure to create DIEs
20776 for abstract instances of a decl first.
20777 (dwarf2out_late_global_decl): Adjust comment.
20778 (output_macinfo_op): With multiple macro sections use
20779 macinfo_label_base to distinguish labels.
20780 (output_macinfo): Likewise. Update macinfo_label_base.
20781 Pass in the line info label.
20782 (note_variable_value_in_expr): When generating LTO resolve
20783 all variable values here by generating DIEs as needed.
20784 (init_sections_and_labels): Add early LTO debug flag parameter
20785 and generate different sections and names if set. Add generation
20786 counter for the labels so we can have multiple of them.
20787 (reset_dies): Helper to allow DIEs to be output multiple times.
20788 (dwarf2out_finish): When outputting DIEs to the fat part of an
20789 LTO object first reset DIEs.
20790 (dwarf2out_early_finish): Output early DIEs when generating LTO.
20791 (modified_type_die): Check for decl_ultimate_origin being self
20793 (gen_type_die_with_usage): Likewise.
20794 (gen_typedef_die): Allow decl_ultimate_origin being self.
20795 (set_decl_abstract_flags): Remove.
20796 (set_block_abstract_flags): Likewise.
20797 (dwarf2out_abstract_function): Treat the early generated DIEs
20798 as the abstract copy and only add DW_AT_inline and
20799 DW_AT_artificial here and call set_decl_origin_self.
20800 If the DIE has an abstract origin don't do anything.
20801 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
20802 if we have none yet (Go fails to build one, PR78628).
20803 (variably_modified_type_p): Prevent endless recursion for Ada
20804 cyclic pointer types.
20805 * lto-streamer-in.c: Include debug.h.
20806 (dref_queue): New global.
20807 (lto_read_tree_1): Stream in DIE references.
20808 (lto_input_tree): Register DIE references.
20809 (input_function): Stream DECL_DEBUG_ARGS.
20810 * lto-streamer-out.c: Include debug.h.
20811 (lto_write_tree_1): Output DIE references.
20812 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
20813 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
20814 (output_function): Stream DECL_DEBUG_ARGS.
20815 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20816 Stream DECL_ABSTRACT_ORIGIN.
20817 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20818 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20819 DECL_CONTEXT for file-scope decls.
20820 * lto-streamer.h (struct dref_entry): Declare.
20821 (dref_queue): Likewise.
20822 * cfgexpand.c (pass_expand::execute): Do not call the
20823 outlining_inline_function hook here.
20824 * lto-wrapper.c (debug_obj): New global.
20825 (tool_cleanup): Unlink it if required.
20826 (debug_objcopy): New function.
20827 (run_gcc): Handle early debug sections in the IL files by
20828 extracting them to separate files, partially linkin them and
20829 feeding the result back as result to the linker.
20830 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20831 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20832 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20833 sections into a separate segment.
20834 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20836 (darwin_asm_dwarf_section): Likewise.
20837 (darwin_asm_output_dwarf_offset): Likewise.
20838 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20840 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20841 Alan Hayward <alan.hayward@arm.com>
20842 David Sherwood <david.sherwood@arm.com>
20844 * read-md.h (md_reader::record_potential_iterator_use): Replace
20845 pointer argument with an rtx and an index.
20846 * read-rtl.c (iterator_group::apply_iterator): Likewise.
20847 (apply_mode_iterator): Likewise.
20848 (apply_code_iterator): Likewise.
20849 (apply_int_iterator): Likewise.
20850 (apply_subst_iterator): Likewise.
20851 (record_iterator_use): Likewise.
20852 (record_attribute_use): Likewise.
20853 (md_reader::record_potential_iterator_use): Likewise. Update calls
20854 to record_iterator_use and apply_iterator.
20855 (iterator_use): Replace ptr with x and index.
20856 (attribute_use): Likewise.
20857 (apply_attribute_uses): Update calls to apply_iterator.
20858 (apply_iterators): Likewise. Update initialization of iterator_use.
20859 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
20860 and record_potential_iterator_use.
20861 (rtx_reader::read_rtx_operand): Likewise.
20863 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20864 Alan Hayward <alan.hayward@arm.com>
20865 David Sherwood <david.sherwood@arm.com>
20867 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
20870 2017-08-21 Richard Biener <rguenther@suse.de>
20872 PR middle-end/81884
20873 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
20874 at struct end conservatively when comparing common bases.
20876 2017-08-21 Richard Biener <rguenther@suse.de>
20878 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20879 (mem_ref_in_stmt): Remove.
20880 (determine_max_movement): Use ref index to get at the reference.
20881 (invariantness_dom_walker::before_dom_children): Deal with
20882 lim data already initialized.
20883 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20885 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20887 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20888 (TARGET_ISA_ROUND): Ditto.
20889 (TARGET_ROUND): Ditto.
20890 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20891 * config/i386/i386.md: Ditto.
20892 * config/i386/sse.md: Ditto.
20893 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20894 with OPTION_MASK_ISA_SSE4_1.
20896 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20899 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20900 __builtin_ia32_lzcnt_u16.
20902 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20905 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20906 (rs6000_set_current_function): Rewrite function to use it.
20908 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20911 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20912 and TYPE_WARN_IF_NOT_ALIGN.
20913 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20914 (handle_warn_if_not_align): New.
20915 (place_union_field): Call handle_warn_if_not_align.
20916 (place_field): Call handle_warn_if_not_align.
20917 Copy TYPE_WARN_IF_NOT_ALIGN.
20918 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20919 (layout_type): Likewise.
20920 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20922 (tree_decl_common): Add warn_if_not_align.
20923 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20924 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20925 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20926 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20927 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20928 * doc/extend.texi: Document warn_if_not_aligned attribute.
20929 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20931 2017-08-17 Martin Liska <mliska@suse.cz>
20934 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20935 (get_data_dependence): Use it as pointer type.
20936 (distribute_loop): Likewise.
20938 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20940 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20941 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20942 (p8_vmrgow_<mode>_direct): New define_insn.
20943 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20944 handle endianness for vmrgew and vmrgow permute patterns.
20946 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20948 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20949 * config/rs6000/rs6000-cpus.def: Remove comment.
20950 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20951 (POWERPC_MASKS): Likewise.
20952 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20953 use of TARGET_VSX_TIMODE.
20954 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20955 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20956 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20957 (rs6000_option_override_internal): Remove dead code.
20958 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20959 (rs6000_legitimize_reload_address): Likewise.
20960 (rs6000_legitimate_address_p): Likewise.
20961 (rs6000_opt_masks): Delete "vsx-timode".
20962 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20963 from function comment.
20964 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20965 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20966 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20968 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20969 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20970 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20971 (VSX_TI): Likewise.
20973 (define_peephole2): Likewise.
20975 2017-08-17 Martin Sebor <msebor@redhat.com>
20978 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20979 past the end of an array.
20980 (test_pp_format): Add test cases.
20982 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20984 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20985 missing ECF_NOTHROW flags.
20987 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20990 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20991 operands residing in integer registers.
20992 (*vsx_le_perm_load_<mode>): Likewise.
20993 (*vsx_le_perm_store_<mode>): Likewise.
20994 (define_peephole2): Add peepholes to optimize the above.
20996 2017-08-17 Marek Polacek <polacek@redhat.com>
20998 PR middle-end/81814
20999 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21000 to mimic what shorten_compare did. Change the return type to bool.
21001 (fold_cond_expr_with_comparison): Update call to
21002 operand_equal_for_comparison_p.
21003 (fold_ternary_loc): Likewise.
21005 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21007 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21009 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21011 2017-08-17 Richard Biener <rguenther@suse.de>
21013 * tree-ssa-structalias.c (solve_graph): When propagating
21014 to successors update the graphs succ edges and avoid duplicate work.
21016 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21019 * config/i386/i386.c (ix86_option_override_internal): Save target
21020 specific options after ix86_stack_protector_guard_reg was changed.
21022 2017-08-17 Richard Biener <rguenther@suse.de>
21024 PR tree-optimization/81827
21025 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21026 (new_var_info): Initialize it conservatively.
21027 (get_call_vi): Mark register vars.
21028 (new_scalar_tmp_constraint_exp): Likewise.
21029 (handle_rhs_call): Likewise.
21030 (handle_const_call): Likewise.
21031 (create_function_info_for): Likewise.
21032 (solve_constraints): Sort varinfos to separate register from
21033 non-register vars to pack points-to solution bitmaps during
21036 2017-08-17 Marek Polacek <polacek@redhat.com>
21038 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21040 2017-08-17 Richard Biener <rguenther@suse.de>
21042 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21043 to true when overflow is undefined and we saturated the result.
21045 2017-08-17 Alan Modra <amodra@gmail.com>
21048 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21049 Don't use store multiple if only one reg needs saving.
21050 (interesting_frame_related_regno): New function.
21051 (rs6000_frame_related): Don't emit frame info for regs that
21053 (rs6000_emit_epilogue): Likewise.
21055 2017-08-16 Nathan Sidwell <nathan@acm.org>
21057 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21058 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21059 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21060 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21061 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21062 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21063 (verify_type): Adjust for TYPE_BINFO move.
21064 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21065 process TYPE_BINFO directly.
21066 (hash_tree): Likewise.
21067 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21069 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21072 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21074 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21076 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21079 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21080 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21081 Add ix86_binary_operator_ok to insn constraint.
21082 (*iordi_1_bts): Ditto.
21083 (*xordi_1_btc): Ditto.
21084 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21085 Update corresponding peephole2 pattern.
21089 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21091 PR tree-optimization/81832
21092 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21093 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21095 2017-08-16 Marek Polacek <polacek@redhat.com>
21098 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21099 regarding size_zero_node.
21101 2017-08-16 Martin Liska <mliska@suse.cz>
21104 * config.gcc: Respect previously set extra_objs in case
21107 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21109 PR tree-optimization/81835
21110 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21111 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21112 not depend on the phi.
21114 2017-08-16 Alan Modra <amodra@gmail.com>
21116 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21119 2017-08-16 Alan Modra <amodra@gmail.com>
21121 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21122 (save_reg_p): ..into this. Update all callers.
21123 (first_reg_to_save): Simplify.
21125 2017-08-16 Alan Modra <amodra@gmail.com>
21127 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21130 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21134 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21135 constant count if that count is less than 32.
21137 2017-08-15 Nathan Sidwell <nathan@acm.org>
21139 * gcc.c (execute): Emit friendlier message if inferior is killed
21140 by an external cause.
21142 2017-08-15 Richard Biener <rguenther@suse.de>
21144 PR tree-optimization/81790
21145 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21146 CONSTRUCTORs from simplifying and VN.
21148 2017-08-14 Martin Sebor <msebor@redhat.com>
21150 * builtin-attrs.def: Add comments.
21152 2017-08-14 Martin Sebor <msebor@redhat.com>
21155 * doc/extend.texi (attribute nonstring): Document new attribute.
21157 2017-08-14 Martin Sebor <msebor@redhat.com>
21160 * tree-diagnostic.c (default_tree_printer): Handle %G.
21161 * gimple-pretty-print.h (percent_G_format): Declare new function.
21162 * gimple-pretty-print.c (percent_G_format): Define.
21163 * tree-pretty-print.c (percent_K_format): Add argument.
21165 2017-08-14 Martin Sebor <msebor@redhat.com>
21167 PR translation/79998
21168 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21169 Remove a stray space.
21171 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21174 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21175 (*iordi_1_bts): Ditto.
21176 (*xordi_1_btc): Ditto.
21178 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21181 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21182 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21184 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21185 quoted strings, and make more translator-friendly.
21186 (darwin_rs6000_override_options): Likewise.
21187 (rs6000_option_override_internal): Likewise.
21188 (rs6000_return_in_memory): Fix overlong line.
21189 (init_cmulative_args): Use quoted strings, and make more
21190 translator-friendly.
21191 (rs6000_pass_by_reference): Fix overlong line.
21192 (def_builtin): Use quoted strings.
21193 (altivec_expand_predicate_builtin): Use quoted strings, and make
21194 more translator-friendly.
21195 (htm_expand_builtin): Use quoted strings.
21196 (cpu_expand_builtin): Use quoted strings, and make more
21197 translator-friendly.
21198 (altivec_expand_builtin): Likewise.
21199 (paired_expand_predicate_builtin): Likewise.
21200 (rs6000_invalid_builtin): Likewise.
21201 (builtin_function_type): Use quoted strings.
21202 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21203 more translator-friendly.
21204 (rs6000_trampoline_init): Likewise.
21205 (rs6000_handle_altivec_attribute): Likewise.
21206 (rs6000_inner_target_options): Use quoted strings.
21207 (rs6000_disable_incompatible_switches): Likewise.
21208 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21209 strings, and make more translator-friendly.
21210 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21212 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21214 PR tree-optimization/81799
21215 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21216 cond_expr to simple gimple operand.
21218 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21220 PR middle-end/46932
21221 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21223 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21227 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21228 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21229 TARGET_GASISR_PROLOGUES.
21230 * config/avr/avr.c (avr_option_override): Same.
21231 (avr_pass_pre_proep::execute): Same.
21233 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21236 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21237 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21239 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21241 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21242 define_insn_and_split. Split to a memory load from 0 in
21243 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21244 using PTR mode iterator.
21245 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21246 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21247 (*add_tp_<mode>): Redefine as define_insn_and_split.
21248 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21249 address space. Merge with *add_tp_x32 using PTR mode iterator.
21250 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21251 Split to an add with a memory load from 0 in
21252 DEFAULT_TLS_SEG_REG address space.
21254 2017-08-12 Andrew Pinski <apinski@cavium.com>
21256 * config/aarch64/aarch64-option-extensions.def (rdma):
21257 Fix feature string to what Linux prints out in /proc/cpuinfo.
21259 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21262 * dwarf2out.c (modified_type_die): For C typedef types that have
21263 an ultimate origin, process the ultimate origin instead of the
21265 (gen_typedef_die): Assert that input DECLs have no ultimate
21267 (gen_type_die_with_usage): For typedef variants that have an
21268 ultimate origin, just call gen_decl_die on the original DECL.
21269 (process_scope_var): Avoid creating DIEs for local typedefs and
21270 concrete static variables.
21272 2017-08-12 Alan Modra <amodra@gmail.com>
21276 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21277 match gnu-user.h startfile.
21278 (ENDFILE_LINUX_SPEC): Similarly.
21280 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21283 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21285 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21287 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21288 * config/aarch64/aarch64.md (mov<mode>): Change.
21289 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21290 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21291 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21293 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21295 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21296 for storage order barriers.
21298 2017-08-11 Martin Liska <mliska@suse.cz>
21301 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21302 variables of void type.
21304 2017-08-11 Martin Liska <mliska@suse.cz>
21306 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21307 TARGET_SUPPORTS_ALIASES.
21308 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21309 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21310 (optimize_weakref): Likewise.
21311 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21312 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21313 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21315 2017-08-11 Martin Liska <mliska@suse.cz>
21318 * config/i386/i386.c (make_resolver_func): Do complete
21319 refactoring of the function.
21321 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21324 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21325 * config/i386/i386.c (ix86_stack_protect_guard): Use
21326 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21327 * doc/invoke.texi (x86 Options): Document
21328 -mstack-protector-guard-symbol= option.
21330 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21332 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21333 * config/i386/i386.c (ix86_split_stack_guard): New function.
21334 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21335 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21336 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21337 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21338 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21339 (split_stack_space_check): Call ix86_split_stack_guard.
21341 2017-08-10 Martin Sebor <msebor@redhat.com>
21343 * print-tree.c (print_node): Print location using the established
21345 Replace spaces with colons.
21346 (debug_raw, debug): Ditto.
21348 2017-08-10 Martin Sebor <msebor@redhat.com>
21351 * pretty-print.c (pp_format): Correct the handling of %s precision.
21353 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21356 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21358 (ix86_finalize_stack_frame_flags): This. Also clear
21359 frame_pointer_needed if -fno-omit-frame-pointer is used without
21361 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21362 with ix86_finalize_stack_frame_flags.
21363 (ix86_expand_epilogue): Likewise.
21364 (ix86_expand_split_stack_prologue): Likewise.
21365 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21367 2017-08-10 Martin Liska <mliska@suse.cz>
21370 * c-attribs.c (handle_target_attribute):
21371 Report warning for an empty string argument of target attribute.
21373 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21376 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21378 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21379 or DECL_NONLOCAL labels.
21380 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21381 or DECL_NONLOCAL labels here.
21383 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21385 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21386 to indicate when early gimple folding has been disabled.
21387 (rs6000_gimple_fold_builtin): Add debug content.
21388 (rs6000_invalid_builtin): Fix whitespace.
21389 (rs6000_expand_builtin): Fix whitespace.
21390 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21392 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21395 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21396 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21398 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21400 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21401 (qdf24xx): Likewise.
21402 * config/aarch64/aarch64.md: Include falkor.md.
21403 * config/aarch64/falkor.md: New.
21405 2017-08-09 Marek Polacek <polacek@redhat.com>
21408 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21409 * diagnostic.c (emit_diagnostic): Add a comment.
21410 (emit_diagnostic_valist): New function.
21412 2017-08-09 Marek Polacek <polacek@redhat.com>
21415 * input.c (make_location): New overload.
21416 * input.h (make_location): Declare.
21418 2017-08-08 Alan Modra <amodra@gmail.com>
21419 H.J. Lu <hongjiu.lu@intel.com>
21422 * gcc.c (NO_PIE_SPEC): Delete.
21423 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21425 (LINK_PIE_SPEC): ..to here.
21426 (LINK_COMMAND_SPEC): Support -no-pie.
21427 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21428 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21430 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21431 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21432 (ENDFILE_CRTEND_SPEC): Similarly.
21434 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21437 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21438 (mstack-protector-guard-offset=): Ditto.
21439 * config/i386/i386.c (ix86_option_override): Handle
21440 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21442 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21443 ix86_stack_protect_guard_offset variables.
21444 (TARGET_STACK_PROTECT_GUARD): Always define.
21445 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21446 and -mstack-protector-guard-offset= options.
21448 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21450 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21451 boundary case for the last candidate.
21453 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21455 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21456 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21458 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21460 PR middle-end/19706
21461 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21462 * config/aarch64/aarch64-builtins.c
21463 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21464 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21465 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21467 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21468 Andrew Pinski <pinskia@gmail.com>
21470 PR middle-end/19706
21471 * internal-fn.def (XORSIGN): New.
21472 * optabs.def (xorsign_optab): New.
21473 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21474 (convert_expand_mult_copysign): New.
21475 (pass_optimize_widening_mul::execute): Call
21476 convert_expand_mult_copysign.
21478 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21480 PR tree-optimization/81354
21481 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21482 Insert on edges rather than explicitly creating landing pads.
21483 (analyze_candidates_and_replace): Commit edge inserts.
21485 2017-08-08 Richard Biener <rguenther@suse.de>
21487 PR middle-end/81719
21488 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21489 (expand_simple_operations): Also look through ADDR_EXPRs with
21490 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21492 2017-08-08 Richard Biener <rguenther@suse.de>
21494 PR tree-optimization/81723
21495 * tree-vect-slp.c (struct bst_traits): New hash traits.
21496 (bst_fail): New global.
21497 (vect_build_slp_tree_2): New worker, split out from ...
21498 (vect_build_slp_tree): ... this now wrapping it with using
21499 bst_fail set to cache SLP tree build fails. Properly handle
21501 (vect_analyze_slp_instance): Allocate and free bst_fail.
21503 2017-08-08 Martin Liska <mliska@suse.cz>
21506 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21507 LABEL_DECLs that can be from a different function.
21509 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21511 PR tree-optimization/81744
21512 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21513 loop's number of iterations.
21515 2017-08-08 Martin Liska <mliska@suse.cz>
21517 * asan.c: Include header files.
21518 * attribs.c (build_decl_attribute_variant): New function moved
21520 (build_type_attribute_qual_variant): Likewise.
21521 (cmp_attrib_identifiers): Likewise.
21522 (simple_cst_list_equal): Likewise.
21523 (omp_declare_simd_clauses_equal): Likewise.
21524 (attribute_value_equal): Likewise.
21525 (comp_type_attributes): Likewise.
21526 (build_type_attribute_variant): Likewise.
21527 (lookup_ident_attribute): Likewise.
21528 (remove_attribute): Likewise.
21529 (merge_attributes): Likewise.
21530 (merge_type_attributes): Likewise.
21531 (merge_decl_attributes): Likewise.
21532 (merge_dllimport_decl_attributes): Likewise.
21533 (handle_dll_attribute): Likewise.
21534 (attribute_list_equal): Likewise.
21535 (attribute_list_contained): Likewise.
21536 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21537 (lookup_attribute_by_prefix): Likewise.
21538 * bb-reorder.c: Include header files.
21539 * builtins.c: Likewise.
21540 * calls.c: Likewise.
21541 * cfgexpand.c: Likewise.
21542 * cgraph.c: Likewise.
21543 * cgraphunit.c: Likewise.
21544 * convert.c: Likewise.
21545 * dwarf2out.c: Likewise.
21546 * final.c: Likewise.
21547 * fold-const.c: Likewise.
21548 * function.c: Likewise.
21549 * gimple-expr.c: Likewise.
21550 * gimple-fold.c: Likewise.
21551 * gimple-pretty-print.c: Likewise.
21552 * gimple.c: Likewise.
21553 * gimplify.c: Likewise.
21554 * hsa-common.c: Likewise.
21555 * hsa-gen.c: Likewise.
21556 * internal-fn.c: Likewise.
21557 * ipa-chkp.c: Likewise.
21558 * ipa-cp.c: Likewise.
21559 * ipa-devirt.c: Likewise.
21560 * ipa-fnsummary.c: Likewise.
21561 * ipa-inline.c: Likewise.
21562 * ipa-visibility.c: Likewise.
21564 * lto-cgraph.c: Likewise.
21565 * omp-expand.c: Likewise.
21566 * omp-general.c: Likewise.
21567 * omp-low.c: Likewise.
21568 * omp-offload.c: Likewise.
21569 * omp-simd-clone.c: Likewise.
21570 * opts-global.c: Likewise.
21571 * passes.c: Likewise.
21572 * predict.c: Likewise.
21573 * sancov.c: Likewise.
21574 * sanopt.c: Likewise.
21575 * symtab.c: Likewise.
21576 * toplev.c: Likewise.
21577 * trans-mem.c: Likewise.
21578 * tree-chkp.c: Likewise.
21579 * tree-eh.c: Likewise.
21580 * tree-into-ssa.c: Likewise.
21581 * tree-object-size.c: Likewise.
21582 * tree-parloops.c: Likewise.
21583 * tree-profile.c: Likewise.
21584 * tree-ssa-ccp.c: Likewise.
21585 * tree-ssa-live.c: Likewise.
21586 * tree-ssa-loop.c: Likewise.
21587 * tree-ssa-sccvn.c: Likewise.
21588 * tree-ssa-structalias.c: Likewise.
21589 * tree-ssa.c: Likewise.
21590 * tree-streamer-in.c: Likewise.
21591 * tree-vectorizer.c: Likewise.
21592 * tree-vrp.c: Likewise.
21593 * tsan.c: Likewise.
21594 * ubsan.c: Likewise.
21595 * varasm.c: Likewise.
21596 * varpool.c: Likewise.
21597 * tree.c: Remove functions moved to attribs.[ch].
21598 * tree.h: Likewise.
21599 * config/aarch64/aarch64.c: Add attrs.h header file.
21600 * config/alpha/alpha.c: Likewise.
21601 * config/arc/arc.c: Likewise.
21602 * config/arm/arm.c: Likewise.
21603 * config/avr/avr.c: Likewise.
21604 * config/bfin/bfin.c: Likewise.
21605 * config/c6x/c6x.c: Likewise.
21606 * config/cr16/cr16.c: Likewise.
21607 * config/cris/cris.c: Likewise.
21608 * config/darwin.c: Likewise.
21609 * config/epiphany/epiphany.c: Likewise.
21610 * config/fr30/fr30.c: Likewise.
21611 * config/frv/frv.c: Likewise.
21612 * config/ft32/ft32.c: Likewise.
21613 * config/h8300/h8300.c: Likewise.
21614 * config/i386/winnt.c: Likewise.
21615 * config/ia64/ia64.c: Likewise.
21616 * config/iq2000/iq2000.c: Likewise.
21617 * config/lm32/lm32.c: Likewise.
21618 * config/m32c/m32c.c: Likewise.
21619 * config/m32r/m32r.c: Likewise.
21620 * config/m68k/m68k.c: Likewise.
21621 * config/mcore/mcore.c: Likewise.
21622 * config/microblaze/microblaze.c: Likewise.
21623 * config/mips/mips.c: Likewise.
21624 * config/mmix/mmix.c: Likewise.
21625 * config/mn10300/mn10300.c: Likewise.
21626 * config/moxie/moxie.c: Likewise.
21627 * config/msp430/msp430.c: Likewise.
21628 * config/nds32/nds32-isr.c: Likewise.
21629 * config/nds32/nds32.c: Likewise.
21630 * config/nios2/nios2.c: Likewise.
21631 * config/nvptx/nvptx.c: Likewise.
21632 * config/pa/pa.c: Likewise.
21633 * config/pdp11/pdp11.c: Likewise.
21634 * config/powerpcspe/powerpcspe.c: Likewise.
21635 * config/riscv/riscv.c: Likewise.
21636 * config/rl78/rl78.c: Likewise.
21637 * config/rx/rx.c: Likewise.
21638 * config/s390/s390.c: Likewise.
21639 * config/sh/sh.c: Likewise.
21640 * config/sol2.c: Likewise.
21641 * config/sparc/sparc.c: Likewise.
21642 * config/spu/spu.c: Likewise.
21643 * config/stormy16/stormy16.c: Likewise.
21644 * config/tilegx/tilegx.c: Likewise.
21645 * config/tilepro/tilepro.c: Likewise.
21646 * config/v850/v850.c: Likewise.
21647 * config/vax/vax.c: Likewise.
21648 * config/visium/visium.c: Likewise.
21649 * config/xtensa/xtensa.c: Likewise.
21651 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21654 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21655 constraints since the -mupper-regs-* switches have been
21657 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21658 into a vector from a double word element that was extracted from
21659 another vector, and eliminate extra XXPERMDI instructions.
21660 (vsx_concat_<mode>_2): Likewise.
21661 (vsx_concat_<mode>_3): Likewise.
21662 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21663 concat to allow optimizing inserts from previous extracts.
21665 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21667 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21668 memory reference to a SSP offset in TLS address space.
21669 (ix86_print_operand) <case '@'>: Remove.
21670 (ix86_print_operand_punct_valid_p): Remove '@' code.
21671 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21672 UNSPEC_SP_TLS_TEST.
21673 (stack_tls_protect_set_<mode>): Remove.
21674 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21675 (stack_tls_protect_test_<mode>): Remove.
21676 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21678 2017-08-07 Olivier Hainque <hainque@adacore.com>
21681 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21683 2017-08-07 Douglas Rupp <rupp@adacore.com>
21685 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21686 variable was referenced as multidir in command.
21688 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21691 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21694 2017-08-07 Martin Liska <mliska@suse.cz>
21696 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21697 * config/rl78/rl78.c: Add include of attribs.h.
21698 * config/sh/sh.c: Likewise.
21699 * config/v850/v850.c: Likewise.
21701 2017-08-07 Tom de Vries <tom@codesourcery.com>
21703 PR middle-end/78266
21704 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21706 2017-08-07 Martin Liska <mliska@suse.cz>
21708 * config/mips/mips.c: Include attribs.h.
21710 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21713 * doc/invoke.texi: Document change in behvaior for -Ofast for
21716 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21718 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21720 (aarch64_pop_regs): Likewise.
21721 (aarch64_gen_load_pair): Likewise.
21722 (aarch64_save_callee_saves): Likewise.
21723 (aarch64_restore_callee_saves): Likewise.
21725 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21727 * config/i386/i386.c: Revert the last change.
21729 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21732 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21734 (ix86_finalize_stack_frame_flags): This. Also clear
21735 frame_pointer_needed if -fno-omit-frame-pointer is used without
21737 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21738 with ix86_finalize_stack_frame_flags.
21739 (ix86_expand_epilogue): Likewise.
21740 (ix86_expand_split_stack_prologue): Likewise.
21742 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21745 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21746 to P_AES for Westmere.
21748 2017-08-07 Jonathan Yong <10walls@gmail.com>
21750 * config/i386/mingw.opt (fset-stack-executable): Removed.
21751 * config/i386/cygming.opt (fset-stack-executable): Moved
21753 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21755 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21757 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21759 2017-08-07 Marek Polacek <polacek@redhat.com>
21761 PR middle-end/81737
21762 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21764 2017-08-07 Martin Liska <mliska@suse.cz>
21766 * attribs.h (canonicalize_attr_name): New function.
21767 (cmp_attribs): Move from c-format.c and adjusted.
21768 (is_attribute_p): Moved from tree.h.
21769 * tree-inline.c: Add new includes.
21770 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21771 (private_is_attribute_p): Remove.
21772 (private_lookup_attribute): Likewise.
21773 (private_lookup_attribute_by_prefix): Simplify.
21774 (remove_attribute): Use is_attribute_p.
21775 * tree.h: Remove removed declarations.
21777 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21779 PR middle-end/81698
21780 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21781 instead of computing it in the function. Formatting fix.
21782 (expand_case): Don't rely on default_edge being the first edge,
21783 clear it if removing it, pass default_edge to
21784 emit_case_dispatch_table.
21785 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21788 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
21790 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
21791 insn in the function, emit NOP after the insn.
21793 2017-08-06 Tom de Vries <tom@codesourcery.com>
21795 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
21798 2017-08-06 Tom de Vries <tom@codesourcery.com>
21800 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
21803 2017-08-04 Yury Gribov <tetra2005@gmail.com>
21805 PR tree-optimization/57371
21806 * match.pd: New pattern.
21808 2017-08-04 Marek Polacek <polacek@redhat.com>
21810 PR middle-end/81695
21811 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
21812 perform the computation in offset_int.
21814 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21816 PR tree-optimization/81136
21817 * tree-vectorizer.h: Include tree-hash-traits.h.
21818 (vec_base_alignments): New typedef.
21819 (vec_info): Add a base_alignments field.
21820 (vect_record_base_alignments): Declare.
21821 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21823 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21824 (create_data_ref): Add an is_conditional_in_stmt argument.
21825 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21826 the is_conditional_in_stmt field.
21827 (data_ref_loc): Add an is_conditional_in_stmt field.
21828 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21829 (find_data_references_in_stmt): Update call to create_data_ref.
21830 (graphite_find_data_references_in_stmt): Likewise.
21831 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21832 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21833 (vect_record_base_alignment): New function.
21834 (vect_record_base_alignments): Likewise.
21835 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21836 for nested statements even if we fail to compute a misalignment.
21837 Use pooled base alignments for unconditional references.
21838 (vect_find_same_alignment_drs): Compare base addresses instead
21840 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21841 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
21843 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21845 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
21846 Add an explicit name for the enum. Use auto_vec for slp_instances
21847 and grouped_stores.
21848 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
21850 (_bb_vec_info): Add a constructor and destructor.
21851 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21852 (destroy_loop_vec_info): Delete.
21853 (vect_destroy_datarefs): Likewise.
21854 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21855 (vec_info::vec_info): New function.
21856 (vec_info::~vec_info): Likewise.
21857 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
21858 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
21859 destroy_loop_vec_info.
21860 * tree-vect-loop.c (new_loop_vec_info): Replace with...
21861 (_loop_vec_info::_loop_vec_info): ...this.
21862 (destroy_loop_vec_info): Replace with...
21863 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
21864 the stmt_vec_infos. Leave handling of vec_info information to its
21865 destructor. Remove explicit vector releases.
21866 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
21867 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
21868 * tree-vect-slp.c (new_bb_vec_info): Replace with...
21869 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
21870 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
21871 (destroy_bb_vec_info): Replace with...
21872 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
21873 information to its destructor.
21874 (vect_slp_analyze_bb_1): Use new and delete instead of
21875 new_bb_vec_info and destroy_bb_vec_info.
21876 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21879 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21881 * tree-data-ref.h (subscript): Add access_fn field.
21882 (data_dependence_relation): Add could_be_independent_p.
21883 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21884 (same_access_functions): Move to tree-data-ref.c.
21885 * tree-data-ref.c (ref_contains_union_access_p): New function.
21886 (access_fn_component_p): Likewise.
21887 (access_fn_components_comparable_p): Likewise.
21888 (dr_analyze_indices): Add a reference to access_fn_component_p.
21889 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21891 (constant_access_functions): Likewise.
21892 (add_other_self_distances): Likewise.
21893 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21894 (initialize_data_dependence_relation): Use XCNEW and remove
21895 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21896 of access functions that have the same type. Allow the
21897 subsequence to end with different bases in some circumstances.
21898 Record the chosen access functions in SUB_ACCESS_FN.
21899 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21900 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21901 (subscript_dependence_tester_1): Likewise dra and drb.
21902 (build_classic_dist_vector): Update calls accordingly.
21903 (subscript_dependence_tester): Likewise.
21904 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21905 DDR_COULD_BE_INDEPENDENT_P.
21906 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21907 comp_alias_ddrs instead of may_alias_ddrs.
21908 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21910 (vect_analyze_data_ref_dependence): Use it if
21911 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21912 distance vectors if that fails.
21913 (dependence_distance_ge_vf): New function.
21914 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21915 LOOP_VINFO_MAY_ALIAS_DDRS.
21917 2017-08-04 Richard Biener <rguenther@suse.de>
21919 PR middle-end/81705
21920 * fold-const.c (fold_binary_loc): Properly restrict
21921 minus_var0 && minus_var1 case when associating undefined overflow
21924 2017-08-04 Tom de Vries <tom@codesourcery.com>
21926 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21928 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21930 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21931 Don't start diagnostic messages with a capital letter.
21932 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21934 (rs6000_invalid_builtin): Likewise.
21935 (rs6000_trampoline_init): Likewise.
21937 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21940 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21941 after setting changeable df flags.
21943 2017-08-03 Richard Biener <rguenther@suse.de>
21945 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21946 up if the use is in USE - X.
21948 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21950 * toplev.c (dumpfile.h): New include.
21951 (internal_error_reentered): New static function. Use it...
21952 (internal_error_function): ...here to handle reentered internal_error.
21954 2017-08-03 Richard Biener <rguenther@suse.de>
21956 PR middle-end/81148
21957 * fold-const.c (split_tree): Add minus_var and minus_con
21958 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21959 here but always use minus_*.
21960 (associate_trees): Assert we never associate with MINUS_EXPR
21961 and NULL first operand. Do not recurse for PLUS_EXPR operands
21962 when associating as MINUS_EXPR either.
21963 (fold_binary_loc): Track minus_var and minus_con.
21965 2017-08-03 Tom de Vries <tom@codesourcery.com>
21968 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21969 ACCEL_COMPILER, apply finish_options on
21970 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21972 2017-08-03 Tom de Vries <tom@codesourcery.com>
21975 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21976 function_entry_patch_area_size > 0.
21978 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21981 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21982 instead of 10??LU, perform unit multiplication in wide_int,
21983 don't change alloc_object_size_limit if the limit is larger
21986 PR tree-optimization/81655
21987 PR tree-optimization/81588
21988 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21989 the case when ranges[i].low and high are 1 for unsigned type with
21992 PR middle-end/81052
21993 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21994 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21996 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21998 * tree-vrp.h: Add include guard.
22000 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22003 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22004 (ud2): New insn pattern.
22005 * config/i386/i386.c (ix86_expand_epilogue):
22006 For naked functions, generate ud2 instead of trap insn.
22008 2017-08-02 Marek Polacek <polacek@redhat.com>
22011 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22013 2017-08-02 Tom de Vries <tom@codesourcery.com>
22014 Cesar Philippidis <cesar@codesourcery.com>
22016 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22017 Add missing edge probabilities.
22019 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22021 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22022 Correct endianness.
22024 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22026 PR middle-end/79499
22027 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22028 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22029 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22030 prologue_seq sequences - if any.
22032 2017-08-02 Richard Biener <rguenther@suse.de>
22034 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22035 via vectors if supported, integer extracts via punning if supported
22036 or otherwise vector extracts.
22038 2017-08-02 Richard Biener <rguenther@suse.de>
22040 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22042 (bitmap_insert_into_set): ... this.
22044 2017-08-02 Richard Biener <rguenther@suse.de>
22046 PR tree-optimization/81633
22048 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22050 PR tree-optimization/71752
22051 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22053 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22055 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22056 (machine_function::call_ms2sysv_pad_out): Remove field.
22057 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22058 (ix86_compute_frame_layout): Likewise.
22060 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22063 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22064 attribute with interrupt attribute.
22066 2017-08-01 Andrew Pinski <apinski@cavium.com>
22068 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22069 BIT_INSERT_EXPR's operand 1
22070 to see if the types precision matches.
22072 2017-08-01 Martin Liska <mliska@suse.cz>
22074 PR middle-end/70140
22075 * builtins.c (expand_builtin_memcpy_args): Remove.
22076 (expand_builtin_memcpy): Call newly added function
22077 expand_builtin_memory_copy_args.
22078 (expand_builtin_memcpy_with_bounds): Likewise.
22079 (expand_builtin_mempcpy): Remove last argument.
22080 (expand_builtin_mempcpy_with_bounds): Likewise.
22081 (expand_builtin_memory_copy_args): New function created from
22082 expand_builtin_mempcpy_args with small modifications.
22083 (expand_builtin_mempcpy_args): Remove.
22084 (expand_builtin_stpcpy): Remove unused argument.
22085 (expand_builtin): Likewise.
22086 (expand_builtin_with_bounds): Likewise.
22088 2017-08-01 Martin Liska <mliska@suse.cz>
22091 Make mempcpy more optimal (PR middle-end/70140).
22093 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22096 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22097 __builtin_vec_cmpne verify both arguments are compatible vectors
22098 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22099 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22100 move computation of aligned to after checking the argument types.
22104 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22105 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22108 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22109 Doug Rupp <rupp@adacore.com>
22110 Olivier Hainque <hainque@adacore.com>
22112 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22113 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22115 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22116 for TARGET_OS_CPP_BUILTIN.
22117 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22118 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22120 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22121 passing required abi options to the assembler for EABI configurations.
22122 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22123 of .text.hot and .text.unlikely sections for kernel modules when
22124 using ARM style exceptions.
22125 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22126 options. Add EXTRA_CC1_SPEC.
22127 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22129 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22131 (ARM_TARGET2_DWARF_FORMAT): Define.
22132 * config/arm/t-vxworks: Adjust multilib control to removal of the
22133 Diab command line options.
22135 2017-08-01 Martin Liska <mliska@suse.cz>
22137 PR gcov-profile/81561
22138 * gcov.c (unblock): Make unblocking safe as we need to preserve
22139 index correspondence of blocks and block_lists.
22141 2017-08-01 Richard Biener <rguenther@suse.de>
22143 PR tree-optimization/81181
22144 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22145 (compute_antic): ... end of iteration here.
22147 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22149 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22150 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22151 (ftree-slp-vectorize): Likewise.
22152 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22153 can no longer be set independent of flag_tree_loop_vectorize.
22154 * omp-general.c (emp_max_vf): Likewise.
22155 * opts.c (enable_fdo_optimizations): Remove references to
22156 flag_tree_vectorize, these are now implicit.
22157 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22158 and leave it for the options machinery.
22160 2017-08-01 Martin Liska <mliska@suse.cz>
22162 PR middle-end/70140
22163 * builtins.c (expand_builtin_memcpy_args): Remove.
22164 (expand_builtin_memcpy): Call newly added function
22165 expand_builtin_memory_copy_args.
22166 (expand_builtin_memcpy_with_bounds): Likewise.
22167 (expand_builtin_mempcpy): Remove last argument.
22168 (expand_builtin_mempcpy_with_bounds): Likewise.
22169 (expand_builtin_memory_copy_args): New function created from
22170 expand_builtin_mempcpy_args with small modifications.
22171 (expand_builtin_mempcpy_args): Remove.
22172 (expand_builtin_stpcpy): Remove unused argument.
22173 (expand_builtin): Likewise.
22174 (expand_builtin_with_bounds): Likewise.
22176 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22179 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22180 print "ds:" only for immediates in generic address space.
22182 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22185 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22186 (ix86_function_ok_for_sibcall): Return false for naked functions.
22188 2017-08-01 Richard Biener <rguenther@suse.de>
22190 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22191 (compute_antic): Seed worklist with exit block predecessors.
22192 * cfganal.c (dfs_find_deadend): For a cycle return the source
22193 of the edge closing it.
22195 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22197 * config/aarch64/aarch64.c
22198 (aarch64_can_const_movi_rtx_p): Move 0 check.
22200 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22202 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22203 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22205 * match.pd: Ditto in address comparison pattern.
22207 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22209 PR tree-optimization/81627
22210 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22211 closed ssa form for store-store chain.
22213 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22215 PR tree-optimization/81620
22216 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22217 for store-store chain.
22219 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22221 PR tree-optimization/81588
22222 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22223 ranges[i].in_p, invert comparison code ccode. For >/>=,
22224 swap rhs1 and rhs2 and comparison code unconditionally,
22225 for </<= don't do that. Don't swap rhs1/rhs2 again if
22226 ranges[i].in_p, instead invert comparison code ccode if
22227 opcode or oe->rank is BIT_IOR_EXPR.
22230 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22231 a direct optab to conversion optab.
22232 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22233 with GET_MODE_INNER as last argument instead of optab_handler.
22234 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22235 vector extraction if possible and optab is available.
22236 * expr.c (store_constructor): Use convert_optab_handler instead
22237 of optab_handler. Use vector initialization from smaller
22238 vectors if possible and optab is available.
22239 * tree-vect-stmts.c (vectorizable_load): Likewise.
22240 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22241 now have two modes.
22242 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22243 of vec_init from half-sized vectors with the same element mode.
22244 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22245 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22246 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22247 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22248 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22249 after mode in gen_vec_extract* calls.
22250 (vec_extract<mode>): Renamed to ...
22251 (vec_extract<mode><ssescalarmodelower>): ... this.
22252 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22253 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22254 element mode after mode in gen_vec_init* calls.
22255 (VEC_INIT_HALF_MODE): New mode iterator.
22256 (vec_init<mode>): Renamed to ...
22257 (vec_init<mode><ssescalarmodelower>): ... this.
22258 (vec_init<mode><ssehalfvecmodelower>): New expander.
22259 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22260 (vec_extractv2sfsf): ... this.
22261 (vec_initv2sf): Renamed to ...
22262 (vec_initv2sfsf): ... this.
22263 (vec_extractv2si): Renamed to ...
22264 (vec_extractv2sisi): ... this.
22265 (vec_initv2si): Renamed to ...
22266 (vec_initv2sisi): ... this.
22267 (vec_extractv4hi): Renamed to ...
22268 (vec_extractv4hihi): ... this.
22269 (vec_initv4hi): Renamed to ...
22270 (vec_initv4hihi): ... this.
22271 (vec_extractv8qi): Renamed to ...
22272 (vec_extractv8qiqi): ... this.
22273 (vec_initv8qi): Renamed to ...
22274 (vec_initv8qiqi): ... this.
22275 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22276 (vec_init<mode>): Renamed to ...
22277 (vec_init<mode><VEC_base_l>): ... this.
22278 (vec_extract<mode>): Renamed to ...
22279 (vec_extract<mode><VEC_base_l>): ... this.
22280 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22281 (vec_initv2sfsf): ... this.
22282 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22283 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22284 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22285 element mode after mode in gen_vec_init* calls.
22286 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22287 (vec_init<mode><Vel>): ... this.
22288 (vec_extract<mode>): Renamed to ...
22289 (vec_extract<mode><Vel>): ... this.
22290 * config/aarch64/iterators.md (Vel): New mode attribute.
22291 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22292 Add element mode after mode in gen_vec_extract* calls.
22293 * config/s390/vector.md (non_vec_l): New mode attribute.
22294 (vec_extract<mode>): Renamed to ...
22295 (vec_extract<mode><non_vec_l>): ... this.
22296 (vec_init<mode>): Renamed to ...
22297 (vec_init<mode><non_vec_l>): ... this.
22298 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22299 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22301 * config/arm/iterators.md (V_elem_l): New mode attribute.
22302 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22303 (vec_extract<mode><V_elem_l>): ... this.
22304 (vec_extractv2di): Renamed to ...
22305 (vec_extractv2didi): ... this.
22306 (vec_init<mode>): Renamed to ...
22307 (vec_init<mode><V_elem_l>): ... this.
22308 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22309 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22310 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22311 Add element mode after gen_vec_extract* calls.
22312 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22313 (vec_init<mode><unitmode>): ... this.
22314 (vec_extract<mode>): Renamed to ...
22315 (vec_extract<mode><unitmode>): ... this.
22316 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22317 (vec_init<mode><unitmode>): ... this.
22318 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22319 (vec_initv2sfsf): ... this.
22320 (vec_extractv2sf): Renamed to ...
22321 (vec_extractv2sfsf): ... this.
22322 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22323 Add element mode after gen_vec_extract* calls.
22324 * config/mips/mips.md (unitmode): New mode iterator.
22325 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22326 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22327 * config/spu/spu.md (inner_l): New mode attribute.
22328 (vec_init<mode>): Renamed to ...
22329 (vec_init<mode><inner_l>): ... this.
22330 (vec_extract<mode>): Renamed to ...
22331 (vec_extract<mode><inner_l>): ... this.
22332 * config/sparc/sparc.md (veltmode): New mode iterator.
22333 (vec_init<VMALL:mode>): Renamed to ...
22334 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22335 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22336 (vec_initv2sisi): ... this.
22337 (vec_initv2sf): Renamed to ...
22338 (vec_initv2sfsf): ... this.
22339 (vec_extractv2sf): Renamed to ...
22340 (vec_extractv2sfsf): ... this.
22341 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22342 (vec_init<mode>): Renamed to ...
22343 (vec_init<mode><VEC_base_l>): ... this.
22344 (vec_extract<mode>): Renamed to ...
22345 (vec_extract<mode><VEC_base_l>): ... this.
22346 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22347 (vec_initv2sfsf): ... this.
22348 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22349 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22350 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22351 gen_vec_init* calls.
22353 2017-08-01 Richard Biener <rguenther@suse.de>
22355 PR tree-optimization/81297
22356 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22357 TREE_OVERFLOW from INTEGER_CSTs.
22359 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22361 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22363 2017-07-31 Carl Love <cel@us.ibm.com>
22365 * config/rs6000/rs6000-c: Add support for built-in functions
22366 vector signed char vec_xl_be (signed long long, signed char *);
22367 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22368 vector signed int vec_xl_be (signed long long, signed int *);
22369 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22370 vector signed long long vec_xl_be (signed long long, signed long long *);
22371 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22372 vector signed short vec_xl_be (signed long long, signed short *);
22373 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22374 vector double vec_xl_be (signed long long, double *);
22375 vector float vec_xl_be (signed long long, float *);
22376 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22377 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22378 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22380 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22381 (altivec_expand_builtin): Add switch statement to call
22382 altivec_expand_xl_be for each builtin.
22383 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22384 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22385 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22386 __builtin_vsx_le_be_v16qi.
22387 * doc/extend.texi: Update the built-in documentation file for the
22388 new built-in functions.
22390 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22393 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22395 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22397 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22399 * config.gcc: Add z14.
22400 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22401 CPU model numbers for z13s and z14.
22402 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22404 * config/s390/s390-opts.h (enum processor_type): Rename
22405 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22406 * config/s390/s390.c (processor_table): Add field for CPU name to
22407 be passed to Binutils.
22408 (s390_asm_output_machine_for_arch): Use the new field in
22409 processor_table for Binutils.
22410 (s390_expand_builtin): Replace arch12 with z14.
22411 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22412 (s390_get_sched_attrmask): Likewise.
22413 (s390_get_unit_mask): Likewise.
22414 * config/s390/s390.opt: Add z14 to processor_type enum.
22416 2017-07-31 Martin Jambor <mjambor@suse.cz>
22419 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22420 regardless of optimization level.
22422 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22423 Martin Liska <mliska@suse.cz>
22425 * predict.def: Remove old comment and adjust probability.
22426 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22427 PREDICT statements.
22429 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22432 * config/i386/i386.c (ix86_function_naked): New function.
22433 (ix86_can_use_return_insn_p): Return false for naked functions.
22434 (ix86_expand_prologue): Skip prologue for naked functions.
22435 (ix86_expand_epilogue): Skip epilogue for naked functions
22436 and emit trap instruction.
22437 (ix86_warn_func_return): New function.
22438 (ix86_attribute_table): Add "naked" attribute specification.
22439 (TARGET_WARN_FUNC_RETURN): Define.
22440 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22442 2017-07-31 Martin Liska <mliska@suse.cz>
22444 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22445 (dump_gimple_bb_header): Always dump BB info.
22446 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22448 2017-07-31 Martin Liska <mliska@suse.cz>
22451 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22452 also with current_function_decl non-null equality.
22454 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22457 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22458 change type to the element type, instead add eltype variable and
22459 use it where we are interested in the element type.
22461 PR tree-optimization/81603
22462 * ipa-polymorphic-call.c
22463 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22464 offset arithmetic in offset_int, bail out if the resulting bit offset
22465 doesn't fit into shwi.
22467 2017-07-31 Martin Liska <mliska@suse.cz>
22469 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22470 (gimplify_save_expr): Fix comment.
22472 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22475 * config/i386/i386.c (ix86_function_arg): Update arguments for
22477 (ix86_compute_frame_layout): Set the initial stack offset to
22478 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22479 INCOMING_FRAME_SP_OFFSET.
22480 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22481 stack before exception handler returns.
22482 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22483 the 'ERROR_CODE' for exception handler.
22485 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22487 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22488 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22489 (ASM_OUTPUT_REG_POP): Ditto.
22490 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22491 instead of asm_fprintf to output pure string.
22493 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22495 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22496 to imported_module_or_decl hook.
22497 (debug_nothing_tree_tree_tree_bool): Remove.
22498 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22499 * debug.c (do_nothing_debug_hooks): Use
22500 debug_nothing_tree_tree_tree_bool_bool instead of
22501 debug_nothing_tree_tree_tree_bool.
22502 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22503 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22504 * sdbout.c (sdb_debug_hooks): Likewise.
22505 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22506 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22508 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22509 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22510 attribute, don't add anything.
22512 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22514 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22515 (fold_build2_stat_loc): Likewise.
22516 (fold_build3_stat_loc): Likewise.
22517 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22518 (fold_build1_loc): Remove macro.
22519 (fold_build2_loc): Likewise.
22520 (fold_build3_loc): Likewise.
22522 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22524 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22525 (gimple_build_debug_bind_source_stat): Likewise.
22526 * gimple.h (gimple_build_debug_bind): Remove macro.
22527 (gimple_build_debug_bind_source): Likewise.
22529 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22531 * bitmap.c (bitmap_alloc): Adjust.
22532 (bitmap_gc_alloc): Likewise.
22533 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22535 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22537 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22538 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22539 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22540 (bitmap_gc_alloc_stat): Likewise.
22541 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22543 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22545 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22546 * rtl.h (shallow_copy_rtx): Remove macro.
22548 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22550 * emit-rtl.c (gen_raw_REG): Adjust.
22551 * gengenrtl.c (gendef): Likewise.
22552 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22553 * rtl.h (rtx_alloc): Remove macro.
22555 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22557 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22558 (build_tree_list_stat): Likewise.
22559 * tree.h (build_tree_list): Remove macro.
22560 (build_tree_list_vec): Likewise.
22562 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22564 * tree.c (make_vector_stat): Remove _stat from name.
22565 (build_vector_stat): Likewise.
22566 * tree.h (make_vector_stat): Remove macro.
22567 (build_vector_stat): Likewise.
22569 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22571 * tree.h (build_var_debug_value): Remove prototype.
22573 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22575 * tree.c (tree_cons_stat): Remove _stat from name.
22576 * tree.h (tree_cons): Remove macro.
22578 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22580 * tree.c (build_vl_exp_stat): Remove _stat from name.
22581 * tree.h (build_vl_exp): Remove macro.
22583 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22585 * tree.c (build_decl_stat): Remove _stat from name.
22586 * tree.h (build_decl): Remove macro.
22588 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22590 * gimple.c (gimple_build_with_ops_stat): Adjust.
22591 (gimple_alloc_stat): Remove _stat from name.
22592 * gimple.h (gimple_alloc): Remove macro.
22594 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22596 * tree.c (make_tree_vec_stat): Remove _stat from name.
22597 (grow_tree_vec_stat): Likewise.
22598 * tree.h (make_tree_vec_stat): Adjust prototype.
22599 (grow_tree_vec_stat): Likewise.
22600 (make_tree_vec): Remove macro.
22601 (grow_tree_vec): Likewise.
22603 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22605 * fold-const.c (fold_build1_stat_loc): Adjust.
22606 (fold_build2_stat_loc): Likewise.
22607 (fold_build3_stat_loc): Likewise.
22608 * tree.c (build0_stat): Remove _stat from name.
22609 (build1_stat): Likewise.
22610 (build2_stat): Likewise.
22611 (build3_stat): Likewise.
22612 (build4_stat): Likewise.
22613 (build5_stat): Likewise.
22614 * tree.h (build1_loc): Remove macro, and rename _stat function
22616 (build2_loc): Likewise.
22617 (build3_loc): Likewise.
22618 (build4_loc): Likewise.
22619 (build5_loc): Likewise.
22621 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22623 * tree.c (make_int_cst_stat): Remove _stat from name.
22624 * tree.h (make_int_cst_stat): Adjust prototype.
22625 (make_int_cst): Remove macro.
22627 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22629 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22630 * tree.h (make_tree_binfo_stat): Adjust prototype.
22631 (make_tree_binfo): Remove.
22633 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22635 * tree.c (copy_node_stat): Rename to copy_node.
22636 (build_distinct_type_copy): Adjust.
22637 * tree.h (copy_node_stat): Adjust prototype.
22638 (copy_node): Remove macro.
22640 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22642 * tree.c (make_node_stat): rename to make_node.
22643 (build_tree_list_stat): Adjust.
22644 (build0_stat): Likewise.
22645 (build2_stat): Likewise.
22646 (build3_stat): Likewise.
22647 (build4_stat): Likewise.
22648 (build5_stat): Likewise.
22649 (build_decl_stat): Likewise.
22650 * tree.h (make_node_stat): Adjust prototype.
22651 (make_node): remove macro.
22653 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22655 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22656 (PPC_FEATURE2_SCV): Likewise.
22657 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22659 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22661 * config/aarch64/aarch64.c
22662 (aarch64_internal_mov_immediate): Add new special pattern.
22663 * config/aarch64/aarch64.md (*movdi_aarch64):
22664 Add reg/32bit const mov case.
22666 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22667 Richard Sandiford <richard.sandiford@linaro.org>
22669 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22670 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22671 Add integer and movi cases.
22672 (movi-split-hf-df-sf split, fp16): New.
22673 (enabled): Added TARGET_FP_F16INST.
22674 * config/aarch64/iterators.md (GPF_HF): New.
22675 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22677 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22679 * config/aarch64/aarch64.c
22680 (aarch64_simd_container_mode): Add prototype.
22681 (aarch64_expand_mov_immediate): Add HI support.
22682 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22683 (aarch64_can_const_movi_rtx_p): New.
22684 (aarch64_preferred_reload_class):
22685 Remove restrictions of using FP registers for certain SIMD operations.
22686 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22687 (aarch64_valid_floating_const): Add integer move validation.
22688 (aarch64_simd_imm_scalar_p): Remove.
22689 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22690 (aarch64_legitimate_constant_p): Expand list of supported cases.
22691 * config/aarch64/aarch64-protos.h
22692 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22693 (aarch64_reinterpret_float_as_int): New.
22694 (aarch64_simd_imm_scalar_p): Remove.
22695 * config/aarch64/constraints.md (Uvi): New.
22696 (Dd): Split into Ds and new Dd.
22697 * config/aarch64/aarch64.md (*movsi_aarch64):
22699 (*movdi_aarch64): Add SIMD mov case.
22701 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22703 * tree-predcom.c: (struct chain): Handle store-store chain in which
22704 stores for elimination only store loop invariant values.
22705 (execute_pred_commoning_chain): Ditto.
22706 (prepare_initializers_chain_store_elim): Ditto.
22707 (prepare_finalizers): Ditto.
22708 (is_inv_store_elimination_chain): New function.
22709 (initialize_root_vars_store_elim_1): New function.
22711 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22713 * tree-predcom.c: Revise general description of the pass.
22714 (enum chain_type): New enum type for store elimination.
22715 (struct chain): New field supporting store elimination.
22716 (struct component): Ditto.
22717 (dump_chain): Dump store-stores chain.
22718 (release_chain): Release resources.
22719 (split_data_refs_to_components): Compute and create component
22720 contains only stores for elimination.
22721 (get_chain_last_ref_at): New function.
22722 (make_invariant_chain): Initialization.
22723 (make_rooted_chain): Specify chain type in parameter and record it.
22724 (add_looparound_copies): Skip for store-stores chain.
22725 (determine_roots_comp): Compute type of chain and pass it to
22727 (initialize_root_vars_store_elim_2): New function.
22728 (finalize_eliminated_stores): New function.
22729 (remove_stmt): Handle store for elimination.
22730 (execute_pred_commoning_chain): Execute predictive commoning on
22731 store-store chains.
22732 (determine_unroll_factor): Skip unroll for store-stores chain.
22733 (prepare_initializers_chain_store_elim): New function.
22734 (prepare_initializers_chain): Hanlde store-store chain.
22735 (prepare_finalizers_chain, prepare_finalizers): New function.
22736 (tree_predictive_commoning_loop): Return integer value indicating
22737 if loop is unrolled or lcssa form is corrupted.
22738 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22740 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22742 * tree-predcom.c (initialize_root): Delete.
22743 (execute_pred_commoning_chain): Initialize root vars and replace
22744 reference of non-combined chain directly, rather than call above
22747 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22749 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22750 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22752 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22754 * tree-predcom.c (struct chain): New field init_seq.
22755 (release_chain): Release init_seq.
22756 (prepare_initializers_chain): Record intialization stmts in above
22758 (insert_init_seqs): New function.
22759 (tree_predictive_commoning_loop): Call insert_init_seqs.
22761 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22763 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22765 2017-07-28 Richard Biener <rguenther@suse.de>
22767 * match.pd: Remove superfluous :c.
22768 * genmatch.c (simplify::id): Add member.
22769 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22771 (current_id): New global.
22772 (dt_node::parent): Move from ...
22773 (dt_operand::parent): ... here. Add for_id member.
22774 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22775 (decision_tree::find_node): Relax order requirement when
22776 merging DT_TRUE nodes to ones inbetween the current simplify
22777 and the one we try to merge with. Add diagnostic whenever
22778 we need to enforce pattern order by not merging.
22779 (decision_tree::insert): Set current_id.
22780 (decision_tree::print_node): Dump parent node and for_id.
22781 (parser::last_id): Add member.
22782 (parser::push_simplify): Assign unique id.
22783 (parser::parser): Initialize last_id.
22785 2017-07-28 Martin Liska <mliska@suse.cz>
22788 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22789 gimple_build_debug_bind.
22791 2017-07-28 Richard Biener <rguenther@suse.de>
22793 PR tree-optimization/81502
22794 * match.pd: Add pattern combining BIT_INSERT_EXPR with
22796 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
22798 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
22799 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
22800 for BIT_FIELD_REF args.
22801 * fold-const.c (make_bit_field_ref): Likewise.
22802 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
22804 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22807 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
22808 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
22809 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
22810 Or it into SANITIZER_UNDEFINED.
22811 * ubsan.c: Include gimple-fold.h and varasm.h.
22812 (ubsan_expand_ptr_ifn): New function.
22813 (instrument_pointer_overflow): New function.
22814 (maybe_instrument_pointer_overflow): New function.
22815 (instrument_object_size): Formatting fix.
22816 (pass_ubsan::execute): Call instrument_pointer_overflow
22817 and maybe_instrument_pointer_overflow.
22818 * internal-fn.c (expand_UBSAN_PTR): New function.
22819 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22820 * sanitizer.def (__ubsan_handle_pointer_overflow,
22821 __ubsan_handle_pointer_overflow_abort): New builtins.
22822 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22823 * internal-fn.def (UBSAN_PTR): New internal function.
22824 * opts.c (sanitizer_opts): Add pointer-overflow.
22825 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22826 * fold-const.c (build_range_check): Compute pointer range check in
22827 integral type if pointer arithmetics would be needed. Formatting
22830 2017-07-28 Martin Liska <mliska@suse.cz>
22833 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22834 parameters that are of a variable-length.
22836 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22838 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22840 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22841 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22842 (CRT_CALL_STATIC_FUNCTION): Likewise.
22843 (ASM_DEFAULT_SPEC): New define.
22844 (ASM_SPEC32): Likewise.
22845 (ASM_SPEC64): Likewise.
22846 (ASM_SPEC_COMMON): Likewise.
22847 (ASM_SPEC): Likewise.
22848 (INVALID_64BIT): Likewise.
22849 (LINK_OS_DEFAULT_SPEC): Likewise.
22850 (LINK_OS_SPEC32): Likewise.
22851 (LINK_OS_SPEC64): Likewise.
22852 (POWERPC_LINUX): Likewise.
22853 (PTRDIFF_TYPE): Likewise.
22854 (RESTORE_FP_PREFIX): Likewise.
22855 (RESTORE_FP_SUFFIX): Likewise.
22856 (SAVE_FP_PREFIX): Likewise.
22857 (SAVE_FP_SUFFIX): Likewise.
22858 (SIZE_TYPE): Likewise.
22859 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
22860 (TARGET_64BIT): Likewise.
22861 (TARGET_64BIT): Likewise.
22862 (TARGET_AIX): Likewise.
22863 (WCHAR_TYPE_SIZE): Likewise.
22864 (WCHAR_TYPE): Undefine.
22865 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
22866 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
22867 (CPP_OS_RTEMS_SPEC): Delete.
22868 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
22869 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
22871 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
22873 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22875 PR tree-optimization/81578
22876 * tree-parloops.c (build_new_reduction): Bail out if
22877 reduction_code isn't one of the standard OpenMP reductions.
22878 Move the details printing after that decision.
22880 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22882 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22883 related to reload_in_progress.
22884 (splat_input_operand): Likewise.
22885 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22887 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22889 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22890 (TARGET_INSTANTIATE_DECLS): Likewise.
22891 (legitimate_indexed_address_p): Delete reload_in_progress code.
22892 (rs6000_debug_legitimate_address_p): Likewise.
22893 (rs6000_eliminate_indexed_memrefs): Likewise.
22894 (rs6000_emit_le_vsx_store): Likewise.
22895 (rs6000_emit_move_si_sf_subreg): Likewise.
22896 (rs6000_emit_move): Likewise.
22897 (register_to_reg_type): Likewise.
22898 (rs6000_pre_atomic_barrier): Likewise.
22899 (rs6000_machopic_legitimize_pic_address): Likewise.
22900 (rs6000_allocate_stack_temp): Likewise.
22901 (rs6000_address_for_fpconvert): Likewise.
22902 (rs6000_address_for_altivec): Likewise.
22903 (rs6000_secondary_memory_needed_rtx): Delete function.
22904 (rs6000_check_sdmode): Likewise.
22905 (rs6000_alloc_sdmode_stack_slot): Likewise.
22906 (rs6000_instantiate_decls): Likewise.
22907 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22908 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22909 Delete reload_in_progress.
22910 (*vec_reload_and_plus_<mptrsize>): Likewise.
22911 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22912 (vsx_div_v2di): Likewise.
22913 (vsx_udiv_v2di): Likewise.
22915 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22917 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22918 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22919 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22920 (rs6000_debug_reg_global): Delete print of LRA status.
22921 (rs6000_option_override_internal): Delete dead LRA related code.
22922 (rs6000_lra_p): Delete function.
22923 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22925 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22927 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22928 * config/riscv/rtems.h: New file.
22930 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22931 Sudakshina Das <sudi.das@arm.com>
22933 * config/aarch64/aarch64.md
22934 (define_split for and<mode>3nr_compare): Move
22935 non aarch64_logical_operand to a register.
22936 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22937 register immediate operand to a register.
22938 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22940 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22942 PR middle-end/81564
22943 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22945 2017-07-27 Richard Biener <rguenther@suse.de>
22947 PR tree-optimization/81573
22948 PR tree-optimization/81494
22949 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22950 multi defuse cycle case.
22952 2017-07-27 Richard Biener <rguenther@suse.de>
22954 PR tree-optimization/81571
22955 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22958 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22960 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22961 earlier and only if MASK_FPU is set. Adjust formatting.
22963 2017-07-27 Martin Liska <mliska@suse.cz>
22965 * opt-functions.awk: Add validation of value of Init.
22966 * optc-gen.awk: Pass new argument.
22968 2017-07-27 Martin Liska <mliska@suse.cz>
22970 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22971 Fix wrong condition.
22973 2017-07-27 Martin Liska <mliska@suse.cz>
22975 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22976 BBs and edges seen by autoFDO.
22978 2017-07-27 Richard Biener <rguenther@suse.de>
22980 PR tree-optimization/81502
22981 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22982 with incompatible but same sized type.
22983 (execute_update_addresses_taken): Likewise.
22985 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22987 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22988 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22990 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22993 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22994 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22995 Change s_operand to memory_operand.
22997 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22999 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23000 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23001 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23002 Emit instructions rather than returning an expression. Handle TFmode
23003 and KFmode by casting to TImode.
23004 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23005 (rs6000_emit_le_vsx_store): Likewise.
23006 * config/rs6000/vsx.md (VSX_TI): New iterator.
23007 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23008 (*vsx_le_undo_permute_<mode>): Likewise.
23009 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23010 emit the split sequence.
23011 (*vsx_le_perm_store_<mode>): Likewise.
23013 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23015 PR tree-optimization/81555
23016 PR tree-optimization/81556
23017 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23018 if true, force CHANGED for the recursive invocation.
23019 (reassociate_bb): Remember original length of ops array, pass
23020 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23022 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23023 attributes for noipa attribute. For naked attribute use
23024 lookup_attribute first before lookup_attribute_spec.
23025 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23027 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23028 for functions with noipa attribute.
23029 (cgraph_externally_visible_p): Return true for functions with noipa
23031 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23032 for functions with noipa attribute.
23033 * doc/extend.texi: Document noipa function attribute.
23034 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23035 also for functions with noipa attribute.
23036 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23038 2017-07-26 Andrew Pinski <apinski@cavium.com>
23040 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23041 vec_unalign_load_cost and vec_unalign_store_cost.
23043 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23045 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23046 -mvsx-small-integer option.
23047 (ISA_3_0_MASKS_IEEE): Likewise.
23048 (OTHER_VSX_VECTOR_MASKS): Likewise.
23049 (POWERPC_MASKS): Likewise.
23050 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23051 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23052 code, only testing for DImode being allowed in non-VSX floating
23054 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23055 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23057 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23058 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23059 TARGET_P8_VECTOR test.
23060 (rs6000_secondary_reload_simple_move): Likewise.
23061 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23062 since TARGET_P9_VECTOR was already tested.
23063 (rs6000_opt_masks): Remove -mvsx-small-integer.
23064 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23065 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23067 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23068 test for TARGET_VEXTRACTUB was used, and that uses
23070 (p9 extract splitter): Likewise.
23071 (vsx_extract_<mode>_di_p9): Likewise.
23072 (vsx_extract_<mode>_store_p9): Likewise.
23073 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23074 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23075 the elimination of TARGET_VSX_SMALL_INTEGER.
23076 (vsx_extract_<mode>_p8): Likewise.
23077 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23078 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23079 (vsx_set_<mode>_p9): Likewise.
23080 (vsx_set_v4sf_p9): Likewise.
23081 (vsx_set_v4sf_p9_zero): Likewise.
23082 (vsx_insert_extract_v4sf_p9): Likewise.
23083 (vsx_insert_extract_v4sf_p9_2): Likewise.
23084 * config/rs6000/rs6000.md (sign extend splitter): Change
23085 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23086 (floatsi<mode>2_lfiwax_mem): Likewise.
23087 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23088 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23089 since a test for TARGET_P9_VECTOR was used.
23090 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23091 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23092 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23093 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23094 TARGET_P8_VECTOR test.
23095 (fix_trunc<mode>si2_stfiwx): Likewise.
23096 (fix_trunc<mode>si2_internal): Likewise.
23097 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23098 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23100 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23101 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23102 TARGET_P8_VECTOR test.
23103 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23104 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23105 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23107 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23108 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23109 since a test for TARGET_P9_VECTOR was used.
23110 (splitter for loading small constants): Likewise.
23112 2017-07-26 Andrew Pinski <apinski@cavium.com>
23114 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23117 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23120 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23121 (fp_valid_at): Likewise.
23123 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23125 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23126 (qdf24xx_addrcost_table): Likewise.
23127 (cortexa57_tunings): Update to use generic_branch_cost.
23128 (cortexa72_tunings): Likewise.
23129 (cortexa73_tunings): Likewise.
23130 (qdf24xx_tunings): Likewise.
23132 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23134 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23135 (thunderx2t99_branch_cost): Likewise.
23136 (cortexa35_tunings): Update to use generic_branch_cost.
23137 (cortexa53_tunings): Likewise.
23138 (cortexa57_tunings): Likewise.
23139 (cortexa72_tunings): Likewise.
23140 (cortexa73_tunings): Likewise.
23141 (thunderx2t99_tunings): Likewise.
23143 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23145 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23146 (sparc_option_override): Honour MASK_FSMULD.
23147 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23148 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23149 * config/sparc/sparc.opt (mfsmuld): New option.
23150 * doc/invoke.texi (mfsmuld): Document option.
23152 2017-07-26 Marek Polacek <polacek@redhat.com>
23154 PR middle-end/70992
23155 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23157 2017-07-26 Richard Biener <rguenther@suse.de>
23159 * gimple-match-head.c (do_valueize): Return OP if valueize
23161 (get_def): New helper to get at the def stmt of a SSA name
23162 if valueize allows.
23163 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23164 do_valueize to get at the def stmt.
23165 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23167 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23169 PR middle-end/46932
23170 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23172 2017-07-26 Martin Liska <mliska@suse.cz>
23175 * function.c (expand_function_start): Make expansion of
23176 nonlocal_goto_save_area after parm_birth_insn.
23178 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23180 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23181 from all CPU target flags enable members.
23183 2017-07-26 Richard Biener <rguenther@suse.de>
23185 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23186 (decision_tree::gen): Make 'type' const.
23187 (write_predicate): Likewise.
23189 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23191 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23192 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23193 (rs6000_option_override_internal): Likewise.
23194 (rs6000_expand_vector_set): Likewise.
23195 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23196 (TARGET_UPPER_REGS_SF): Likewise.
23197 (TARGET_UPPER_REGS_DI): Likewise.
23198 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23199 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23200 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23201 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23202 (Splitters for DI constants in Altivec registers): Likewise.
23203 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23204 (vsx_set_v4sf_p9): Likewise.
23205 (vsx_set_v4sf_p9_zero): Likewise.
23206 (vsx_insert_extract_v4sf_p9): Likewise.
23207 (vsx_insert_extract_v4sf_p9_2): Likewise.
23209 2017-07-25 Carl Love <cel@us.ibm.com>
23211 * doc/extend.texi: Update the built-in documentation file for the
23212 existing built-in functions
23213 vector signed char vec_cnttz (vector signed char);
23214 vector unsigned char vec_cnttz (vector unsigned char);
23215 vector signed short vec_cnttz (vector signed short);
23216 vector unsigned short vec_cnttz (vector unsigned short);
23217 vector signed int vec_cnttz (vector signed int);
23218 vector unsigned int vec_cnttz (vector unsigned int);
23219 vector signed long long vec_cnttz (vector signed long long);
23220 vector unsigned long long vec_cnttz (vector unsigned long long);
23222 2017-07-25 Andrew Pinski <apinski@cavium.com>
23224 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23225 accesses where the use is for the first operand of a BIT_INSERT.
23227 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23230 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23231 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23233 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23235 * config/i386/gstabs.h: Delete.
23236 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23238 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23240 * config/i386/i386.c (ix86_decompose_address): Do not check for
23241 register RTX when looking at index_reg or base_reg.
23242 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23244 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23246 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23247 to update EH info here.
23249 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23251 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23253 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23255 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23257 2017-07-25 Torsten Duwe <duwe@suse.de>
23259 * common.opt: Introduce -fpatchable-function-entry
23260 command line option, and its variables function_entry_patch_area_size
23261 and function_entry_patch_area_start.
23262 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23263 including a two-value parser.
23264 * target.def (print_patchable_function_entry): New target hook.
23265 * targhooks.h (default_print_patchable_function_entry): New function.
23266 * targhooks.c (default_print_patchable_function_entry): Likewise.
23267 * toplev.c (process_options): Switch off IPA-RA if
23268 patchable function entries are being generated.
23269 * varasm.c (assemble_start_function): Look at the
23270 patchable-function-entry command line switch and current
23271 function attributes and maybe generate NOP instructions by
23272 calling the print_patchable_function_entry hook.
23273 * doc/extend.texi: Document patchable_function_entry attribute.
23274 * doc/invoke.texi: Document -fpatchable_function_entry
23275 command line option.
23276 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23278 * doc/tm.texi: Re-generate.
23280 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23283 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23284 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23286 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23288 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23290 2017-07-25 Richard Biener <rguenther@suse.de>
23292 PR tree-optimization/81455
23293 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23294 not walk in cycles when looking for guards.
23296 2017-07-25 Richard Biener <rguenther@suse.de>
23298 PR tree-optimization/81529
23299 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23300 when optimizing backedge uses.
23302 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23304 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23306 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23307 to dl_section_ref. On AIX, append an expression to subtract
23308 the size of the section length to dl_section_ref.
23310 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23312 * configure.ac: If any of the config.* scripts fail, exit 1.
23313 * configure: Regenerate.
23315 2017-07-25 Richard Biener <rguenther@suse.de>
23317 PR middle-end/81546
23318 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23319 of immediate uses, be more verbose on errors.
23321 2017-07-25 Richard Biener <rguenther@suse.de>
23323 PR tree-optimization/81510
23324 * tree-vect-loop.c (vect_is_simple_reduction): When the
23325 reduction stmt is not inside the loop bail out.
23327 2017-07-25 Richard Biener <rguenther@suse.de>
23329 PR tree-optimization/81303
23330 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23331 profitability check against LOOP_VINFO_NITERSM1.
23333 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23335 * domwalk.c (cmp_bb_postorder): Simplify.
23336 (sort_bbs_postorder): New function. Use it...
23337 (dom_walker::walk): ...here to optimize common cases.
23339 2017-07-25 Martin Liska <mliska@suse.cz>
23342 * ipa-visibility.c (function_and_variable_visibility): Make the
23343 redirection just on target that supports aliasing.
23344 Fix GNU coding style.
23346 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23349 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23351 * config/arm/rtems.h: Likewise.
23352 * config/bfin/rtems.h: Likewise.
23353 * config/i386/rtemself.h: Likewise.
23354 * config/lm32/rtems.h: Likewise.
23355 * config/m32c/rtems.h: Likewise.
23356 * config/m68k/rtemself.h: Likewise.
23357 * config/microblaze/rtems.h: Likewise.
23358 * config/mips/rtems.h: Likewise.
23359 * config/moxie/rtems.h: Likewise.
23360 * config/nios2/rtems.h: Likewise.
23361 * config/powerpcspe/rtems.h: Likewise.
23362 * config/rs6000/rtems.h: Likewise.
23363 * config/rtems.h: Likewise.
23364 * config/sh/rtems.h: Likewise.
23365 * config/sh/rtemself.h: Likewise.
23366 * config/sparc/rtemself.h: Likewise.
23368 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23371 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23372 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23373 * tree-ssa-structalias.c (alias_get_name): Same.
23375 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23378 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23379 instructions if no du chain is found.
23381 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23383 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23385 2017-07-25 Richard Biener <rguenther@suse.de>
23387 PR middle-end/81505
23388 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23391 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23393 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23394 upper-regs options.
23395 (ISA_2_7_MASKS_SERVER): Likewise.
23396 (ISA_3_0_MASKS_IEEE): Likewise.
23397 (OTHER_P8_VECTOR_MASKS): Likewise.
23398 (OTHER_VSX_VECTOR_MASKS): Likewise.
23399 (POWERPC_MASKS): Likewise.
23400 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23401 duplicate list of options.
23402 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23403 explicit -mupper-regs options.
23404 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23405 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23406 alias for -mupper-regs-df.
23407 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23408 (rs6000_init_hard_regno_mode_ok): Likewise.
23409 (rs6000_option_override_internal): Likewise.
23410 (rs6000_opt_masks): Likewise.
23411 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23412 options in terms of whether -mvsx or -mpower8-vector was used.
23413 (TARGET_UPPER_REGS_DI): Likewise.
23414 (TARGET_UPPER_REGS_SF): Likewise.
23415 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23416 -mupper-regs-* options.
23418 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23420 * passes.c (emergency_dump_function): Print some empty lines and a
23421 header before the RTL dump.
23423 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23425 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23427 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23430 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23431 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23433 2017-07-24 Carl Love <cel@us.ibm.com>
23435 * config/rs6000/rs6000-c.c: Add support for built-in functions
23436 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23437 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23438 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23439 vec_extract_fp_from_shortl): Add defines for the two builtins.
23440 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23441 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23443 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23444 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23445 * doc/extend.texi: Update the built-in documentation file for the
23446 new built-in function.
23448 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23451 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23453 * doc/generic.texi: Likewise.
23454 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23455 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23457 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23459 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23460 (aarch64_mls_elt_merge<mode>): Likewise.
23462 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23464 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23465 having __cxa_atexit.
23467 2017-07-23 Michael Collison <michael.collison@arm.com>
23469 * config/arm/arm.c (arm_option_override): Deprecate
23470 use of -mstructure-size-boundary.
23471 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23472 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23474 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23477 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23478 Reduce cost estimate for direct moves.
23480 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23483 * config/i386/i386.c (ix86_option_override_internal): Disable
23484 BMI, BMI2 and TBM instructions for -m16.
23486 2017-07-21 Carl Love <cel@us.ibm.com>
23488 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23489 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23490 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23491 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23492 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23493 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23494 VMULOSW): New enum "unspec" values.
23495 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23496 altivec_vmulosw): New patterns.
23497 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23498 VMULOSW): Add definitions.
23500 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23502 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23503 (qdf24xx): Likewise.
23504 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23505 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23506 (AARCH64_FL_V8_1): Renumber.
23507 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23508 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23509 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23510 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23511 rdma to feature modifiers list.
23513 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23515 PR middle-end/56727
23516 * ipa-visibility (function_and_variable_visibility): Convert
23517 recursive PLT call to direct call if appropriate.
23519 2017-07-21 Andrew Pinski <apinski@cavium.com>
23521 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23522 operand 1 to see if the types precision matches.
23523 * fold-const.c (operand_equal_p): Likewise.
23525 2017-07-21 Richard Biener <rguenther@suse.de>
23527 PR tree-optimization/81303
23528 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23529 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23530 (vect_peeling_hash_get_lowest_cost): Adjust.
23531 (vect_enhance_data_refs_alignment): Likewise. Use
23532 vect_get_peeling_costs_all_drs to compute the penalty for no
23533 peeling to match up costs.
23535 2017-07-21 Richard Biener <rguenther@suse.de>
23537 PR tree-optimization/81500
23538 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23539 we didn't identify a reduction path.
23541 2017-07-21 Tom de Vries <tom@codesourcery.com>
23542 Cesar Philippidis <cesar@codesourcery.com>
23544 PR gcov-profile/81442
23545 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23548 2017-07-21 Tom de Vries <tom@codesourcery.com>
23551 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23553 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23554 nvptx_override_options_after_change.
23556 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23558 * dwarf2out.c (output_file_names): Avoid double testing for
23559 dwarf_version >= 5.
23561 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23563 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23565 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23567 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23569 (try_crossjump_to_edge): Do not punt on partitioned functions.
23571 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23573 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23574 Put all BBs reachable only via paths crossing cold region to cold
23576 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23578 2016-07-21 Richard Biener <rguenther@suse.de>
23580 PR tree-optimization/81303
23581 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23582 into account prologue and epilogue iterations when raising
23583 min_profitable_iters to sth at least covering one vector iteration.
23585 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23587 * config/arm/arm.c (arm_test_cpu_arch_dat):
23590 2017-07-20 Nathan Sidwell <nathan@acm.org>
23592 Remove TYPE_METHODS.
23593 * tree.h (TYPE_METHODS): Delete.
23594 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23595 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23596 (dbxout_type_methods): Scan TYPE_FIELDS.
23597 (dbxout_type): Don't check TYPE_METHODS here.
23598 * function.c (use_register_for_decl): Always ignore register for
23599 class types when not optimizing.
23600 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23601 * tree.c (free_lang_data_in_type): Stitch out member functions and
23602 templates from TYPE_FIELDS.
23603 (build_distinct_type_copy, verify_type_variant,
23604 verify_type): Member fns are on TYPE_FIELDS.
23605 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23606 * tree-pretty-print.c (dump_generic_node): Likewise.
23608 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23611 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23612 V2TImode and V4TImode.
23613 (ix86_expand_vector_extract): Likewise.
23614 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23615 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23616 (ssescalarmode): Handle V4TImode and V2TImode.
23617 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23618 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23619 (VEXTRACTI128_MODE): New mode iterator.
23620 (splitter for *vec_extractv?ti first element): New.
23621 (VEC_INIT_MODE): New mode iterator.
23622 (vec_init<mode>): Consolidate 3 expanders into one using
23623 VEC_INIT_MODE mode iterator.
23625 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23627 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23628 non_spilled_static_chain_regno_p.
23630 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23632 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23634 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23636 * bb-reorder.c (connect_traces): Allow copying of blocks within
23639 2017-07-20 Richard Biener <rguenther@suse.de>
23641 * gimple.h (gimple_phi_result): Add gphi * overload.
23642 (gimple_phi_result_ptr): Likewise.
23643 (gimple_phi_arg): Likewise. Adjust index assert to only
23644 allow actual argument accesses rather than all slots available
23646 (gimple_phi_arg_def): Add gphi * overload.
23647 * tree-phinodes.c (make_phi_node): Initialize only actual
23649 (resize_phi_node): Clear memory not covered by old node,
23650 do not initialize excess argument slots.
23651 (reserve_phi_args_for_new_edge): Initialize new argument slot
23654 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23656 PR tree-optimization/81388
23658 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23660 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23661 by removing computation of may_be_zero.
23663 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23664 Tom de Vries <tom@codesourcery.com>
23666 PR middle-end/81030
23667 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23668 when gimple level profile disagrees with what RTL expander did.
23670 2017-07-20 Richard Biener <rguenther@suse.de>
23672 PR tree-optimization/61171
23673 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23674 (vect_analyze_stmt): Add slp instance parameter.
23675 (vectorizable_reduction): Likewise.
23676 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23677 (vect_is_simple_reduction): Deal with chains not detected
23678 as SLP reduction chain, specifically not properly associated
23679 chains containing a mix of plus/minus.
23680 (get_reduction_op): Remove.
23681 (get_initial_defs_for_reduction): Simplify, pass in whether
23682 this is a reduction chain, pass in the SLP node for the PHIs.
23683 (vect_create_epilog_for_reduction): Get the SLP instance as
23685 (vectorizable_reduction): Get the SLP instance as arg.
23686 During analysis remember the SLP node with the PHIs in the
23687 instance. Simplify getting at the vectorized reduction PHIs.
23688 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23689 through SLP instance.
23690 (vect_slp_analyze_operations): Likewise.
23691 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23692 (vect_transform_stmt): Likewise.
23694 2017-07-20 Tom de Vries <tom@codesourcery.com>
23696 PR tree-optimization/81489
23697 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23698 read of phi arg location to before loop that modifies phi.
23700 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23702 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23705 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23707 PR middle-end/81331
23708 * except.c (execute): Fix ordering issue.
23710 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23712 PR rtl-optimization/81423
23713 * combine.c (make_compound_operation_int): Don't try to optimize
23714 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23716 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23718 PR rtl-optimization/81423
23719 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23720 with a constant that is -1 in the truncated to mode.
23722 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23724 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23725 (determine_unlikely_bbs): ... here.
23726 * predict.h (propagate_unlikely_bbs_forward): Declare.
23727 * cfgexpand.c (pass_expand::execute): Use it.
23728 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23730 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23731 propagate_unlikely_bbs_forward.
23733 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23735 PR middle-end/81331
23736 * except.c (maybe_add_nop_after_section_switch): New function.
23739 2017-07-19 Tom de Vries <tom@codesourcery.com>
23741 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23743 2017-07-19 Tom de Vries <tom@codesourcery.com>
23745 * gimple.h (gimple_phi_arg): Make assert more strict.
23747 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23749 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23750 * config/rs6000/mmintrin.h: New file.
23751 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23753 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23755 PR tree-optimization/81346
23756 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23758 2017-07-19 Tom de Vries <tom@codesourcery.com>
23760 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23762 2017-07-19 Tom de Vries <tom@codesourcery.com>
23764 * config/nvptx/nvptx-modes.def: Add V2DImode.
23765 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23766 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23767 (nvptx_output_mov_insn): Handle lack of mov.b128.
23768 (nvptx_print_operand): Handle 'H' and 'L' codes.
23769 (nvptx_vector_mode_supported): Allow V2DImode.
23770 (nvptx_preferred_simd_mode): New function.
23771 (nvptx_data_alignment): New function.
23772 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23773 nvptx_preferred_simd_mode.
23774 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23776 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23778 2017-07-19 Tom de Vries <tom@codesourcery.com>
23780 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23781 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23782 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23783 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23784 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23785 (mov<VECIM>_insn): New define_insn.
23786 (define_expand "mov<VECIM>): New define_expand.
23788 2017-07-19 Tom de Vries <tom@codesourcery.com>
23790 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
23792 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23794 PR tree-optimization/81346
23795 * fold-const.h (fold_div_compare, range_check_type): Declare.
23796 * fold-const.c (range_check_type): New function.
23797 (build_range_check): Use range_check_type.
23798 (fold_div_compare): No longer static, rewritten into
23799 a match.pd helper function.
23800 (fold_comparison): Don't call fold_div_compare here.
23801 * match.pd (X / C1 op C2): New optimization using fold_div_compare
23802 as helper function.
23804 2017-07-19 Nathan Sidwell <nathan@acm.org>
23806 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
23807 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
23808 * tree.c (find_decls_types_r, verify_type): Use
23809 TYPE_{MIN,MAX}_VALUE_RAW.
23810 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
23811 (hash_tree): Likewise.
23812 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
23814 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
23817 2017-07-18 Tom de Vries <tom@codesourcery.com>
23819 PR middle-end/81464
23820 * omp-expand.c (expand_omp_for_static_chunk): Handle
23821 equal-argument loop exit phi.
23823 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23826 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23827 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23828 operand 2 predicate.
23829 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23830 operand 2 predicate.
23831 (ror,rol -> rorx splitters): Use const_int_operand as
23832 operand 2 predicate.
23834 2017-06-18 Richard Biener <rguenther@suse.de>
23836 PR tree-optimization/81410
23837 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23838 the gap in the ! slp_perm SLP case after each group.
23840 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23842 PR middle-end/81463
23843 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
23846 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23848 PR middle-end/81462
23849 * predict.c (set_even_probabilities): Cleanup; do not affect
23850 probabilities that are already known.
23851 (combine_predictions_for_bb): Call even when count is set.
23853 2017-07-18 Nathan Sidwell <nathan@acm.org>
23855 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
23858 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23861 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
23862 optimization for loop niter analysis.
23864 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
23867 * config/avr/avr.c (avr_optimize_casesi): Don't use
23868 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
23870 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
23872 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
23873 body_cost_vec from _vect_peel_extended_info.
23874 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23875 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23878 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23880 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23882 2017-07-18 Richard Biener <rguenther@suse.de>
23884 PR tree-optimization/80620
23885 PR tree-optimization/81403
23886 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23887 info when re-using a VN table entry.
23889 2017-07-18 Richard Biener <rguenther@suse.de>
23891 PR tree-optimization/81418
23892 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23893 vectype_in. Verify that with lane-reducing reduction operations
23894 we have a single def-use cycle.
23896 2017-07-17 Carl Love <cel@us.ibm.com>
23898 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23900 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23901 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23902 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23903 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23904 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23905 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23906 VMULOSW): New enum "unspec" values.
23907 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23908 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23909 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23910 altivec_vmulosw): New patterns.
23911 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23912 VMULOSW): Add definitions.
23914 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23916 * config/alpha/alpha.c: Include predict.h.
23918 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23920 * tree-vrp.c (compare_assert_loc): Fix comparison function
23921 to return predictable results.
23923 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23925 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23927 (subdi3): Likewise.
23928 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23929 * doc/invoke.texi (mexpand-adddi): Update text.
23931 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23933 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23934 that also clobbers the CC register. The old expand code is moved
23936 (*arc_clzsi2): ... here.
23937 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23938 the CC register. The old expand code is moved to ...
23939 (arc_ctzsi2): ... here.
23941 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23943 * config/arc/arc.opt (mindexed-loads): Use initial value
23944 TARGET_INDEXED_LOADS_DEFAULT.
23945 (mauto-modify-reg): Use initial value
23946 TARGET_AUTO_MODIFY_REG_DEFAULT.
23947 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23948 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23949 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23950 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23952 2017-07-17 Martin Liska <mliska@suse.cz>
23955 * opts.c (finish_options): Do not allow -fgnu-tm
23956 w/ -fsanitize={kernel-,}address. Say sorry.
23958 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23961 * tree-loop-distribution.c (classify_partition): Only assert on
23962 numer of iterations.
23963 (merge_dep_scc_partitions): Delete prameter. Update function call.
23964 (distribute_loop): Remove code handling loop with unknown niters.
23965 (pass_loop_distribution::execute): Skip loop with unknown niters.
23967 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23970 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23971 function sort_partitions_by_post_order.
23973 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23975 PR tree-optimization/81374
23976 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23977 the max index of basic blocks, rather than number of basic blocks.
23979 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23981 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23983 (arc_legitimate_pic_operand_p): Likewise.
23984 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23986 (arc_needs_pcl_p): Likewise.
23987 (arc_legitimate_pc_offset_p): Likewise.
23988 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23989 function is also used in constrains.md.
23990 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23991 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23992 PLUS. Only return true/false in known cases, otherwise assert.
23993 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23994 is already called in arc_legitimate_constant_p.
23995 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23997 (LEGITIMATE_PIC_OPERAND_P): Use
23998 arc_raw_symbolic_reference_mentioned_p function.
23999 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24004 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24005 Andrew Burgess <andrew.burgess@embecosm.com>
24007 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24008 (arc_return_address_register): New function.
24009 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24010 (arc_handle_fndecl_attribute): Add naked attribute.
24011 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24012 (TARGET_WARN_FUNC_RETURN): Likewise.
24013 (arc_allocate_stack_slots_for_args): New function.
24014 (arc_warn_func_return): Likewise.
24015 (machine_function): Change type fn_type.
24016 (arc_compute_function_type): Consider new naked function type,
24017 change function return type.
24018 (arc_must_save_register): Adapt to handle new
24019 arc_compute_function_type's return type.
24020 (arc_expand_prologue): Likewise.
24021 (arc_expand_epilogue): Likewise.
24022 (arc_return_address_regs): Delete.
24023 (arc_return_address_register): New function.
24024 (arc_epilogue_uses): Use above function.
24025 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24026 (arc_function_type): Change encoding, add naked type.
24027 (ARC_INTERRUPT_P): Change to handle the new encoding.
24028 (ARC_FAST_INTERRUPT_P): Likewise.
24029 (ARC_NORMAL_P): Define.
24030 (ARC_NAKED_P): Likewise.
24031 (arc_compute_function_type): Delete prototype.
24032 * config/arc/arc.md (in_ret_delay_slot): Use
24033 arc_return_address_register function.
24034 (simple_return): Likewise.
24035 (p_return_i): Likewise.
24037 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24039 PR tree-optimization/81428
24040 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24041 can't be built for those types.
24043 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24045 Remove stuff dead since r239246.
24047 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24048 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24049 (avr_inform_devices): Remove dead stuff.
24051 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24053 * config/arm/arm_neon.h: Fix softp typo.
24055 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24057 PR tree-optimization/81365
24058 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24059 aggregate moves onto bb predecessor edges, make sure there are no
24060 loads that could alias the lhs in between the start of bb and the
24063 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24066 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24067 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24068 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24070 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24072 PR tree-optimization/81396
24073 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24074 (init_symbolic_number): Initialize it to 1.
24075 (perform_symbolic_merge): Add n_ops from both operands into the new
24077 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24078 without base_addr as useless if they need more than one operation.
24079 (bswap_replace): Handle !bswap case for NULL base_addr.
24081 2017-07-17 Tom de Vries <tom@codesourcery.com>
24084 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24087 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24089 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24090 conditional builtin define __FIX_LEON3FT_B2BST.
24092 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24094 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24095 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24098 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24100 PR rtl-optimization/81424
24101 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24102 to remove potential trapping from operands if -fnon-call-exceptions.
24104 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24106 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24107 profile_proability for scalling.
24108 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24110 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24112 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24114 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24116 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24117 fixpoint arithmetics.
24119 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24121 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24122 fixpoint arithmetics.
24124 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24126 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24127 fixpoint arithmetics.
24129 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24131 * profile-count.h (profile_probability::from_reg_br_prob_note,
24132 profile_probability::to_reg_br_prob_note): New functions.
24133 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24134 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24135 * predict.c (probability_reliable_p): Update.
24136 (edge_probability_reliable_p): Update.
24137 (br_prob_note_reliable_p): Update.
24138 (invert_br_probabilities): Update.
24139 (add_reg_br_prob_note): New function.
24140 (combine_predictions_for_insn): Update.
24141 * asan.c (asan_clear_shadow): Update.
24142 * cfgbuild.c (compute_outgoing_frequencies): Update.
24143 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24144 (update_br_prob_note): Update.
24145 (rtl_verify_edges): Update.
24146 (purge_dead_edges): Update.
24147 (fixup_reorder_chain): Update.
24148 * emit-rtl.c (try_split): Update.
24149 * ifcvt.c (cond_exec_process_insns): Update.
24150 (cond_exec_process_if_block): Update.
24151 (dead_or_predicable): Update.
24152 * internal-fn.c (expand_addsub_overflow): Update.
24153 (expand_neg_overflow): Update.
24154 (expand_mul_overflow): Update.
24155 * loop-doloop.c (doloop_modify): Update.
24156 * loop-unroll.c (compare_and_jump_seq): Update.
24157 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24158 * predict.h: Update.
24159 * reorg.c (mostly_true_jump): Update.
24161 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24162 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24163 * config/arc/arc.c: (emit_unlikely_jump): Update.
24164 * config/arm/arm.c: (emit_unlikely_jump): Update.
24165 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24166 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24167 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24168 (ix86_print_operand): Update.
24169 (ix86_split_fp_branch): Update.
24170 (predict_jump): Update.
24171 * config/ia64/ia64.c (ia64_print_operand): Update.
24172 * config/mmix/mmix.c (mmix_print_operand): Update.
24173 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24174 (rs6000_expand_split_stack_prologue): Update.
24175 * config/rs6000/rs6000.c: Update.
24176 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24177 (s390_expand_vec_movstr): Update.
24178 (s390_expand_cs_tdsi): Update.
24179 (s390_expand_split_stack_prologue): Update.
24180 * config/sh/sh.c (sh_print_operand): Update.
24181 (expand_cbranchsi4): Update.
24182 (expand_cbranchdi4): Update.
24183 * config/sparc/sparc.c (output_v9branch): Update.
24184 * config/spu/spu.c (get_branch_target): Update.
24185 (ea_load_store_inline): Update.
24186 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24187 * config/tilepro/tilepro.c: Update.
24189 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24191 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24192 (gimplify_save_expr): Likewise.
24194 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24196 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24198 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24200 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24202 * timevar.def (TV_IPA_FNSUMMARY): Define.
24204 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24206 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24207 to back store errata sensitive sequence from being generated.
24208 (sqrtdf2_fix): Likewise.
24210 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24212 * tree-ssa-threadupdate.c (compute_path_counts,
24213 update_joiner_offpath_counts): Use profile_probability.
24215 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24218 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24220 * config/arm/arm-c.c (arm_cpu_builtins): Define
24221 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24223 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24225 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24226 array entries to represent __ieee128 versions of the
24227 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24228 scalar_extract_sig, and scalar_insert_exp built-in functions.
24229 (altivec_resolve_overloaded_builtin): Add special case handling
24230 for the __builtin_scalar_insert_exp function, as represented by
24231 the P9V_BUILTIN_VEC_VSIEDP constant.
24232 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24233 exponent support for __ieee128 argument.
24234 (VSESQP): Add scalar extract signature support for __ieee128
24236 (VSTDCNQP): Add scalar test negative support for __ieee128
24238 (VSIEQP): Add scalar insert exponent support for __int128 argument
24239 with __ieee128 result.
24240 (VSIEQPF): Add scalar insert exponent support for __ieee128
24241 argument with __ieee128 result.
24242 (VSTDCQP): Add scalar test data class support for __ieee128
24244 (VSTDCNQP): Add overload support for scalar test negative with
24245 __ieee128 argument.
24246 (VSTDCQP): Add overload support for scalar test data class
24247 __ieee128 argument.
24248 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24249 UNSPEC_VSX_SXSIGDP.
24250 (UNSPEC_VSX_SIEXPQP): New constant.
24251 (xsxexpqp): New insn for VSX scalar extract exponent quad
24253 (xsxsigqp): New insn for VSX scalar extract significand quad
24255 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24256 precision with floating point argument.
24257 (xststdcqp): New expand for VSX scalar test data class quad
24259 (xststdcnegqp): New expand for VSX scalar test negative quad
24261 (xststdcqp): New insn to match expansions for VSX scalar test data
24262 class quad precision and VSX scalar test negative quad precision.
24263 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24264 special case operand checking to enforce that second operand of
24265 VSX scalar test data class with quad precision argument is a 7-bit
24267 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24268 prototypes and descriptions of __ieee128 versions of
24269 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24270 scalar_test_data_class, and scalar_test_neg built-in functions.
24272 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24274 PR tree-optimization/81162
24275 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24276 replace a negate with an add.
24278 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24280 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24282 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24284 * config/arm/arm-c.c (arm_cpu_builtins): Define
24285 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24287 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24289 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24290 (armv8-r): Set ARM Cortex-R52 as default CPU.
24291 * config/arm/arm-tables.opt: Regenerate.
24292 * config/arm/arm-tune.md: Regenerate.
24293 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24295 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24296 extension for -mcpu=cortex-r52.
24298 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24300 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24301 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24302 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24303 (fp-armv8): Define it as FP_ARMv8 only.
24304 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24305 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24307 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24308 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24309 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24310 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24311 than TARGET_FPU_ARMV8.
24312 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24313 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24314 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24316 * config/arm/neon.md (neon_vrint): Likewise.
24317 (neon_vcvt): Likewise.
24318 (neon_<fmaxmin_op><mode>): Likewise.
24319 (<fmaxmin><mode>3): Likewise.
24320 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24321 * config/arm/predicates.md (arm_cond_move_operator): Check against
24322 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24324 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24326 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24327 to top of function.
24329 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24331 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24332 loop in comment with memset.
24334 2017-07-14 Martin Liska <mliska@suse.cz>
24336 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24337 * dwarf2out.c (is_java): Remove the function.
24338 (output_pubname): Remove usage of the function.
24339 (lower_bound_default): Remove usage of DW_LANG_Java.
24340 (gen_compile_unit_die): Likewise.
24341 * gcc.c: Remove compiler defaults for .java and .zip files.
24342 * gimple-expr.c (remove_suffix): Change as there's no longer
24343 extension than 4-letter one.
24344 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24345 (gimplify_save_expr): Likewise.
24346 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24347 as it's possible even for other languages than Java.
24348 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24349 * lto-opts.c (lto_write_options): Remove reference to Java.
24350 * opts.c (strip_off_ending): Update file extension handling.
24351 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24352 * tree-eh.c (lower_resx): Likewise.
24353 * tree.c (free_lang_data_in_type): Remove dead code.
24354 (find_decls_types_r): Likewise.
24355 (build_common_builtin_nodes): Remove Java from a comment.
24356 (verify_type): Remove dead code.
24357 * varasm.c (assemble_external): Remove Java from a comment.
24359 2017-07-14 Martin Liska <mliska@suse.cz>
24361 * opts.c (finish_options): Add quotes.
24362 (common_handle_option): Likewise.
24364 2017-07-14 Martin Liska <mliska@suse.cz>
24366 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24367 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24368 Remove N_SO_PASCAL.
24369 * dwarf2out.c (lower_bound_default): Do not handle
24371 (gen_compile_unit_die): Likewise.
24372 * gcc.c: Remove default extension binding for GNU Pascal.
24373 * stmt.c: Remove Pascal language from a comment.
24374 * xcoffout.c: Likewise.
24376 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24379 * diagnostic-show-locus.c (fixit_cmp): New function.
24380 (layout::layout): Sort m_fixit_hints.
24381 (column_range::column_range): Assert that the values are valid.
24382 (struct char_span): New struct.
24383 (correction::overwrite): New method.
24384 (struct source_line): New struct.
24385 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24386 calls in terms of classes source_line and char_span, and
24387 correction::overwrite.
24388 (selftest::test_overlapped_fixit_printing_2): New function.
24389 (selftest::diagnostic_show_locus_c_tests): Call it.
24391 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24393 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24394 early if there is no lhs.
24396 2017-07-13 Martin Liska <mliska@suse.cz>
24398 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24399 (gen_reference_type_die): Likewise.
24400 * stor-layout.c: Remove Pascal-related comment.
24402 2017-07-13 Martin Liska <mliska@suse.cz>
24404 * opts.c (finish_options): Add quotes to error messages.
24405 (parse_sanitizer_options): Likewise.
24407 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24409 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24411 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24413 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24415 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24417 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24419 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24421 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24424 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24425 provides the hardware capability bits, define the macro
24426 __BUILTIN_CPU_SUPPORTS__.
24427 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24428 if GLIBC does not provide the hardware capability bits. Add a
24429 gcc_unreachable call if the built-in cpu function is neither
24430 __builtin_cpu_is nor __builtin_cpu_supports.
24431 (rs6000_get_function_versions_dispatcher): Change the warning
24432 that an old GLIBC is used which does not export the capability
24433 bits to be an error.
24434 * doc/extend.texi (target_clones attribute): Document the
24435 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24436 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24437 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24438 the macros defined by GCC if the newer GLIBC is available.
24440 2017-07-12 Jeff Law <law@redhat.com>
24442 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24443 remaining includes slightly.
24444 * config/riscv/riscv-builtins.c: Include profile-count.h.
24446 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24449 * config/avr/avr.c (avr_set_current_function): In diagnostic
24450 messages: Quote keywords and (parts of) identifiers.
24451 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24454 2017-07-12 Carl Love <cel@us.ibm.com>
24456 * config/rs6000/rs6000-c.c: Add support for built-in functions
24457 vector bool char vec_revb (vector bool char);
24458 vector bool short vec_revb (vector short char);
24459 vector bool int vec_revb (vector bool int);
24460 vector bool long long vec_revb (vector bool long long);
24461 * doc/extend.texi: Update the built-in documentation file for the
24462 new built-in functions.
24464 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24466 * config/s390/s390.md: Remove movcc splitter.
24468 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24470 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24471 load/store on condition.
24473 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24476 * config/avr/avr.c (avr_encode_section_info)
24477 [progmem && !TREE_READONLY]: Error if progmem object needs
24480 2017-07-11 Michael Collison <michael.collison@arm.com>
24482 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24485 2017-07-11 Carl Love <cel@us.ibm.com>
24487 * config/rs6000/rs6000-c.c: Add support for builtins
24488 vector unsigned int vec_parity_lsbb (vector signed int);
24489 vector unsigned int vec_parity_lsbb (vector unsigned int);
24490 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24491 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24492 vector unsigned long long vec_parity_lsbb (vector signed long long);
24493 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24494 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24495 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24496 * doc/extend.texi: Update the built-in documentation file for the
24497 new built-in functions.
24499 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24501 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24502 (layout::m_primary_loc): New field.
24503 (layout::layout): Initialize new field. Move location filtering
24504 logic from here to...
24505 (layout::maybe_add_location_range): ...this new method. Add
24506 support for filtering to just the lines already specified by other
24508 (layout::will_show_line_p): New method.
24509 (gcc_rich_location::add_location_if_nearby): New method.
24510 (selftest::test_add_location_if_nearby): New test function.
24511 (selftest::diagnostic_show_locus_c_tests): Call it.
24512 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24515 2017-07-11 Tom de Vries <tom@codesourcery.com>
24517 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24518 (bb_first_real_insn): New function.
24519 (nvptx_single): Add extra initialization of broadcasted condition
24522 2017-07-11 Nathan Sidwell <nathan@acm.org>
24524 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24526 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24528 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24529 Binutils doc as TEXI will mess them up.
24530 * doc/invoke.texi (AVR Options): Same here.
24532 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24534 * config/sparc/sparc.opt (mfix-ut700): New option.
24535 (mfix-gr712rc): Likewise.
24536 (sparc_fix_b2bst): New variable.
24537 * doc/invoke.texi (SPARC options): Document them.
24538 (ARM options): Fix warnings.
24539 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24540 instructions to prevent sequences that can trigger the store-store
24541 errata for certain LEON3FT processors.
24542 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24543 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24544 * config/sparc/sparc.md (fix_b2bst): New attribute.
24545 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24547 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24550 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24552 (*rsqrtsf2_sse): Ditto.
24554 (div<mode>3): Macroize insn from divdf3 and divsf3
24555 using MODEF mode iterator.
24557 2017-07-10 Martin Sebor <msebor@redhat.com>
24559 PR tree-optimization/80397
24560 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24561 instead of testing for equality to INTEGER_TYPE.
24563 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24565 * config.gcc: Remove uclibc from arc target spec.
24567 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24569 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24571 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24574 * lto-wrapper.c (remove_option): New function.
24575 (merge_and_complain): Merge PIC/PIE options more realistically.
24577 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24579 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24583 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24584 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24585 * config.in: Regenerate.
24586 * configure: Regenerate.
24587 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24588 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24589 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24590 (TARGET_GASISR_PROLOGUES): ...target mask.
24591 * common/config/avr/avr-common.c
24592 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24593 Set -mgas-isr-prologues.
24594 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24595 INSERT_PASS_BEFORE for it.
24596 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24597 * config/avr/avr.c (avr_option_override)
24598 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24599 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24600 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24601 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24602 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24604 (make_avr_pass_pre_proep): New function.
24605 (emit_push_sfr) <treg>: Add argument to function and use it
24606 instead of TMP_REG.
24607 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24608 and set machine->gasisr.yes.
24609 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24610 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24611 __gcc_isr.n_pushed to .L__stack_usage.
24612 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24613 (avr_asm_final_postscan_insn): ...this new static function.
24614 * config/avr/avr.h (machine_function)
24615 <is_no_gccisr, use_L__stack_usage>: New fields.
24616 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24617 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24618 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24619 (gasisr, *gasisr): New expander and insn.
24620 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24621 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24622 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24624 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24626 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24629 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24631 Move jump-tables out of .text again.
24634 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24635 (ASM_OUTPUT_ADDR_VEC): New function.
24636 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24637 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24638 INSN_ADDRESSes as asm comment.
24639 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24640 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24641 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24642 * config/avr/avr.md (*tablejump): Adjust comment.
24643 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24644 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24646 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24647 (avr_output_addr_vec): New proto.
24648 (avr_log_t) <insn_addresses>: New field.
24650 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24653 * config/i386/i386.c (ix86_function_arg_advance): Set
24654 outgoing_args_on_stack to true if there are outgoing arguments
24656 (ix86_function_arg): Likewise.
24657 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24658 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24659 * config/i386/i386.h (machine_function): Add
24660 outgoing_args_on_stack.
24662 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24664 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24665 supporting pthreds.
24666 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24668 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24670 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24671 (REAL_H): Remove $(MACHMODE_H).
24672 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24674 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24675 $(MACHMODE_H) and double-int.h.
24676 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24678 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24681 2017-07-07 Andrew Pinski <apinski@cavium.com>
24683 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24684 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24686 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24688 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24689 Add warning if GCC was not configured to link against a GLIBC that
24690 exports the hardware capability bits.
24691 (make_resolver_func): Make resolver function private and not a
24692 COMDAT function. Create the name with clone_function_name instead
24693 of make_unique_name.
24696 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24697 correct operand in doing the split.
24699 2017-07-07 Carl Love <cel@us.ibm.com>
24701 * config/rs6000/rs6000-c: Add support for built-in function
24702 vector unsigned short vec_pack_to_short_fp32 (vector float,
24704 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24705 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24706 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24707 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24708 (convert_4f32_8i16): Add define_expand.
24709 * doc/extend.texi: Update the built-in documentation file for the
24710 new built-in function.
24712 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24714 * config/sparc/m8.md: New file.
24715 * config/sparc/sparc.md: Include m8.md.
24717 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24719 * config/sparc/sparc.opt: New option -mvis4b.
24720 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24721 (sparc_option_override): Handle VIS4B.
24722 (enum sparc_builtins): Define
24723 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24724 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24725 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24726 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24727 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24728 (check_constant_argument): New function.
24729 (sparc_vis_init_builtins): Define builtins
24730 __builtin_vis_dictunpack{8,16,32},
24731 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24732 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24733 __builtin_vis_fpcmpde{8,16,32}shl and
24734 __builtin_vis_fpcmpur{8,16,32}shl.
24735 (sparc_expand_builtin): Check that the constant operands to
24736 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24737 constant and in range.
24738 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24740 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24741 (SPARC_IMM5_P): Likewise.
24742 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24743 (enabled): Handle vis4b.
24744 (UNSPEC_DICTUNPACK): New unspec.
24745 (UNSPEC_FPCMPSHL): Likewise.
24746 (UNSPEC_FPUCMPSHL): Likewise.
24747 (UNSPEC_FPCMPDESHL): Likewise.
24748 (UNSPEC_FPCMPURSHL): Likewise.
24749 (cpu_feature): New CPU feature `vis4b'.
24750 (dictunpack{8,16,32}): New insns.
24751 (FPCSMODE): New mode iterator.
24752 (fpcscond): New code iterator.
24753 (fpcsucond): Likewise.
24754 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24755 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24756 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24757 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24758 * config/sparc/constraints.md: Define constraints `q' for unsigned
24759 2-bit integer constants and `t' for unsigned 5-bit integer
24761 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24763 (imm5_operand_dictunpack16): Likewise.
24764 (imm5_operand_dictunpack32): Likewise.
24765 (imm2_operand): Likewise.
24766 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24767 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24768 ditunpack* and fpcmp*shl builtins.
24770 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24772 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24773 * config.in: Add HAVE_AS_SPARC6 define.
24774 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24776 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24778 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24779 (CPP_CPU_SPEC): Handle m8.
24780 (ASM_CPU_SPEC): Likewise.
24781 * config/sparc/sparc-opts.h (enum processor_type): Add
24783 * config/sparc/sparc.c (m8_costs): New struct.
24784 (sparc_option_override): Handle TARGET_CPU_m8.
24785 (sparc32_initialize_trampoline): Likewise.
24786 (sparc64_initialize_trampoline): Likewise.
24787 (sparc_issue_rate): Likewise.
24788 (sparc_register_move_cost): Likewise.
24789 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
24790 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
24791 (ASM_CPU64_DEFAULT_SPEC): Likewise.
24792 (CPP_CPU_SPEC): Handle M8.
24793 (ASM_CPU_SPEC): Likewise.
24794 (AS_M8_FLAG): Define.
24795 * config/sparc/sparc.md: Add m8 to the cpu attribute.
24796 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
24797 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
24799 * configure: Regenerate.
24800 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
24803 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24805 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
24807 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
24808 ("*movdi_insn_sp32"): Do not set v3pipe.
24809 ("*movsi_insn"): Likewise.
24810 ("*movdi_insn_sp64"): Likewise.
24811 ("*movsf_insn"): Likewise.
24812 ("*movdf_insn_sp32"): Likewise.
24813 ("*movdf_insn_sp64"): Likewise.
24814 ("*zero_extendsidi2_insn_sp64"): Likewise.
24815 ("*sign_extendsidi2_insn"): Likewise.
24816 ("*mov<VM32:mode>_insn"): Likewise.
24817 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24818 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24819 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24820 ("<vlop:code><VL:mode>3"): Likewise.
24821 ("*not_<vlop:code><VL:mode>3"): Likewise.
24822 ("*nand<VL:mode>_vis"): Likewise.
24823 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24824 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24825 ("one_cmpl<VL:mode>2"): Likewise.
24826 ("faligndata<VM64:mode>_vis"): Likewise.
24827 ("alignaddrsi_vis"): Likewise.
24828 ("alignaddrdi_vis"): Likweise.
24829 ("alignaddrlsi_vis"): Likewise.
24830 ("alignaddrldi_vis"): Likewise.
24831 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24832 ("bmaskdi_vis"): Likewise.
24833 ("bmasksi_vis"): Likewise.
24834 ("bshuffle<VM64:mode>_vis"): Likewise.
24835 ("cmask8<P:mode>_vis"): Likewise.
24836 ("cmask16<P:mode>_vis"): Likewise.
24837 ("cmask32<P:mode>_vis"): Likewise.
24838 ("pdistn<P:mode>_vis"): Likewise.
24839 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24841 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24843 * config/sparc/sparc.md ("subtype"): New insn attribute.
24844 ("*wrgsr_sp64"): Set insn subtype.
24845 ("*rdgsr_sp64"): Likewise.
24846 ("alignaddrsi_vis"): Likewise.
24847 ("alignaddrdi_vis"): Likewise.
24848 ("alignaddrlsi_vis"): Likewise.
24849 ("alignaddrldi_vis"): Likewise.
24850 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24851 ("fexpand_vis"): Likewise.
24852 ("fpmerge_vis"): Likewise.
24853 ("faligndata<VM64:mode>_vis"): Likewise.
24854 ("bshuffle<VM64:mode>_vis"): Likewise.
24855 ("cmask8<P:mode>_vis"): Likewise.
24856 ("cmask16<P:mode>_vis"): Likewise.
24857 ("cmask32<P:mode>_vis"): Likewise.
24858 ("fchksm16_vis"): Likewise.
24859 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
24860 ("fmean16_vis"): Likewise.
24861 ("fp<plusminus_insn>64_vis"): Likewise.
24862 ("<plusminus_insn>v8qi3"): Likewise.
24863 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24864 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
24865 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
24866 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
24867 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
24868 ("*movqi_insn"): Likewise.
24869 ("*movhi_insn"): Likewise.
24870 ("*movsi_insn"): Likewise.
24871 ("movsi_pic_gotdata_op"): Likewise.
24872 ("*movdi_insn_sp32"): Likewise.
24873 ("*movdi_insn_sp64"): Likewise.
24874 ("movdi_pic_gotdata_op"): Likewise.
24875 ("*movsf_insn"): Likewise.
24876 ("*movdf_insn_sp32"): Likewise.
24877 ("*movdf_insn_sp64"): Likewise.
24878 ("*zero_extendhisi2_insn"): Likewise.
24879 ("*zero_extendqihi2_insn"): Likewise.
24880 ("*zero_extendqisi2_insn"): Likewise.
24881 ("*zero_extendqidi2_insn"): Likewise.
24882 ("*zero_extendhidi2_insn"): Likewise.
24883 ("*zero_extendsidi2_insn_sp64"): Likewise.
24884 ("ldfsr"): Likewise.
24885 ("prefetch_64"): Likewise.
24886 ("prefetch_32"): Likewise.
24887 ("tie_ld32"): Likewise.
24888 ("tie_ld64"): Likewise.
24889 ("*tldo_ldub_sp32"): Likewise.
24890 ("*tldo_ldub1_sp32"): Likewise.
24891 ("*tldo_ldub2_sp32"): Likewise.
24892 ("*tldo_ldub_sp64"): Likewise.
24893 ("*tldo_ldub1_sp64"): Likewise.
24894 ("*tldo_ldub2_sp64"): Likewise.
24895 ("*tldo_ldub3_sp64"): Likewise.
24896 ("*tldo_lduh_sp32"): Likewise.
24897 ("*tldo_lduh1_sp32"): Likewise.
24898 ("*tldo_lduh_sp64"): Likewise.
24899 ("*tldo_lduh1_sp64"): Likewise.
24900 ("*tldo_lduh2_sp64"): Likewise.
24901 ("*tldo_lduw_sp32"): Likewise.
24902 ("*tldo_lduw_sp64"): Likewise.
24903 ("*tldo_lduw1_sp64"): Likewise.
24904 ("*tldo_ldx_sp64"): Likewise.
24905 ("*mov<VM32:mode>_insn"): Likewise.
24906 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24907 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24909 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24911 * config/sparc/sparc.md ("type"): New insn type viscmp.
24912 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24914 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24915 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24916 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24917 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24919 ("n7_vis_logical_11cycle"): Likewise.
24920 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24921 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24922 * config/sparc/niagara.md ("niag_vis"): Likewise.
24923 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24924 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24926 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24928 * config/sparc/sparc.md: New instruction type `bmask'.
24929 (bmaskdi_vis): Use the `bmask' type.
24930 (bmasksi_vis): Likewise.
24931 * config/sparc/ultra3.md (us3_array): Likewise.
24932 * config/sparc/niagara7.md (n7_array): Likewise.
24933 * config/sparc/niagara4.md (n4_array): Likewise.
24934 * config/sparc/niagara2.md (niag2_vis): Likewise.
24935 (niag3_vis): Likewise.
24936 * config/sparc/niagara.md (niag_vis): Likewise.
24938 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24940 * ipa-comdats.c: Remove optimize check from gate.
24941 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24942 for functions not optimized.
24943 (ipa_fn_summary_read): Skip optimize check.
24944 (ipa_fn_summary_write): Likewise.
24945 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24947 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24949 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24950 (check_callers): Give up on caller which is not optimized.
24951 (inline_small_functions): Likewise.
24952 (ipa_inline): Do not give up when not optimizing.
24953 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24954 away unoptimizes cdtors.
24955 (whole_program_function_and_variable_visibility): Do
24956 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24957 * ipa.c (process_references): Do not check optimize.
24958 (symbol_table::remove_unreachable_nodes): Update optimize check.
24959 (set_writeonly_bit): Update optimize check.
24960 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24961 (pass_ipa_single_use::gate): Remove.
24963 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24965 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24966 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24967 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24968 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24969 permute_load, permute_store, adjust_extract, adjust_splat,
24970 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24971 replace_swap_with_copy, dump_swap_insn_table,
24972 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24973 recombine_lvx_pattern, recombine_stvx_pattern,
24974 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24975 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24976 to file rs6000-p8swap.c.
24977 * config/rs6000/rs6000-p8swap.c: New file.
24978 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24979 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24980 and rs6000*-*-* targets.
24982 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24984 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24986 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24988 * lto-wrapper.c (merge_and_complain): Do not merge
24989 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24990 fsigned_zeros, ftrapping_math, fwrapv.
24991 (append_compiler_options): Do not track these options.
24992 (append_linker_options): Likewie
24994 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24996 * cgraphunit.c (cgraph_node::finalize_function): When
24997 !flag_toplevel_reorde set no_reorder flag.
24998 (varpool_node::finalize_decl): Likewise.
24999 (symbol_table::compile): Drop no toplevel reorder path.
25001 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25003 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25004 edges; zero probability is not better than uninitialized.
25006 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25008 * asan.h (asan_sanitize_allocas_p): Declare.
25009 * asan.c (asan_sanitize_allocas_p): New function.
25010 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25011 (handle_builtin_alloca): Likewise.
25012 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25013 if !asan_sanitize_allocas_p.
25014 * params.def (asan-instrument-allocas): Add new option.
25015 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25016 * opts.c (common_handle_option): Disable allocas sanitization for
25019 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25021 * asan.c: Include gimple-fold.h.
25022 (get_last_alloca_addr): New function.
25023 (handle_builtin_stackrestore): Likewise.
25024 (handle_builtin_alloca): Likewise.
25025 (asan_emit_allocas_unpoison): Likewise.
25026 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25027 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25028 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25029 (instrument_builtin_call): Pass gimple iterator to
25030 get_mem_refs_of_builtin_call.
25031 (last_alloca_addr): New global.
25032 * asan.h (asan_emit_allocas_unpoison): Declare.
25033 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25034 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25035 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25036 if function calls alloca.
25037 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25038 * gimple-fold.h (replace_call_with_value): Declare.
25039 * internal-fn.c: Include asan.h.
25040 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25041 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25043 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25045 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25046 (C_SELFTEST_FLAGS): New.
25047 (CPP_SELFTEST_FLAGS): New.
25048 (SELFTEST_DEPS): New, from deps of s-selftest.
25049 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25050 (CPP_SELFTEST_DEPS): New.
25051 (selftest): Add dependency on s-selftest-c++.
25052 (s-selftest): Rename to...
25053 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25054 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25055 than SELFTEST_FLAGS.
25056 (selftest-gdb): Rename to...
25057 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25059 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25060 (selftest-valgrind): Rename to...
25061 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25063 (selftest-valgrind): Reintroduce as an alias for
25064 selftest-c-valgrind.
25065 (s-selftest-c++): New.
25066 (selftest-c++-gdb): New.
25067 (selftest-c++-valgrind): New.
25069 2017-07-06 Olivier Hainque <hainque@adacore.com>
25071 * gcc.c (process_command): When deciding if undefined variables
25072 should be ignored when processing specs, accept "gcc -v" as well.
25074 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25076 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25077 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25079 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25081 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25082 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25083 * config/arm/arm-tables.opt: Regenerate.
25084 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25086 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25088 2017-07-06 Carl Love <cel@us.ibm.com>
25090 * ChangeLog: Clean up from mid air collision
25092 2017-07-06 Carl Love <cel@us.ibm.com>
25094 * config/rs6000/rs6000-c.c: Add support for built-in functions
25095 vector signed int vec_subc (vector signed int, vector signed int);
25096 vector signed __int128 vec_subc (vector signed __int128,
25097 vector signed __int128);
25098 vector unsigned __int128 vec_subc (vector unsigned __int128,
25099 vector unsigned __int128);
25100 vector signed int vec_sube (vector signed int, vector signed int,
25101 vector signed int);
25102 vector unsigned int vec_sube (vector unsigned int,
25103 vector unsigned int,
25104 vector unsigned int);
25105 vector signed __int128 vec_sube (vector signed __int128,
25106 vector signed __int128,
25107 vector signed__int128);
25108 vector unsigned __int128 vec_sube (vector unsigned __int128,
25109 vector unsigned __int128,
25110 vector unsigned __int128);
25111 vector signed int vec_subec (vector signed int, vector signed int,
25112 vector signed int);
25113 vector unsigned int vec_subec (vector unsigned int,
25114 vector unsigned int,
25115 vector unsigned int);
25116 vector signed __int128 vec_subec (vector signed __int128,
25117 vector signed __int128,
25118 vector signed__int128);
25119 vector unsigned __int128 vec_subec (vector unsigned __int128,
25120 vector unsigned __int128,
25121 vector unsigned __int128);
25122 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25123 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25124 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25125 BU_ALTIVEC_OVERLOAD_X definitions.
25126 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25127 * doc/extend.texi: Update the built-in documentation file for the new
25128 built-in functions.
25130 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25133 * diagnostic-show-locus.c (layout::layout): Use start and finish
25134 spelling location for the start and finish of each range.
25135 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25136 Add unused aspect param.
25137 * input.c (expand_location_1): Add "aspect" param, and use it
25138 to access the correct part of the location.
25139 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25141 (expand_location_to_spelling_point): Likewise.
25142 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25143 param, and pass it to expand_location_1.
25145 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25147 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25148 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25149 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25150 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25151 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25152 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25153 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25154 _mm_maskz_getmant_ss): New intrinsics.
25155 (__builtin_ia32_getexpss128_mask): Changed to ...
25156 __builtin_ia32_getexpss128_round ... this.
25157 (__builtin_ia32_getexpsd128_mask): Changed to ...
25158 __builtin_ia32_getexpsd128_round ... this.
25159 * config/i386/i386-builtin-types.def
25160 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25161 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25162 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25163 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25164 __builtin_ia32_getmantss_mask_round): New builtins.
25165 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25166 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25167 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25168 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25169 * config/i386/sse.md
25170 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25171 avx512f_sgetexp<mode><mask_scalar_name>
25172 <round_saeonly_scalar_name> ... this.
25173 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25174 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25175 vgetexp<ssescalarmodesuffix>
25176 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25177 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25178 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25179 avx512f_vgetmant<mode><mask_scalar_name>
25180 <round_saeonly_scalar_name> ... this.
25181 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25182 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25183 vgetmant<ssescalarmodesuffix>
25184 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25185 %0<mask_scalar_operand4>, %1, %2
25186 <round_saeonly_scalar_mask_op4>, %3} ... this.
25187 * config/i386/subst.md (mask_scalar_operand4,
25188 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25189 round_saeonly_scalar_nimm_predicate): New subst attributes.
25191 2017-07-06 Julia Koval <julia.koval@intel.com>
25193 * config/i386/i386.c (ix86_erase_embedded_rounding):
25194 Remove code for old rounding pattern.
25196 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25198 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25200 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25202 * doc/sourcebuild.texi (Test Directives, Variants of
25203 dg-require-support): Add documentation for dg-require-stack-check.
25205 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25207 * config/i386/subst.md (mask_scalar, round_scalar,
25208 round_saeonly_scalar): New meta-templates.
25209 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25210 round_scalar_mask_operand3, round_scalar_mask_op3,
25211 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25212 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25213 round_saeonly_scalar_constraint,
25214 round_saeonly_scalar_prefix): New subst attribute.
25215 * config/i386/sse.md
25216 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25217 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25218 <round_scalar_name> ... this.
25219 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25220 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25221 <round_scalar_name> ... this.
25222 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25223 <sse>_vm<code><mode>3<mask_scalar_name>
25224 <round_saeonly_scalar_name> ... this.
25225 (v<plusminus_mnemonic><ssescalarmodesuffix>
25226 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25227 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25228 v<plusminus_mnemonic><ssescalarmodesuffix>
25229 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25230 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25231 (v<multdiv_mnemonic><ssescalarmodesuffix>
25232 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25233 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25234 v<multdiv_mnemonic><ssescalarmodesuffix>
25235 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25236 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25237 (v<maxmin_float><ssescalarmodesuffix>
25238 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25239 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25240 v<maxmin_float><ssescalarmodesuffix>
25241 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25242 %0<mask_scalar_operand3>, %1, %<iptr>2
25243 <round_saeonly_scalar_mask_op3>} ... this.
25245 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25247 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25248 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25250 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25251 Alan Hayward <alan.hayward@arm.com>
25252 David Sherwood <david.sherwood@arm.com>
25254 * combine.c (simplify_if_then_else): Remove "enum" before
25256 * compare-elim.c (can_eliminate_compare): Likewise.
25257 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25259 (aarch64_lookup_simd_builtin_type): Likewise.
25260 (aarch64_simd_builtin_type): Likewise.
25261 (aarch64_init_simd_builtin_types): Likewise.
25262 (aarch64_simd_expand_args): Likewise.
25263 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25265 (aarch64_reverse_mask): Likewise.
25266 (aarch64_simd_emit_reg_reg_move): Likewise.
25267 (aarch64_gen_adjusted_ldpstp): Likewise.
25268 (aarch64_ccmp_mode_to_code): Likewise.
25269 (aarch64_operands_ok_for_ldpstp): Likewise.
25270 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25271 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25273 (aarch64_min_divisions_for_recip_mul): Likewise.
25274 (aarch64_reassociation_width): Likewise.
25275 (aarch64_get_condition_code_1): Likewise.
25276 (aarch64_simd_emit_reg_reg_move): Likewise.
25277 (aarch64_simd_attr_length_rglist): Likewise.
25278 (aarch64_reverse_mask): Likewise.
25279 (aarch64_operands_ok_for_ldpstp): Likewise.
25280 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25281 (aarch64_gen_adjusted_ldpstp): Likewise.
25282 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25284 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25285 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25286 (arm_lookup_simd_builtin_type): Likewise.
25287 (arm_simd_builtin_type): Likewise.
25288 (arm_init_simd_builtin_types): Likewise.
25289 (arm_expand_builtin_args): Likewise.
25290 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25291 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25292 (ft32_setup_incoming_varargs): Likewise.
25293 (ft32_function_arg): Likewise.
25294 (ft32_function_arg_advance): Likewise.
25295 (ft32_pass_by_reference): Likewise.
25296 (ft32_arg_partial_bytes): Likewise.
25297 (ft32_valid_pointer_mode): Likewise.
25298 (ft32_addr_space_pointer_mode): Likewise.
25299 (ft32_addr_space_legitimate_address_p): Likewise.
25300 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25302 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25303 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25304 (iamcu_alignment): Likewise.
25305 (canonicalize_vector_int_perm): Likewise.
25306 (ix86_noce_conversion_profitable_p): Likewise.
25307 (ix86_mpx_bound_mode): Likewise.
25308 (ix86_operands_ok_for_move_multiple): Likewise.
25309 * config/microblaze/microblaze-protos.h
25310 (microblaze_expand_conditional_branch_reg): Likewise.
25311 * config/microblaze/microblaze.c
25312 (microblaze_expand_conditional_branch_reg): Likewise.
25313 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25315 (rs6000_reassociation_width): Likewise.
25316 (rs6000_invalid_binary_op): Likewise.
25317 (fusion_p9_p): Likewise.
25318 (emit_fusion_p9_load): Likewise.
25319 (emit_fusion_p9_store): Likewise.
25320 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25322 (riscv_hard_regno_mode_ok_p): Likewise.
25323 (riscv_address_insns): Likewise.
25324 (riscv_split_symbol): Likewise.
25325 (riscv_legitimize_move): Likewise.
25326 (riscv_function_value): Likewise.
25327 (riscv_hard_regno_nregs): Likewise.
25328 (riscv_expand_builtin): Likewise.
25329 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25330 (riscv_build_integer): Likewise.
25331 (riscv_split_integer): Likewise.
25332 (riscv_legitimate_constant_p): Likewise.
25333 (riscv_cannot_force_const_mem): Likewise.
25334 (riscv_regno_mode_ok_for_base_p): Likewise.
25335 (riscv_valid_base_register_p): Likewise.
25336 (riscv_valid_offset_p): Likewise.
25337 (riscv_valid_lo_sum_p): Likewise.
25338 (riscv_classify_address): Likewise.
25339 (riscv_legitimate_address_p): Likewise.
25340 (riscv_address_insns): Likewise.
25341 (riscv_load_store_insns): Likewise.
25342 (riscv_force_binary): Likewise.
25343 (riscv_split_symbol): Likewise.
25344 (riscv_force_address): Likewise.
25345 (riscv_legitimize_address): Likewise.
25346 (riscv_move_integer): Likewise.
25347 (riscv_legitimize_const_move): Likewise.
25348 (riscv_legitimize_move): Likewise.
25349 (riscv_address_cost): Likewise.
25350 (riscv_subword): Likewise.
25351 (riscv_output_move): Likewise.
25352 (riscv_canonicalize_int_order_test): Likewise.
25353 (riscv_emit_int_order_test): Likewise.
25354 (riscv_function_arg_boundary): Likewise.
25355 (riscv_pass_mode_in_fpr_p): Likewise.
25356 (riscv_pass_fpr_single): Likewise.
25357 (riscv_pass_fpr_pair): Likewise.
25358 (riscv_get_arg_info): Likewise.
25359 (riscv_function_arg): Likewise.
25360 (riscv_function_arg_advance): Likewise.
25361 (riscv_arg_partial_bytes): Likewise.
25362 (riscv_function_value): Likewise.
25363 (riscv_pass_by_reference): Likewise.
25364 (riscv_setup_incoming_varargs): Likewise.
25365 (riscv_print_operand): Likewise.
25366 (riscv_elf_select_rtx_section): Likewise.
25367 (riscv_save_restore_reg): Likewise.
25368 (riscv_for_each_saved_reg): Likewise.
25369 (riscv_register_move_cost): Likewise.
25370 (riscv_hard_regno_mode_ok_p): Likewise.
25371 (riscv_hard_regno_nregs): Likewise.
25372 (riscv_class_max_nregs): Likewise.
25373 (riscv_memory_move_cost): Likewise.
25374 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25375 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25376 (rl78_addr_space_address_mode): Likewise.
25377 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25379 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25380 (rs6000_reassociation_width): Likewise.
25381 (rs6000_invalid_binary_op): Likewise.
25382 (fusion_p9_p): Likewise.
25383 (emit_fusion_p9_load): Likewise.
25384 (emit_fusion_p9_store): Likewise.
25385 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25386 (ok_for_simple_move_operands): Likewise.
25387 (ok_for_simple_move_strict_operands): Likewise.
25388 (ok_for_simple_arith_logic_operands): Likewise.
25389 (visium_legitimize_reload_address): Likewise.
25390 (visium_select_cc_mode): Likewise.
25391 (output_cbranch): Likewise.
25392 (visium_split_double_move): Likewise.
25393 (visium_expand_copysign): Likewise.
25394 (visium_expand_int_cstore): Likewise.
25395 (visium_expand_fp_cstore): Likewise.
25396 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25397 (visium_function_arg): Likewise.
25398 (visium_function_arg_advance): Likewise.
25399 (visium_libcall_value): Likewise.
25400 (visium_setup_incoming_varargs): Likewise.
25401 (visium_legitimate_constant_p): Likewise.
25402 (visium_legitimate_address_p): Likewise.
25403 (visium_legitimize_address): Likewise.
25404 (visium_secondary_reload): Likewise.
25405 (visium_register_move_cost): Likewise.
25406 (visium_memory_move_cost): Likewise.
25407 (prepare_move_operands): Likewise.
25408 (ok_for_simple_move_operands): Likewise.
25409 (ok_for_simple_move_strict_operands): Likewise.
25410 (ok_for_simple_arith_logic_operands): Likewise.
25411 (visium_function_value_1): Likewise.
25412 (rtx_ok_for_offset_p): Likewise.
25413 (visium_legitimize_reload_address): Likewise.
25414 (visium_split_double_move): Likewise.
25415 (visium_expand_copysign): Likewise.
25416 (visium_expand_int_cstore): Likewise.
25417 (visium_expand_fp_cstore): Likewise.
25418 (visium_split_cstore): Likewise.
25419 (visium_select_cc_mode): Likewise.
25420 (visium_split_cbranch): Likewise.
25421 (output_cbranch): Likewise.
25422 (visium_print_operand_address): Likewise.
25423 * expmed.c (flip_storage_order): Likewise.
25424 * expmed.h (emit_cstore): Likewise.
25425 (flip_storage_order): Likewise.
25426 * genrecog.c (validate_pattern): Likewise.
25427 * hsa-gen.c (gen_hsa_addr): Likewise.
25428 * internal-fn.c (expand_arith_overflow): Likewise.
25429 * ira-color.c (allocno_copy_cost_saving): Likewise.
25430 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25431 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25432 (process_invariant_for_inheritance): Likewise.
25433 * lra-eliminations.c (move_plus_up): Likewise.
25434 * omp-low.c (lower_oacc_reductions): Likewise.
25435 * simplify-rtx.c (simplify_subreg): Likewise.
25436 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25437 (TARGET_CHKP_BOUND_MODE): Likewise..
25438 * targhooks.c (default_chkp_bound_mode): Likewise.
25439 (default_setup_incoming_vararg_bounds): Likewise.
25440 * targhooks.h (default_chkp_bound_mode): Likewise.
25441 (default_setup_incoming_vararg_bounds): Likewise.
25442 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25443 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25444 (have_whole_vector_shift): Likewise.
25445 * tree-vect-stmts.c (vectorizable_load): Likewise.
25446 * doc/tm.texi: Regenerate.
25448 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25450 Graceful degrade if Binutils PR21472 is not available.
25453 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25454 .rodata in flash test fails.
25455 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25456 * confgure: Regenerate.
25457 * config.in: Regenerate.
25458 * config/avr/avr.c (avr_asm_named_section)
25459 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25460 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25461 (avr_asm_init_sections): Same.
25463 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25465 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25466 (fma<VH:mode>4_intrinsic): Likewise.
25467 (*fmsub<VCVTF:mode>4): Likewise.
25468 (*fmsub<VH:mode>4_intrinsic): Likewise.
25470 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25473 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25474 Don't depend on "optimize > 0".
25475 (out_movhi_r_mr, out_movqi_mr_r): Same.
25476 (out_movhi_mr_r, out_movqi_r_mr): Same.
25477 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25478 io_address_operand on "optimize > 0".
25480 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25482 * tree-loop-distribution.c: Add general explanantion on the pass.
25483 (generate_loops_for_partition): Mark distributed loop.
25484 (pg_add_dependence_edges): New parameter. Handle alias data
25485 dependence specially and record it in the parameter if asked.
25486 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25487 (init_partition_graph_vertices, add_partition_graph_edge): New.
25488 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25489 (free_partition_graph_vdata, build_partition_graph): New.
25490 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25491 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25492 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25493 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25494 (version_for_distribution_p, finalize_partitions): New.
25495 (distribute_loop): Handle alias data dependence specially. Factor
25496 out loop fusion code as functions and call these functions.
25498 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25500 * tree-loop-distribution.c (classify_partition): New parameter and
25501 better handle reduction statement.
25502 (rdg_build_partitions): Revise comment.
25503 (distribute_loop): Compute statements in all partitions and pass it
25504 to classify_partition.
25506 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25508 * tree-loop-distribution.c (enum partition_type): New.
25509 (struct partition): New field type.
25510 (partition_merge_into): Add parameter. Update partition type.
25511 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25512 (build_rdg_partition_for_vertex): Compute partition type.
25513 (rdg_build_partitions): Dump partition type.
25514 (distribute_loop): Update calls to partition_merge_into.
25516 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25518 * tree-loop-distribution.c (struct ddr_hasher): New.
25519 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25521 (classify_partition): Call get_data_dependence.
25522 (pg_add_dependence_edges): Ditto.
25523 (distribute_loop): Release data dependence hash table.
25525 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25527 * tree-loop-distribution.c (ref_base_address): Delete.
25528 (similar_memory_accesses): Rename ...
25529 (share_memory_accesses): ... to this. Check if partitions access
25530 the same memory reference.
25531 (distribute_loop): Call share_memory_accesses.
25533 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25535 * tree-loop-distribution.c (struct partition): New field recording
25536 its data reference.
25537 (partition_alloc, partition_free): Init and release data refs.
25538 (partition_merge_into): Merge data refs.
25539 (build_rdg_partition_for_vertex): Collect data refs for partition.
25540 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25542 (distribute_loop): Remve data refs from vertice data of partition
25545 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25547 * tree-loop-distribution.c (params.h): Include header file.
25548 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25549 (datarefs_vec): New global var.
25550 (create_rdg_vertices): Use datarefs_vec directly.
25551 (free_rdg): Don't free data references.
25552 (build_rdg): Update use. Don't free data references.
25553 (distribute_loop): Compute global variable for data references.
25554 Bail out if there are too many data references.
25556 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25558 * tree-loop-distribution.c (loop_nest): New global var.
25559 (build_rdg): Use loop directly, rather than loop nest.
25560 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25562 (distribute_loop): Compute global variable loop nest. Update use.
25564 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25566 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25567 (partition_merge_into): New parameter. Dump reason for fusion.
25568 (distribute_loop): Update use of partition_merge_into.
25570 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25572 * tree-loop-distribution.c (bb_top_order_index): New.
25573 (bb_top_order_index_size, bb_top_order_cmp): New.
25574 (stmts_from_loop): Use topological order.
25575 (pass_loop_distribution::execute): Compute and release topological
25576 order for basic blocks.
25578 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25580 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25583 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25585 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25586 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25587 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25588 * internal-fn.def (LOOP_DIST_ALIAS): New.
25589 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25590 (fold_loop_internal_call): ... this.
25591 (vect_loop_dist_alias_call): New function.
25592 (set_uid_loop_bbs): Call fold_loop_internal_call.
25593 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25596 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25599 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25600 Require dead FLAGS_REG at the beginning of a peephole.
25602 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25605 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25606 arguments in the call to __builtin_ia32_sbb_u32.
25607 (_subborrow_u64): Swap _X and _Y arguments in the call to
25608 __builtin_ia32_sbb_u64.
25610 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25613 * tree-vrp.c (compare_assert_loc): Turn into a function template
25614 with stable template parameter. Only test if a->e is NULL,
25615 !a->e == !b->e has been verified already. Use e == NULL or
25616 e != NULL instead of e or ! e tests. If stable is true, don't use
25617 iterative_hash_expr, on the other side allow a or b or both NULL
25618 and sort the NULLs last.
25619 (process_assert_insertions): Sort using compare_assert_loc<false>
25620 instead of compare_assert_loc, later sort using
25621 compare_assert_loc<true> before calling process_assert_insertions_for
25622 in a loop. Use break instead of continue once seen NULL pointer.
25624 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25626 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25627 Cortex-R7 and Cortex-R8 processors.
25629 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25631 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25632 uninitialized while src is not.
25634 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25636 * common/config/arm/arm-common.c: Adjust include path for
25638 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25639 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25640 (arm-cpu-data.h): Likewise.
25641 (arm-cpu-cdata.h): Likewise.
25642 * config/arm/arm-cpu.h: Delete.
25643 * config/arm/arm-cpu-cdata.h: Delete.
25644 * config/arm/arm-cpu-data.h: Delete.
25646 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25648 * config/arm/arm-cpus.in (cortex-a55): New.
25649 (cortex-a75): Likewise.
25650 (cortex-a75.cortex-a55): Likewise.
25651 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25653 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25654 * config/arm/arm-cpu-cdata.h: Regenerate.
25655 * config/arm/arm-cpu-data.h: Regenerate.
25656 * config/arm/arm-cpu.h: Regenerate.
25657 * config/arm/arm-tables.opt: Regenerate.
25658 * config/arm/arm-tune.md: Regenerate.
25660 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25662 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25664 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25666 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25669 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25671 PR tree-optimization/81292
25672 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25673 full_string_p, also call adjust_related_strinfos if the adjustment
25674 is simple, otherwise invalidate related strinfos.
25676 2017-07-04 Martin Liska <mliska@suse.cz>
25679 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25680 newly created variable as DECL_IGNORED_P.
25682 2017-07-04 Martin Liska <mliska@suse.cz>
25685 * ipa-inline.c (inline_small_functions):
25686 Use xstrdup_for_dump.
25688 2017-07-04 Tom de Vries <tom@codesourcery.com>
25690 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25692 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25695 * config/darwin.c (darwin_function_switched_text_sections):
25698 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25700 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25702 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25704 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25706 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25708 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25709 min_profitable_iters, and th as inclusive lower bounds.
25710 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25711 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25712 for min_profitable_iters and min_profitable_estimate.
25713 (vect_transform_loop): Treat th as an inclusive lower bound.
25714 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25716 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25719 * config/darwin.c (darwin_function_switched_text_sections):
25720 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25721 in two pieces, and suppress the use of buf.
25723 2017-07-03 Nathan Sidwell <nathan@acm.org>
25725 * hash-table.h (hash_table_mod1): Fix indentation.
25727 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25729 PR middle-end/81290
25730 * predict.c (force_edge_cold): Be more careful about propagation
25732 * profile-count.h (profile_probability::guessed,
25733 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25735 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25737 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25739 * doc/invoke.texi (rcpc architecture extension): Document it.
25741 2017-07-03 Richard Biener <rguenther@suse.de>
25743 PR tree-optimization/60510
25744 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25745 the scalar reduction PHI and use it.
25746 (vectorizable_reduction): Properly guard the single_defuse_cycle
25747 path for non-SLP reduction chains where we cannot use it.
25748 Rework reduc_def/index and vector type deduction. Rework
25749 vector operand gathering during reduction op code-gen.
25750 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25751 chains dissolve the chain and leave it to non-SLP reduction
25754 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25756 * tree-data-ref.h (dr_alignment): Declare.
25757 * tree-data-ref.c (dr_alignment): New function.
25758 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25759 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25761 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25763 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25765 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25766 and base_misalignment fields.
25767 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25768 * tree-data-ref.c: Include builtins.h.
25769 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25770 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25771 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25772 * tree-vect-data-refs.c: Include tree-cfg.h.
25773 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25774 fields instead of calculating an alignment here.
25775 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25776 innermost_loop_behavior fields.
25778 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25780 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25782 (DR_STEP_ALIGNMENT): New macro.
25783 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25784 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25785 (create_data_ref): Print it.
25786 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25787 to tell whether the step preserves vector (mis)alignment.
25788 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25789 Move the check for an integer step and generalise to all INTEGER_CST.
25790 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
25791 Print the outer step alignment.
25793 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25795 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
25796 with offset_alignment.
25797 (DR_ALIGNED_TO): Delete.
25798 (DR_OFFSET_ALIGNMENT): New macro.
25799 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
25800 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
25801 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
25802 (create_data_ref): Likewise.
25803 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25804 (vect_analyze_data_refs): Likewise.
25805 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
25806 creating dummy innermost behavior.
25808 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25810 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
25811 with a "innermost_loop_behavior *" and refeence tree.
25812 * tree-data-ref.c (dr_analyze_innermost): Likewise.
25813 (create_data_ref): Update call accordingly.
25814 * tree-predcom.c (find_looparound_phi): Likewise.
25816 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25818 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25819 fields with dr_wrt_vec_loop.
25820 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25821 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25822 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25823 (vect_dr_behavior): New function.
25824 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25825 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25826 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25827 track whether the step preserves the misalignment.
25828 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25829 Use vect_dr_behavior.
25830 (vect_setup_realignment): Update call accordingly.
25831 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25832 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25833 call to vect_create_addr_base_for_vector_ref.
25834 (vect_create_cond_for_align_checks): Likewise.
25835 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25836 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25837 (vect_recog_mask_conversion_pattern): Likewise.
25838 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25839 (new_stmt_vec_info): Remove redundant zeroing.
25841 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25843 * common/config/arm/arm-common.c (arm_be8_option): New function.
25844 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
25845 (ISA_ARMv6): Add isa_bit_be8.
25846 * config/arm/arm.h (arm_be8_option): Add prototype.
25847 (BE8_SPEC_FUNCTION): New define.
25848 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
25849 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
25850 (mlittle-endian): Similarly.
25851 (mbe8, mbe32): New options.
25852 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25853 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25855 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25857 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
25859 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25861 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
25862 (cleanup_tree_cfg_bb): Use it.
25863 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
25865 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
25867 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25870 * loop-doloop.c (add_test): Update profile.
25872 2017-07-03 Martin Liska <mliska@suse.cz>
25875 * sanopt.c (rewrite_usage_of_param): New function.
25876 (sanitize_rewrite_addressable_params): Likewise.
25877 (pass_sanopt::execute): Call rewrite_usage_of_param.
25879 2017-07-03 Richard Biener <rguenther@suse.de>
25881 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25882 back to using VIEW_CONVERT_EXPR.
25884 2017-07-03 Martin Liska <mliska@suse.cz>
25887 * doc/extend.texi: Document when a resolver function is
25888 generated for target_clones.
25890 2017-07-03 Martin Liska <mliska@suse.cz>
25892 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25893 and shadow memory of auto variables which are subject of
25894 use-after-scope sanitization.
25895 (asan_expand_mark_ifn): Add do set only when is_poison.
25897 2016-07-03 Richard Biener <rguenther@suse.de>
25899 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25901 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25902 (vectorizable_reduction): Perform reduction PHI creation when
25903 visiting a reduction PHI and adjust and simplify code generation
25904 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25905 (vect_transform_loop): Visit reduction PHIs.
25906 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25907 defs into the SLP tree.
25908 (vect_build_slp_tree): Reduction defs terminate the recursion.
25909 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25911 (vect_get_vec_defs_for_stmt_copy): Export.
25912 (vect_get_vec_defs): Likewise.
25913 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25915 (vect_get_vec_defs_for_stmt_copy): Declare.
25916 (vect_get_vec_defs): Likewise.
25918 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25920 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25921 parameter with a "loop" parameter and use it instead of the
25922 loop containing DR_STMT. Don't check simple_iv when doing
25923 BB analysis. Describe the two analysis modes in the comment.
25925 2017-07-03 Tom de Vries <tom@codesourcery.com>
25927 PR tree-optimization/69468
25928 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25929 (find_same_succ_bb): Handle ignore_edge_flags.
25931 2017-07-03 Tom de Vries <tom@codesourcery.com>
25933 PR tree-optimization/81192
25934 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25936 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25938 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25940 2017-07-03 Tom de Vries <tom@codesourcery.com>
25942 PR tree-optimization/81192
25943 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25944 BB_SAME_SUCC (bb) == NULL.
25946 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25948 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25951 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25953 * dumpfile.c: Include profile-count.h
25954 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25956 (insert_cond_bb): Update profile.
25957 * tree-cfg.h (insert_cond_bb): Update prototype.
25958 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25959 * tree-dump.c: Do not include tree-cfg.
25961 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25963 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25965 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25967 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25970 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25972 * tree-complex.c (expand_complex_div_wide): update profile.
25974 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25975 Alan Hayward <alan.hayward@arm.com>
25976 David Sherwood <david.sherwood@arm.com>
25978 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25979 (CORETYPES_H): New define.
25980 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25981 (insn-modes-inline.h, s-modes-inline-h): New rules.
25982 (generated_files): Add insn-modes-inline.h.
25983 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25984 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25985 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25986 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25987 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25988 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25989 (build/gencodes.o, build/genconditions.o): Likewise.
25990 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25991 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25992 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25993 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25994 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25995 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25996 * coretypes.h: Include everything up to real.h for generators.
25997 Include insn-modes.h first. Include wide-int-print.h after
25998 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25999 * machmode.h: Don't include insn-modes.h here.
26000 * function-tests.c: Remove includes of signop.h, machmode.h,
26001 double-int.h and wide-int.h.
26003 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26005 * optc-save-gen.awk: Likewise.
26006 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26007 * godump.c: Remove include of wide-int-print.h.
26008 * pretty-print.h: Likewise.
26009 * wide-int-print.cc: Likewise.
26010 * wide-int.cc: Likewise.
26011 * hash-map-tests.c: Remove include of signop.h.
26012 * hash-set-tests.c: Likewise.
26013 * rtl-tests.c: Likewise.
26014 * mkconfig.sh: Remove include of machmode.h.
26015 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26017 (emit_insn_modes_inline_h): ...this new function. Emit the code
26018 into an insn-modes-inline.h header file, adding appropriate
26019 include guards and end comments.
26020 (emit_insn_modes_c_header): Remove include of machmode.h.
26021 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26023 (main): Handle -i flag and call emit_insn_modes_inline_h when
26026 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26028 * tree-ssa-strlen.c (strinfo): Rename the length field to
26029 nonzero_chars. Add a full_string_p field.
26030 (compare_nonzero_chars, zero_length_string_p): New functions.
26031 (get_addr_stridx): Add an offset_out parameter.
26032 Use compare_nonzero_chars.
26033 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26034 (new_strinfo): Update after above changes to strinfo.
26035 (set_endptr_and_length): Set full_string_p.
26036 (get_string_length): Update after above changes to strinfo.
26037 (unshare_strinfo): Update call to new_strinfo.
26038 (maybe_invalidate): Likewise.
26039 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26040 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26041 as a uhwi instead of an shwi. Update after above changes to
26042 strinfo and new_strinfo.
26043 (zero_length_string): Assert that chainsi contains full strings.
26044 Use zero_length_string_p. Update call to new_strinfo.
26045 (adjust_related_strinfos): Update after above changes to strinfo.
26046 Copy full_string_p from origsi.
26047 (adjust_last_stmt): Use zero_length_string_p.
26048 (handle_builtin_strlen): Update after above changes to strinfo and
26049 new_strinfo. Install the lhs as the string length if the previous
26050 entry didn't describe a full string.
26051 (handle_builtin_strchr): Update after above changes to strinfo
26053 (handle_builtin_strcpy): Likewise.
26054 (handle_builtin_strcat): Likewise.
26055 (handle_builtin_malloc): Likewise.
26056 (handle_pointer_plus): Likewise.
26057 (handle_builtin_memcpy): Likewise. Track nonzero characters
26058 that aren't necessarily followed by a nul terminator.
26059 (handle_char_store): Likewise.
26061 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26063 PR tree-optimization/80769
26064 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26065 for malloc and calloc. Document the new invariant that all related
26066 strinfos have delayed lengths or none do.
26067 (verify_related_strinfos): Move earlier in file.
26068 (set_endptr_and_length): New function, split out from...
26069 (get_string_length): ...here. Also set the lengths of related
26071 (zero_length_string): Assert that chainsi has known (rather than
26073 (adjust_related_strinfos): Likewise.
26075 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26077 PR tree-optimization/81136
26078 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26079 assert that two references with the same misalignment have the same
26080 compile-time misalignment if those compile-time misalignments
26083 2017-07-01 Andi Kleen <ak@linux.intel.com>
26085 * print-tree.c (print_node): Print all attributes.
26087 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26089 * cfg.c (scale_bbs_frequencies): New function.
26090 * cfg.h (scale_bbs_frequencies): Declare it.
26091 * cfgloopanal.c (single_likely_exit): Cleanup.
26092 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26094 (scale_loop_profile): Likewise.
26095 (loop_version): Likewise.
26096 (create_empty_loop_on_edge): Update.
26097 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26098 scale_loop_frequencies, scale_loop_profile, loopify,
26099 loop_version): Update prototypes.
26100 * modulo-sched.c (sms_schedule): Update.
26101 * predict.c (unlikely_executed_edge_p): Also check probability.
26102 (probably_never_executed_edge_p): Fix typo.
26103 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26104 * tree-parloops.c (gen_parallel_loop): Update.
26105 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26106 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26107 * tree-ssa-loop-split.c (split_loop): Update.
26108 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26109 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26110 (vect_loop_versioning): Update.
26111 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26113 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26115 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26117 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26119 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26120 to keep profile consistent.
26122 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26124 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26125 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26126 * profile-count.h (max_safe_multiplier): Make unsigned.
26127 (profile_count::guessed_zero): New.
26129 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26131 * bb-reorder.c (fix_up_crossing_landing_pad,
26132 fix_crossing_conditional_branches): Use make_single_succ_edge
26133 to keep profile consistent.
26135 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26137 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26140 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26143 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26144 the right scopes, make sure cond_jump isn't preserved between multiple
26145 iterations. Search for fallthru edge whenever there are 3+ edges and
26146 use find_fallthru_edge for it.
26148 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26150 Patch by Alexander Monakov <amonakov@ispras.ru>
26151 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26152 probabilities consistently.
26154 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26156 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26157 emit_cmp_and_jump_insns.
26159 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26162 * tree-inline.c (expand_call_inline): Combine profile statuses.
26164 2017-06-30 Andrew Pinski <apinski@cavium.com>
26166 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26167 fold_stmt returned true.
26169 2017-06-30 Nathan Sidwell <nathan@acm.org>
26171 * ggc.h (empty_string): Delete.
26172 * cfgexpand.c (expand_asm_stmt): Use plain "".
26173 * optabs.c (expand_asm_memory_barrier): Likewise.
26174 * stringpool.c (empty_string): Delete.
26175 (digit_vector, digit_string): Delete.
26176 (ggc_alloc_string): Use plain "", don't optimize single digit
26177 strings. Use ggc_alloc_atomic.
26179 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26181 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26182 comparison set and one other set, use the cost of the non-comparison
26185 2017-06-30 Nathan Sidwell <nathan@acm.org>
26187 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26190 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26192 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26193 loops. Remove now unneeded calls to gimple_switch_set_label() that
26194 just set removed labels to NULL_TREE.
26196 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26198 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26199 (set_range_info): ...here. Only call set_range_info_raw if domain
26201 (set_nonzero_bits): Call set_range_info_raw.
26202 * tree-ssanames.h (set_range_info_raw): New.
26204 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26207 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26208 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26209 of nonimmediate_operand and <store_mask_constraint> instead of m
26210 for the input operand. For V8FI iterator, always split if input
26211 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26212 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26213 <store_mask_predicate> instead of register_operand and
26214 <store_mask_constraint> instead of v for the input operand. Make
26215 sure both operands aren't MEMs for if not <mask_applied>.
26217 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26219 * lto-wrapper.c (copy_file) Close both file descriptors before
26222 2017-06-30 Martin Liska <mliska@suse.cz>
26225 * multiple_target.c (create_dispatcher_calls): Make ifunc
26226 also for function that don't have calls or are not referenced.
26228 2017-06-30 Richard Biener <rguenther@suse.de>
26230 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26231 analyze the first scalar stmt. Move vector type computation
26232 for the BB case here from ...
26233 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26234 live operation processing in the SLP case properly.
26236 2017-06-30 Richard Biener <rguenther@suse.de>
26238 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26240 2017-06-30 Martin Liska <mliska@suse.cz>
26243 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26244 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26246 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26248 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26251 2017-06-30 Richard Biener <rguenther@suse.de>
26253 PR tree-optimization/81249
26254 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26255 condition reduction result to original scalar type.
26257 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26259 * profile-count.h (enum profile_quality): Fix typos and whitespace
26262 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26264 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26265 type for branch probabilities.
26267 2017-06-29 Julian Brown <julian@codesourcery.com>
26268 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26270 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26271 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26272 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26273 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26275 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26277 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26278 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26279 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26280 CC usage from generic code to here.
26281 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26282 CC usage into the target macros.
26284 2017-06-29 Maya Rashish <coypu@sdf.org>
26286 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26289 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26291 * arm/arm-builtins.c: Include profile-count.h
26292 * except.c (sjlj_emit_function_enter): Use
26293 profile_probability::unlikely.
26295 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26297 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26298 and tocrel_offset be pointer args rather than implicitly using
26300 (legitimate_constant_pool_address_p, rs6000_emit_move,
26301 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26302 tocrel_offset and use in toc_relative_expr_p call.
26303 (print_operand, print_operand_address): Use static tocrel_base_oac
26304 and tocrel_offset_oac.
26305 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26308 2017-06-29 Maya Rashish <coypu@sdf.org>
26310 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26312 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26314 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26315 objects, take into account only the alignment of 'op0' and 'mode1' if
26318 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26320 * ccmp.c (ccmp_tree_comparison_p): New function.
26321 (ccmp_candidate_p): Update to use above function.
26322 (get_compare_parts): New function.
26323 (expand_ccmp_next): Update to use new functions.
26324 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26326 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26327 take mode as argument.
26328 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26329 * expr.c (expand_expr_real_1): Pass mode as argument.
26331 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26333 * combine.c (combine_instructions): Print insns to dump_file, together
26336 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26338 * asan.c (asan_emit_stack_protection): Update.
26339 (create_cond_insert_point): Update.
26340 * auto-profile.c (afdo_propagate_circuit): Update.
26341 * basic-block.h (struct edge_def): Turn probability to
26342 profile_probability.
26343 (EDGE_FREQUENCY): Update.
26344 * bb-reorder.c (find_traces_1_round): Update.
26345 (better_edge_p): Update.
26346 (sanitize_hot_paths): Update.
26347 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26348 (make_single_succ_edge): Update.
26349 (check_bb_profile): Update.
26350 (dump_edge_info): Update.
26351 (update_bb_profile_for_threading): Update.
26352 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26353 probabilitycount to 0.
26354 * cfgbuild.c (compute_outgoing_frequencies): Update.
26355 * cfgcleanup.c (try_forward_edges): Update.
26356 (outgoing_edges_match): Update.
26357 (try_crossjump_to_edge): Update.
26358 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26359 (expand_gimple_tailcall): Update.
26360 (construct_init_block): Use make_single_succ_edge.
26361 (construct_exit_block): Use make_single_succ_edge.
26362 * cfghooks.c (verify_flow_info): Update.
26363 (redirect_edge_succ_nodup): Update.
26364 (split_edge): Update.
26365 (account_profile_record): Update.
26366 * cfgloopanal.c (single_likely_exit): Update.
26367 * cfgloopmanip.c (scale_loop_profile): Update.
26368 (set_zero_probability): Remove.
26369 (duplicate_loop_to_header_edge): Update.
26370 * cfgloopmanip.h (loop_version): Update prototype.
26371 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26372 (force_nonfallthru_and_redirect): Update.
26373 (update_br_prob_note): Update.
26374 (rtl_verify_edges): Update.
26375 (purge_dead_edges): Update.
26376 (rtl_lv_add_condition_to_bb): Update.
26377 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26378 * cgraphunit.c (init_lowered_empty_function): Update.
26379 (cgraph_node::expand_thunk): Update.
26380 * cilk-common.c: Include profile-count.h
26381 * dojump.c (inv): Remove.
26382 (jumpifnot): Update.
26383 (jumpifnot_1): Update.
26384 (do_jump_1): Update.
26386 (do_jump_by_parts_greater_rtx): Update.
26387 (do_compare_rtx_and_jump): Update.
26388 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26389 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26390 * dwarf2cfi.c: Include profile-count.h
26391 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26392 (sjlj_emit_dispatch_table): Likewise.
26393 * explow.c: Include profile-count.h
26394 * expmed.c (emit_store_flag_force): Update.
26395 (do_cmp_and_jump): Update.
26396 * expr.c (compare_by_pieces_d::generate): Update.
26397 (compare_by_pieces_d::finish_mode): Update.
26398 (emit_block_move_via_loop): Update.
26399 (store_expr_with_bounds): Update.
26400 (store_constructor): Update.
26401 (expand_expr_real_2): Update.
26402 (expand_expr_real_1): Update.
26403 * expr.h (try_casesi, try_tablejump): Update prototypes.
26404 * gimple-pretty-print.c (dump_probability): Update.
26405 (dump_profile): New.
26406 (dump_gimple_label): Update.
26407 (dump_gimple_bb_header): Update.
26408 * graph.c (draw_cfg_node_succ_edges): Update.
26409 * hsa-gen.c (convert_switch_statements): Update.
26410 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26411 (find_if_case_1): Update.
26412 (find_if_case_2): Update.
26413 * internal-fn.c (expand_arith_overflow_result_store): Update.
26414 (expand_addsub_overflow): Update.
26415 (expand_neg_overflow): Update.
26416 (expand_mul_overflow): Update.
26417 (expand_vector_ubsan_overflow): Update.
26418 * ipa-cp.c (good_cloning_opportunity_p): Update.
26419 * ipa-split.c (split_function): Use make_single_succ_edge.
26420 * ipa-utils.c (ipa_merge_profiles): Update.
26421 * loop-doloop.c (add_test): Update.
26422 (doloop_modify): Update.
26423 * loop-unroll.c (compare_and_jump_seq): Update.
26424 (unroll_loop_runtime_iterations): Update.
26425 * lra-constraints.c (lra_inheritance): Update.
26426 * lto-streamer-in.c (input_cfg): Update.
26427 * lto-streamer-out.c (output_cfg): Update.
26428 * mcf.c (adjust_cfg_counts): Update.
26429 * modulo-sched.c (sms_schedule): Update.
26430 * omp-expand.c (expand_omp_for_init_counts): Update.
26431 (extract_omp_for_update_vars): Update.
26432 (expand_omp_ordered_sink): Update.
26433 (expand_omp_for_ordered_loops): Update.
26434 (expand_omp_for_generic): Update.
26435 (expand_omp_for_static_nochunk): Update.
26436 (expand_omp_for_static_chunk): Update.
26437 (expand_cilk_for): Update.
26438 (expand_omp_simd): Update.
26439 (expand_omp_taskloop_for_outer): Update.
26440 (expand_omp_taskloop_for_inner): Update.
26441 * omp-simd-clone.c (simd_clone_adjust): Update.
26442 * optabs.c (expand_doubleword_shift): Update.
26443 (expand_abs): Update.
26444 (emit_cmp_and_jump_insn_1): Update.
26445 (expand_compare_and_swap_loop): Update.
26446 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26447 * predict.c (predictable_edge_p): Update.
26448 (edge_probability_reliable_p): Update.
26449 (set_even_probabilities): Update.
26450 (combine_predictions_for_insn): Update.
26451 (combine_predictions_for_bb): Update.
26452 (propagate_freq): Update.
26453 (estimate_bb_frequencies): Update.
26454 (force_edge_cold): Update.
26455 * profile-count.c (profile_count::dump): Add missing space into dump.
26456 (profile_count::debug): Add newline.
26457 (profile_count::differs_from_p): Explicitly convert to unsigned.
26458 (profile_count::stream_in): Update.
26459 (profile_probability::dump): New member function.
26460 (profile_probability::debug): New member function.
26461 (profile_probability::differs_from_p): New member function.
26462 (profile_probability::differs_lot_from_p): New member function.
26463 (profile_probability::stream_in): New member function.
26464 (profile_probability::stream_out): New member function.
26465 * profile-count.h (profile_count_quality): Rename to ...
26466 (profile_quality): ... this one.
26467 (profile_probability): New.
26468 (profile_count): Update.
26469 * profile.c (compute_branch_probabilities): Update.
26470 * recog.c (peep2_attempt): Update.
26471 * sched-ebb.c (schedule_ebbs): Update.
26472 * sched-rgn.c (find_single_block_region): Update.
26473 (compute_dom_prob_ps): Update.
26474 (schedule_region): Update.
26475 * sel-sched-ir.c (compute_succs_info): Update.
26476 * stmt.c (struct case_node): Update.
26477 (do_jump_if_equal): Update.
26478 (get_outgoing_edge_probs): Update.
26479 (conditional_probability): Update.
26480 (emit_case_dispatch_table): Update.
26481 (expand_case): Update.
26482 (expand_sjlj_dispatch_table): Update.
26483 (emit_case_nodes): Update.
26484 * targhooks.c: Update.
26485 * tracer.c (better_p): Update.
26486 (find_best_successor): Update.
26487 * trans-mem.c (expand_transaction): Update.
26488 * tree-call-cdce.c: Update.
26489 * tree-cfg.c (gimple_split_edge): Upate.
26490 (move_sese_region_to_fn): Upate.
26491 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26492 * tree-eh.c (lower_resx): Upate.
26493 (cleanup_empty_eh_move_lp): Upate.
26494 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26495 * tree-inline.c (copy_edges_for_bb): Update.
26496 (copy_cfg_body): Update.
26497 * tree-parloops.c (gen_parallel_loop): Update.
26498 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26499 (gimple_gen_time_profiler): Update.
26500 * tree-ssa-dce.c (remove_dead_stmt): Update.
26501 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26502 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26503 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26504 (unloop_loops): Update.
26505 (try_peel_loop): Update.
26506 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26507 * tree-ssa-loop-split.c (connect_loops): Update.
26508 (split_loop): Update.
26509 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26510 (hoist_guard): Update.
26511 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26512 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26513 (value_replacement): Update.
26514 * tree-ssa-reassoc.c (branch_fixup): Update.
26515 * tree-ssa-tail-merge.c (replace_block_by): Update.
26516 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26517 (create_edge_and_update_destination_phis): Update.
26518 (compute_path_counts): Update.
26519 (recompute_probabilities): Update.
26520 (update_joiner_offpath_counts): Update.
26521 (freqs_to_counts_path): Update.
26522 (duplicate_thread_path): Update.
26523 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26524 (struct switch_conv_info): Update.
26525 (gen_inbound_check): Update.
26526 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26527 (vect_do_peeling): Update.
26528 (vect_loop_versioning): Update.
26529 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26530 (optimize_mask_stores): Update.
26531 * ubsan.c (ubsan_expand_null_ifn): Update.
26532 * value-prof.c (gimple_divmod_fixed_value): Update.
26533 (gimple_divmod_fixed_value_transform): Update.
26534 (gimple_mod_pow2): Update.
26535 (gimple_mod_pow2_value_transform): Update.
26536 (gimple_mod_subtract): Update.
26537 (gimple_mod_subtract_transform): Update.
26538 (gimple_ic): Update.
26539 (gimple_stringop_fixed_value): Update.
26540 (gimple_stringops_transform): Update.
26541 * value-prof.h: Update.
26543 2017-06-29 Carl Love <cel@us.ibm.com>
26545 * config/rs6000/rs6000-c.c: Add support for built-in functions
26546 vector signed int vec_signed (vector float);
26547 vector signed long long vec_signed (vector double);
26548 vector signed int vec_signed2 (vector double, vector double);
26549 vector signed int vec_signede (vector double);
26550 vector signed int vec_signedo (vector double);
26551 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26552 instruction generator.
26553 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26554 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26555 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26557 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26558 vunsignede_v2df): Add define_expands.
26559 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26560 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26561 VEC_UNSIGNEDO): Add definitions.
26562 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26563 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26564 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26565 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26566 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26567 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26568 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26569 vec_unsignede, vec_unsignedo): Add builtin defines.
26570 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26572 * doc/extend.texi: Update the built-in documentation file for the
26573 new built-in functions.
26575 2017-06-29 Richard Biener <rguenther@suse.de>
26577 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26578 reduction chains to LOOP_VINFO_REDUCTIONS.
26579 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26580 SLP reductions after processing reduction chains.
26582 2017-06-29 Nathan Sidwell <nathan@acm.org>
26584 * builtins.c (fold_builtin_FUNCTION): Use
26585 lang_hooks.decl_printable_name.
26587 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26589 PR middle-end/81194
26590 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26591 with only one label.
26592 * stmt.c (expand_case): Assert NCASES is greater than one.
26594 2017-06-29 Richard Biener <rguenther@suse.de>
26596 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26598 (group_case_labels): Likewise.
26599 (find_taken_edge): Push sanity checking on val to workers...
26600 (find_taken_edge_cond_expr): ... here
26601 (find_taken_edge_switch_expr): ... and here, handle cases
26602 with just a default label.
26603 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26604 (group_case_labels): Likewise.
26605 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26606 group_case_labels does anything cleanup the CFG again.
26608 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26610 PR tree-optimization/81196
26611 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26612 exit condition comparing two IVs.
26614 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26616 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26617 profile to the dummy entry at the end of the list of architectures.
26618 * config/arm/arm-cpu-cdata.h: Regenerated.
26620 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26621 Michael Collison <michael.collison@arm.com>
26624 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26626 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26627 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26628 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26629 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26630 of shift when the shift amount is masked with constant equal to
26631 the size of the mode.
26632 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26635 2017-06-29 Martin Liska <mliska@suse.cz>
26637 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26639 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26642 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26645 (edge_growth_cache_entry::edge_growth_cache_entry): New
26647 (reset_edge_growth_cache): Update to use constructor.
26649 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26651 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26652 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26653 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26655 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26657 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26658 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26660 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26662 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26663 (*-linux-uclibc*): Add t-uclibc tmake_file.
26664 * config/t-musl: New.
26665 * config/t-uclibc: New.
26667 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26669 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26671 (gen_comm_data): Emit architectural setting of arch_prof.
26672 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26674 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26675 (armv8-m.base, armv8-m.main): Likewise.
26676 * arm-protos.h (arm_build_target): Add profile field.
26677 (arch_option): Likewise.
26678 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26680 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26681 arm_active_target.profile.
26683 2017-06-28 Richard Biener <rguenther@suse.de>
26685 PR middle-end/81227
26686 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26687 TYPE_OVERFLOW_WRAPS.
26688 * match.pd (negate_expr_p): Likewise.
26689 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26690 fold_build2, not fold_binary.
26692 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26694 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26695 Convert memory address to Pmode.
26696 (aarch64_print_operand): Assert MEM operands are always Pmode.
26698 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26701 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26702 Remove redundant if.
26703 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26704 * config/arm/aarch-common-protos.h
26705 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26706 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26708 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26711 * multiple_target.c (create_dispatcher_calls): Set the default
26712 clone to be static, not public.
26714 2017-06-28 Richard Biener <rguenther@suse.de>
26716 * tree-vect-loop.c (vectorizable_reduction): Move special
26717 cond reduction IV var creation ...
26718 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26719 parameter. Use STMT_VINFO_VECTYPE.
26720 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26723 2017-06-28 Martin Liska <mliska@suse.cz>
26726 * ipa-visibility.c (non_local_p): Handle visibility.
26728 2017-06-28 Martin Liska <mliska@suse.cz>
26731 * common.opt: Add IntegerRange to various options.
26732 * opt-functions.awk (integer_range_info): New function.
26733 * optc-gen.awk: Add integer_range_info to cl_options struct.
26734 * opts-common.c (decode_cmdline_option): Handle
26735 CL_ERR_INT_RANGE_ARG.
26736 (cmdline_handle_error): Likewise.
26737 * opts.c (print_filtered_help): Show valid interval in
26738 when --help is provided.
26739 * opts.h (struct cl_option): Add range_min and range_max fields.
26740 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26742 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26744 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26745 (x * C EQ/NE y * C): New transformation.
26747 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26749 * genmultilib (combination_space): Accept '+' in option names.
26751 2017-06-28 Martin Liska <mliska@suse.cz>
26754 * asan.c (instrument_derefs): Bail out inner references
26755 that are hard register variables.
26757 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26760 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26761 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26763 2017-06-28 Richard Biener <rguenther@suse.de>
26765 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26766 (vect_get_slp_defs): Adjust.
26767 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26769 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26771 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26772 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26773 (vectorizable_reduction): Adjust.
26774 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26776 (vect_get_slp_defs): Likewise.
26777 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26778 (vectorizable_bswap): Adjust.
26779 (vectorizable_call): Likewise.
26780 (vectorizable_conversion): Likewise.
26781 (vectorizable_assignment): Likewise.
26782 (vectorizable_shift): Likewise.
26783 (vectorizable_operation): Likewise.
26784 (vectorizable_store): Likewise.
26785 (vectorizable_condition): Likewise.
26786 (vectorizable_comparison): Likewise.
26788 2017-06-28 Michael Collison <michael.collison@arm.com>
26791 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
26793 (arm_gen_movmemqi): Removed unused variable 'i'.
26794 Convert 'for' loop into 'while' loop.
26795 (arm_expand_prologue): Remove last unnecessary set of insn.
26796 (thumb_pop): Remove unused variable 'pushed_words'.
26797 (thumb_exit): Remove last unnecessary set of regs_to_pop.
26799 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26801 * config/s390/predicates.md: Use s390_rel_address_ok_p.
26802 * config/s390/s390-protos.h: Add prototype of
26803 s390_rel_address_ok_p.
26804 * config/s390/s390.c (s390_got_symbol): New function.
26805 (s390_rel_address_ok_p): New function.
26806 (legitimize_pic_address): Use s390_rel_address_ok_p.
26807 (s390_load_got): Use s390_got_symbol.
26808 (s390_option_override): Issue error if
26809 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
26810 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
26812 * config/s390/s390.opt: New option mpic-data-is-text-relative.
26814 2017-06-27 Andrew Pinski <apinski@cavium.com>
26816 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
26817 (X * copysign (1.0, X)): New pattern.
26818 (X * copysign (1.0, -X)): New pattern.
26819 (copysign (-1.0, CST)): New pattern.
26821 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26823 * genmultilib (combination_space): Remove variable.
26824 Validate reuse rules against regular expression for any sequence
26825 of multilib options in any order.
26827 2017-06-27 Michael Collison <michael.collison@arm.com>
26829 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26830 call aarch64_split_simd_combine.
26831 * (aarch64_combine_internal<mode>): Delete pattern.
26832 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26833 Allow register and subreg operands.
26835 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26837 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26838 specific need, just fallback on defaults.
26839 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26841 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26842 Olivier Hainque <hainque@adacore.com>
26844 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
26846 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
26847 targets. Pick a default if no particular attempt applied.
26848 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26851 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26853 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26854 (x86_64-wrs-vxworks7): Likewise.
26856 2017-06-27 Marek Polacek <polacek@redhat.com>
26859 * ubsan.c (instrument_null): Check get_base_address's result for null.
26861 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26863 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
26865 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26867 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
26868 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
26869 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
26870 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
26871 New function types.
26872 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
26873 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
26874 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26875 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26876 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26877 BUILT_IN_FEUPDATEENV): New builtins.
26878 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26879 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26880 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26881 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26883 (builtin_structptr_types): Adjust size.
26884 * tree.c (builtin_structptr_types): Add four entries.
26886 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26887 Olivier Hainque <hainque@adacore.com>
26889 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26890 (TLS_SYM): New local macro, forcing reference to __tls__ on
26891 link command lines for VxWorks 7 RTPs, triggering initialization
26893 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26894 OS features TLS support, true for RTPs on VxWorks 7.
26895 * config/vxworks.c (vxworks_override_options): Setup emutls
26898 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26900 * predict.c (test_prediction_value_range): Use -1U instead of -1
26901 to avoid narrowing conversion warning.
26902 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26903 to avoid narrowing conversion warning.
26904 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26906 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26908 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26910 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26911 64bit configurations.
26912 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26913 (SIZE_TYPE): Likewise.
26914 * config/vxworks.c (vxworks_emutls_var_fields): Use
26915 long_unsigned_type_node instead of unsigned_type_node as the offset
26916 field type, which is "pointer" mode in emutls.c.
26918 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26921 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26923 PR middle-end/81207
26924 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26925 gimple_vuse copying separately from gimple_vdef copying.
26927 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26929 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26932 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26933 Olivier Hainque <hainque@adacore.com>
26935 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26936 for all vxworks7 targets.
26937 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26938 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26939 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26940 variations for VX6/VX7 and 32/64bits later on in ...
26941 (VXWORKS_LIB_SPEC): Leverage new macros.
26942 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26943 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26945 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26947 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26948 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26950 2017-06-26 Carl Love <cel@us.ibm.com>
26952 * config/rs6000/rs6000-c.c: Add support for built-in functions
26953 vector bool char vec_reve (vector bool char);
26954 vector signed char vec_reve (vector signed char);
26955 vector unsigned char vec_reve (vector unsigned char);
26956 vector bool int vec_reve (vector bool int);
26957 vector signed int vec_reve (vector signed int);
26958 vector unsigned int vec_reve (vector unsigned int);
26959 vector bool long long vec_reve (vector bool long long);
26960 vector signed long long vec_reve (vector signed long long);
26961 vector unsigned long long vec_reve (vector unsigned long long);
26962 vector bool short vec_reve (vector bool short);
26963 vector signed short vec_reve (vector signed short);
26964 vector double vec_reve (vector double);
26965 vector float vec_reve (vector float);
26966 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26967 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26968 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26969 (altivec_vreve): New pattern.
26970 * config/rs6000/altivec.h (vec_reve): New define.
26971 * doc/extend.texi (vec_rev): Update the built-in documentation file
26972 for the new built-in functions.
26974 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26976 PR tree-optimization/71815
26977 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26979 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26981 (slsr_process_phi): Likewise.
26982 (replace_uncond_cands_and_profitable_phis): Don't replace a
26983 multiply candidate with a stride of 1 (copy or cast).
26984 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26986 (lowest_cost_path): Likewise.
26987 (total_savings): Likewise.
26989 2017-06-26 Richard Biener <rguenther@suse.de>
26992 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26993 Use def_builtin_pure for all gather builtins.
26995 2017-06-26 Richard Biener <rguenther@suse.de>
26997 PR tree-optimization/81203
26998 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26999 non-dominating BBs.
27001 2017-06-26 Marek Polacek <polacek@redhat.com>
27004 * doc/invoke.texi: Document -Wmultistatement-macros.
27006 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27008 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27009 arm_neon_ok_no_float_abi effective target.
27011 2017-06-26 Richard Biener <rguenther@suse.de>
27013 PR tree-optimization/80928
27014 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27015 (copy_bbs): Set BB_DUPLICATED flag early.
27016 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27018 (execute_on_shrinking_pred): Likewise.
27019 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27020 BB_DUPLICATED blocks.
27021 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27022 iterate over all PHIs considering removal of *gsi.
27024 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27026 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27029 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27031 * config/rs6000/rs6000-string.c: (expand_block_clear,
27032 do_load_for_compare, select_block_compare_mode,
27033 compute_current_alignment, expand_block_compare,
27034 expand_strncmp_align_check, expand_strn_compare,
27035 expand_block_move, rs6000_output_load_multiple)
27036 Move functions related to string/block move/compare
27037 to a separate file.
27038 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27039 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27040 for this function which is now used in two files.
27041 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27042 * config.gcc: Add rs6000-string.o to extra_objs for
27043 targets powerpc*-*-* and rs6000*-*-*.
27045 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27048 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27049 32-bit, since indexed is not valid for DImode.
27050 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27051 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27052 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27053 (define_peephole2 for Altivec d-form store): Likewise.
27056 * multiple_target.c (create_dispatcher_calls): Only create the
27057 dispatcher call if the function is the default clone of a
27058 versioned function.
27060 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27062 PR middle-end/80902
27063 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27064 a call, force the call to not be a tail call.
27066 2017-06-23 Jeff Law <law@redhat.com>
27068 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27071 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27073 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27074 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27075 handling for early expansion of vector shifts (sl,sr,sra,rl).
27076 (builtin_function_type): Add vector shift right instructions
27077 to the unsigned argument list.
27079 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27081 rtl-optimizatoin/79286
27082 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27083 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27084 trap. PIC register plus a const unspec without offset can never trap.
27086 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27088 * tree.h (builtin_structptr_type): New type.
27089 (builtin_structptr_types): Declare new array.
27090 * tree.c (builtin_structptr_types): New array.
27091 (free_lang_data, build_common_tree_nodes): Use it.
27093 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27096 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27099 2017-06-22 Matt Turner <mattst88@gmail.com>
27101 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27102 Lake models to skylake case. Assume skylake for unknown
27103 models with clflushopt.
27105 2017-06-22 Jeff Law <law@redhat.com>
27107 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27108 frame sizes that do not satisfy aarch64_uimm12_shift.
27110 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27112 * profile-count.h (apply_probability,
27113 apply_scale, probability_in): Fix checks for zero.
27115 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27117 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27118 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27120 2016-06-22 Richard Biener <rguenther@suse.de>
27122 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27123 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27124 REDUC_MAX_EXPR support.
27125 (vectorizable_reduction): Likewise.
27126 (vect_create_epilog_for_reduction): Likewise.
27128 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27130 * match.pd (A / (1 << B) -> A >> B): New.
27131 * generic-match-head.c: Include optabs-tree.h.
27132 * gimple-match-head.c: Likewise.
27133 * optabs-tree.h (target_supports_op_p): New.
27134 * optabs-tree.c (target_supports_op_p): New.
27136 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27138 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27139 $gcc_cv_ld --help output.
27140 (gcc_cv_ld_demangle): Likewise.
27141 (gcc_cv_ld_eh_frame_hdr): Likewise.
27142 (gcc_cv_ld_pie): Likewise.
27143 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27144 (gcc_cv_ld_buildid): Likewise.
27145 (gcc_cv_ld_sysroot): Likewise.
27146 (ld_bndplt_support): Likewise.
27147 (ld_pushpopstate_support): Likewise.
27148 * configure: Regenerate.
27149 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27151 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27154 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27155 operands indexes to avoid gap between operands and match_dups.
27157 2017-06-21 Andrew Pinski <apinski@cavium.com>
27159 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27160 Increment Arith_shift and Arith_shift_reg by 1.
27161 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27163 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27164 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27165 (aarch64_strip_extend): Add new argument and test for it.
27166 (aarch64_cheap_mult_shift_p): New function.
27167 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27168 add a cost if it is true.
27169 Update calls to aarch64_strip_extend.
27170 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27172 2017-06-21 Andrew Pinski <apinski@cavium.com>
27174 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27176 (thunderxt88): Likewise.
27177 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27178 (thunderx_prefetch_tune): New variable.
27179 (thunderx2t99_prefetch_tune): Update for the correct values.
27180 (thunderxt88_tunings): New variable.
27181 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27182 generic_prefetch_tune.
27183 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27185 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27187 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27188 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27189 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27190 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27191 (aarch64_atomic_cas<mode>, GPI): Likewise.
27193 2017-06-21 Martin Liska <mliska@suse.cz>
27195 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27196 statements on cold and hot labels.
27197 * predict.c (tree_estimate_probability_bb): Remove the
27198 prediction from this place.
27200 2017-06-21 Martin Liska <mliska@suse.cz>
27202 PR tree-optimization/79489
27203 * gimplify.c (maybe_add_early_return_predict_stmt): New
27205 (gimplify_return_expr): Call the function.
27206 * predict.c (tree_estimate_probability_bb): Remove handling
27208 * predict.def: Update comment about early return predictor.
27209 * gimple-predict.h (is_gimple_predict): New function.
27210 * predict.def: Change default value of early return to 66.
27211 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27213 * passes.def: Put pass_strip_predict_hints to the beginning of
27216 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27218 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27219 FUNCTION_DECL declarations.
27220 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27222 (dwaf2out_decl): Likewise.
27223 * godump.c (go_early_global_decl): Skip call to the real debug hook
27224 for FUNCTION_DECL declarations.
27225 * passes.c (rest_of_decl_compilation): Skip call to the
27226 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27227 -fdump-go-spec is passed.
27229 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27231 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27232 Reorder for compactness.
27233 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27234 (def_builtin_pure, def_builtin_pure2): New functions.
27235 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27237 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27239 * match.pd (nop_convert): New predicate.
27240 ((A +- CST1) +- CST2): Allow some NOP conversions.
27242 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27245 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27246 with ctors/dtors if GOVD_SHARED is set.
27248 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27250 * config/aarch64/aarch64.md (movti_aarch64):
27251 Emit mov rather than orr.
27252 (movtf_aarch64): Likewise.
27253 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27254 Emit mov rather than orr.
27256 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27258 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27259 Swap alternatives, make integer dup more expensive.
27261 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27263 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27264 Return true for non-tls symbols.
27266 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27268 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27269 (cortex-a75): Likewise.
27270 (cortex-a75.cortex-a55): Likewise.
27271 * config/aarch64/aarch64-tune.md: Regenerate.
27272 * doc/invoke.texi (-mtune): Document new values for -mtune.
27274 2017-06-21 Tom de Vries <tom@codesourcery.com>
27276 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27277 stack_size feature.
27278 (Effective-Target Keywords, Other attributes): Suggest using
27279 dg-add-options stack_size feature to get stack limit in stack_size
27280 effective target documentation.
27282 2017-06-21 Julian Brown <julian@codesourcery.com>
27283 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27285 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27286 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27287 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27289 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27290 attribute type list for neon_multiply.
27291 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27292 attribute type list for neon_multiply.
27293 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27294 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27295 attribute type list for neon_multiply.
27296 * config/arm/types.md (crypto_pmull): Add.
27297 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27298 attribute type list.
27300 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27302 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27305 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27307 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27308 to make sure not to dereference a NULL cost_classes_ptr pointer.
27310 2017-06-20 Carl Love <cel@us.ibm.com>
27312 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27313 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27314 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27315 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27316 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27317 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27318 VMULOSW): New enum "unspec" values.
27319 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27320 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27321 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27322 altivec_vmulosw): New patterns.
27323 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27324 VMULOSW): Add definitions.
27326 2017-06-20 Julia Koval <julia.koval@intel.com>
27328 * config/i386/i386.c: Fix rounding expand for new pattern.
27329 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27331 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27333 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27334 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27336 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27338 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27341 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27343 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27344 architecture, then by implementer ID.
27345 * config/aarch64/aarch64-tune.md: Regenerate.
27347 2017-06-20 Richard Biener <rguenther@suse.de>
27349 PR middle-end/81097
27350 * fold-const.c (split_tree): Fold to type before negating.
27352 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27354 * diagnostic-show-locus.c
27355 (selftest::test_fixit_deletion_affecting_newline): New function.
27356 (selftest::diagnostic_show_locus_c_tests): Call it.
27358 2017-06-20 Andreas Schwab <schwab@suse.de>
27361 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27364 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27366 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27367 __ARM_FEATURE_COPROC according to support.
27369 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27371 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27372 Rewritten to avoid overflow for > 32-bit pointers.
27375 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27376 by removing enum keyword.
27377 (ubsan_type_descriptor): Likewise. Formatting fix.
27380 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27381 splitter): Require TARGET_SSE2 in the condition.
27383 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27386 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27387 for doing vector set of SFmode on ISA 3.0.
27388 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27389 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27391 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27392 SFmode value into a V4SF variable that was extracted from another
27393 V4SF variable without converting the element to double precision
27394 and back to single precision vector format.
27395 (vsx_insert_extract_v4sf_p9_2): Likewise.
27397 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27399 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27400 in UWHI to avoid undefined overflow.
27403 * ubsan.h (enum ubsan_encode_value_phase): New.
27404 (ubsan_encode_value): Change second argument to
27405 enum ubsan_encode_value_phase with default value of
27406 UBSAN_ENCODE_VALUE_GENERIC.
27407 * ubsan.c (ubsan_encode_value): Change second argument to
27408 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27409 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27410 create_tmp_var_raw instead of create_tmp_var and use a
27412 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27413 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27414 ubsan_encode_value callers.
27417 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27418 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27419 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27421 2017-06-19 Richard Biener <rguenther@suse.de>
27423 PR middle-end/81118
27424 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27425 estimates if we changed anything.
27427 2017-06-19 Richard Biener <rguenther@suse.de>
27429 PR tree-optimization/80887
27430 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27431 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27432 simplified lookups, then reset mprts_hook.
27433 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27435 (try_to_simplify): Likewise.
27437 2017-06-19 Martin Liska <mliska@suse.cz>
27440 * gimplify.c (gimplify_switch_expr):
27441 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27443 2017-06-19 Martin Liska <mliska@suse.cz>
27445 * doc/install.texi: Document that PGO runs in 4 stages.
27447 2017-06-19 Martin Liska <mliska@suse.cz>
27450 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27451 to dispatcher function name.
27452 * multiple_target.c (replace_function_decl): New function.
27453 (create_dispatcher_calls): Redirect both edges and references.
27455 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27457 * profile-count.c (profile_count::dump): Dump quality.
27458 (profile_count::differs_from_p): Update for unsigned val.
27459 * profile-count.h (profile_count_quality): New enum.
27460 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27462 2017-06-19 Richard Biener <rguenther@suse.de>
27464 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27465 struct function as arg.
27466 (estimate_numbers_of_iterations): Export overload with loop arg.
27467 (free_numbers_of_iterations_estimates_loop): Use an overload of
27468 free_numbers_of_iterations_estimates instead.
27469 * tree-cfg.c (remove_bb): Adjust.
27470 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27471 * tree-parloops.c (gen_parallel_loop): Likewise.
27472 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27474 (tree_unroll_loops_completely): Likewise.
27475 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27476 Use an overload instead and export.
27477 (estimated_loop_iterations): Adjust.
27478 (max_loop_iterations): Likewise.
27479 (likely_max_loop_iterations): Likewise.
27480 (estimate_numbers_of_iterations): Take struct function as arg
27482 (loop_exits_before_overflow): Adjust.
27483 (free_numbers_of_iterations_estimates_loop): Use an overload.
27484 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27485 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27487 2017-06-19 Richard Biener <rguenther@suse.de>
27490 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27491 RANGE_EXPR conservatively.
27493 2017-06-16 Carl Love <cel@us.ibm.com>
27495 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27496 definitions for vec_float, vec_float2, vec_floato,
27497 vec_floate built-ins.
27498 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27499 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27501 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27502 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27503 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27504 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27505 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27506 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27507 vec_floato): Add builtin defines.
27508 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27509 Update the built-in documentation file for the new built-in
27512 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27514 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27515 (mthumb): Mark as the negative of -marm.
27517 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27519 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27521 (ARM Options, -mtune): Document that this accepts the same
27522 extension options as -mcpu.
27523 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27525 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27527 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27528 permitted extensions.
27530 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27532 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27533 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27534 (armv8-m.main): Add option +nodsp.
27535 * config/arm/arm-cpu-cdata.h: Regenerated.
27537 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27539 * config/arm/t-fuchsia: New file.
27540 * config.gcc (arm*-*-fuchsia*): Use it.
27542 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27544 * config/arm/t-symbian: Rewrite for new option infrastructure.
27546 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27548 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27549 (MULTILIB_REQUIRED): Likewise.
27551 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27553 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27554 (MULTILIB_RESUE): Likewise.
27555 (MULTILIB_MATCHES): Likewise.
27556 (MULTLIB_REQUIRED): Likewise.
27558 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27560 * config/arm/t-rtems: Rewrite for new option framework.
27562 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27564 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27565 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27566 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27567 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27568 * config/arm/t-multilib: ... here.
27569 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27570 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27571 armv7-a and armv8*-a when A-profile libraries have not been built.
27572 * config/arm/t-rmprofile: Rewrite.
27574 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27576 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27577 with a backslash. Remove the backslash after substituting unescaped
27579 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27581 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27583 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27584 the driver to pass through the default setting of -mfloat-abi.
27585 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27587 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27588 (all_feat_combs): New rule.
27589 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27591 * config/arm/t-aprofile: Rewrite.
27593 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27595 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27596 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27597 fpu is not specified by the user/command-line.
27598 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27599 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27600 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27601 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27602 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27603 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27605 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27607 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27608 * config/arm/t-arm-elf: Rewritten.
27610 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27612 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27613 have some floating-point instructions.
27614 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27615 (TARGET_MAYBE_HARD_FLOAT): New macro.
27616 * config/arm/arm-builtins.c (arm_init_builtins): Use
27617 TARGET_MAYBE_HARD_FLOAT.
27618 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27620 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27622 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27623 (configargs.h): Include it.
27624 (arm_print_hint_for_fpu_option): New function.
27625 (arm_parse_fpu_option): New function.
27626 (candidate_extension): New class.
27627 (arm_canon_for_multilib): New function.
27628 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27629 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27630 (ARCH_CANONICAL_SPECS): New macro.
27631 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27633 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27635 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27636 are set after handling multilib fragments. Set target_cpu_default2
27639 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27641 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27643 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27644 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27646 (check_arch): Likewise.
27647 * config/arm/arm.c (arm_configure_build_target): Handle
27648 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27649 options in the default.
27651 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27653 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27654 when an option is an alias of another.
27655 * config/arm/parsecpu.awk (optalias): New parser token.
27656 (gen_comm_data): Mark non-alias options as such. Emit entries
27657 for extension aliases.
27658 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27659 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27660 (armv6kz, armv6zk, armv6t2): Likewise.
27661 (armv7): Make vfpv3-d16 an alias.
27662 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27664 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27665 Sort in canonical order.
27666 (armv8-a): Sort in canonical order.
27667 (armv8.1-a, armv8.2-a): Likewise.
27668 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27670 (cortex-a9): Sort in canonical order.
27671 * config/arm/arm.c (selftests.h): Include it.
27672 (arm_test_cpu_arch_data): New function.
27673 (arm_run_self_tests): New function.
27674 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27675 (targetm): Move declaration to the end of the file.
27676 * arm-cpu-cdata.h: Regenerated.
27678 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27680 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27681 call to target_mode_check describing the type of option passed.
27682 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27683 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27684 arm_parse_cpu_option_name to match parameters against list of
27686 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27687 arm_arch_core_flags data structure.
27688 * config/arm/arm-cpu_cdata.h: Regenerated.
27690 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27692 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27694 (arm_print_hint_for_cpu_option): Likewise.
27695 (arm_print_hint_for_arch_option): Likewise.
27696 (arm_parse_cpu_option_name): Likewise.
27697 (arm_parse_arch_option_name): Likewise.
27698 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27699 of entries in the all_fpus list.
27700 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27701 (arm_parse_cpu_option_name): Declare.
27702 (arm_parse_arch_option_name): Declare.
27703 (arm_parse_option_features): Declare.
27704 (arm_intialize_isa): Declare.
27705 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27707 (gen_comm_data): ... here. Make definitions non-static.
27708 * config/arm/arm-cpu-data.h: Regenerated.
27709 * config/arm/arm-cpu-cdata.h: Regenerated.
27711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27713 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27714 (cpu_arch_extension): New structure.
27715 (cpu_arch_option, arch_option, cpu_option): New structures.
27716 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27717 architecture types.
27718 (gen_data): Generate new format data tables.
27719 * config/arm/arm.c (cpu_tune): New structure.
27720 (cpu_option, processors): Delete.
27721 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27722 (arm_print_hint_for_cpu_option): ... this and ...
27723 (arm_print_hint_for_arch_option): ... this.
27724 (arm_parse_arch_cpu_name): Delete. Replace with ...
27725 (arm_parse_cpu_option_name): ... this and ...
27726 (arm_parse_arch_option_name): ... this.
27727 (arm_unrecognized_feature): Change type of target parameter to
27729 (arm_parse_arch_cpu_features): Delete. Replace with ...
27730 (arm_parse_option_features): ... this.
27731 (arm_configure_build_target): Rework to use new configuration data
27733 (arm_print_tune_info): Rework for new configuration data tables.
27734 * config/arm/arm-cpu-data.h: Regenerated.
27735 * config/arm/arm-cpu.h: Regenerated.
27737 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27739 * Makefile.in (OBJS): Move sbitmap.o from here ...
27740 (OBJS-libcommon): ... to here.
27742 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27744 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27745 (ISA_ALL_CRYPTO): New macro.
27746 (ISA_ALL_SIMD): New macro
27747 (ISA_ALL_FP): New macro.
27748 * config/arm/arm.c (fpu_bitlist): Update initializer.
27749 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27751 (arm9e): Add fpu. Add option for nofp
27752 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27753 (arm926ej-s, arm1026ej-s): Likewise.
27754 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27755 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27756 neon-fp16, neon-vfpv4, nofp and nosimd.
27757 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27758 (cortex-a8): Add fpu. Add option for nofp.
27759 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27760 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27761 (cortex-r4f): Add fpu.
27762 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27763 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27765 (cortex-r8): Likewise.
27766 (cortex-m4): Add fpu. Add option for nofp.
27767 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27768 (cortex-a17.cortex-a7): Likewise.
27769 (cortex-a32): Add fpu. Add options for crypto and nofp.
27770 (cortex-a35, cortex-a53): Likewise.
27771 (cortex-a57): Add fpu. Add option for crypto.
27772 (cortex-a72, cortex-a73): Likewise.
27773 (exynos-m1): Likewise.
27774 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27775 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27776 (cortex-m33): Add fpu. Add option for nofp.
27777 * config/arm/arm-cpu-cdata.h: Regenerated
27778 * config/arm/arm-cpu-data.h: Regenerated.
27780 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27782 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27783 (armv5te, armv5tej): Likewise.
27784 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27785 (armv7): Add options fp and vfpv3-d16.
27786 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27787 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27789 (armv7ve): Likewise.
27790 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
27791 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
27792 (armv8-a): Add nocrypto option.
27793 (armv8.1-a, armv8.2-a): Likewise.
27794 (armv8-m.main): add options fp, fp.dp and nofp.
27796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27798 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
27800 (armv8-a+crc): Delete.
27801 (armv8.1-a): Add options simd, crypto and nofp.
27802 (armv8.2-a): Add options fp16, simd, crypto and nofp.
27803 (armv8.2-a+fp16): Delete.
27804 (armv8-m.main): Add option dsp.
27805 (armv8-m.main+dsp): Delete.
27806 (cortex-a8): Add fpu. Add nofp option.
27807 (cortex-a9): Add fpu. Add nofp and nosimd options.
27808 * config/arm/parsecpu.awk (gen_data): Generate option tables and
27809 link to main cpu and architecture data structures.
27810 (gen_comm_data): Only put isa attributes from the main architecture
27812 (option): New statement for architecture and CPU entries.
27813 * arm.c (struct cpu_option): New structure.
27814 (struct processors): Add entry for options.
27815 (arm_unrecognized_feature): New function.
27816 (arm_parse_arch_cpu_name): Ignore any characters after the first
27818 (arm_parse_arch_cpu_feature): New function.
27819 (arm_configure_build_target): Separate out any CPU and architecture
27820 features and parse separately. Don't error out if -mfpu=auto is
27821 used with only an architecture string.
27822 (arm_print_asm_arch_directives): New function.
27823 (arm_file_start): Call it.
27824 * config/arm/arm-cpu-cdata.h: Regenerated.
27825 * config/arm/arm-cpu-data.h: Likewise.
27826 * config/arm/arm-tables.opt: Likewise.
27828 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27830 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27831 assembler when it is not -mfpu=auto.
27833 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27835 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27836 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27837 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27838 (ASM_CPU_SPEC): Rewrite.
27839 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27840 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27841 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27842 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
27843 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
27844 copied string is NUL-terminated. Also strip any characters prefixed
27846 (arm_rewrite_selected_arch): New function.
27847 (arm_rewrite_march): New function.
27849 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27851 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27852 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27853 (march, mcpu, mtune): Convert to string-based options.
27854 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27855 (arm_parse_arch_cpu_name): New function.
27856 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
27857 identify selected architecture or CPU.
27858 (arm_option_save): New function.
27859 (TARGET_OPTION_SAVE): Redefine.
27860 (arm_option_restore): Restore string options.
27861 (arm_option_print): Print string options.
27863 2017-06-16 Martin Sebor <msebor@redhat.com>
27865 PR tree-optimization/80933
27866 PR tree-optimization/80934
27867 * builtins.c (fold_builtin_3): Do not handle bcmp here.
27868 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
27869 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
27870 (gimple_fold_builtin): Call them.
27872 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27874 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27875 as unlikely; update profile.
27877 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27879 * predict.c (force_edge_cold): Handle declaring edges impossible
27882 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27884 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27886 (try_unroll_loop_completely): Fix reporting.
27888 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27890 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27892 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27895 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27896 if given a non-constant argument for an intrinsic which requires a
27899 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27901 * profile.c (compare_freqs): New function.
27902 (branch_prob): Sort edge list.
27903 (find_spanning_tree): Assume that the list is priority sorted.
27905 2017-06-16 Richard Biener <rguenther@suse.de>
27907 PR tree-optimization/81090
27908 * passes.def (pass_record_bounds): Remove.
27909 * tree-pass.h (make_pass_record_bounds): Likewise.
27910 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27911 make_pass_record_bounds): Likewise.
27912 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27913 not free niter estimates at the beginning but at the end.
27914 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27916 2017-06-16 Richard Biener <rguenther@suse.de>
27918 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27919 initializer to workaround ICE in host GCC 4.8.
27921 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27923 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27925 (clone_inlined_nodes): Update.
27927 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27929 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27930 prefetch settings, and enable prefetching by default at -O3.
27932 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27934 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27935 Set flag_prefetch_loop_arrays according to tuning data.
27937 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27939 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27940 New tune structure.
27941 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27942 [Unrelated to main purpose of the patch] Place the pointer field last
27943 to enable type checking errors when tune structure are wrongly merged.
27944 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27945 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27946 (thunderx2t99_prefetch_tune): New tune constants.
27947 (tune_params *_tunings): Update all tunings (no functional change).
27948 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27949 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27950 from tunings structures.
27952 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27955 * ubsan.c (instrument_null): Add T argument, use it instead
27956 of computing it based on IS_LHS.
27957 (instrument_object_size): Likewise.
27958 (pass_ubsan::execute): Adjust instrument_null and
27959 instrument_object_size callers to pass gimple_get_lhs or
27960 gimple_assign_rhs1 result to it. Use instrument_null instead of
27961 calling get_base_address and instrument_mem_ref. Handle
27962 aggregate call arguments for object-size sanitization.
27964 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27966 PR tree-optimization/81089
27967 * tree-vrp.c (is_masked_range_test): Validate operands of
27970 2017-06-15 Martin Sebor <msebor@redhat.com>
27973 * dumpfile.c (dump_register): Avoid calling memset to initialize
27974 a class with a default ctor.
27975 * gcc.c (struct compiler): Remove const qualification.
27976 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27977 * hash-table.h: Ditto.
27978 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27980 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27981 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27983 * params.h (struct param_info): Make struct members non-const.
27984 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27985 with default initialization.
27986 * vec.h (vec_copy_construct, vec_default_construct): New helper
27988 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27989 with vec_copy_construct.
27990 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27991 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27992 * doc/invoke.texi (-Wclass-memaccess): Document.
27994 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27996 * emit-rtl.h (is_leaf): Update comment about local
27997 register allocator.
27999 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28002 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28003 for a variable to have a section before checking if the section has a
28005 Set section to.persistent if persistent attribute is set.
28006 Warn if .persistent attribute is used on an automatic variable.
28008 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28010 PR rtl-optimization/80474
28011 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28013 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28015 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28018 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28020 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28021 (casesi): Emit bounds checking as RTL.
28022 (casesi_internal_mips16_<mode>): Remove bounds checking.
28024 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28026 * config/xtensa/xtensa.c (xtensa_option_override): Append
28027 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28028 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28029 xtensa_doloop_hooks): Define unconditionally.
28030 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28032 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28033 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28034 for it in xtensa_option_override.
28035 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28036 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28038 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28040 * doc/cppopts.texi: Document '-' special value to -MF.
28042 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28044 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28045 (cortex_a53_fconst): Likewise.
28046 (cortex_a53_fpmul): Likewise.
28047 (cortex_a53_f_load_64): Likewise.
28048 (cortex_a53_f_load_many): Likewise.
28049 (cortex_a53_advsimd_alu): Likewise.
28050 (cortex_a53_advsimd_alu_q): Likewise.
28051 (cortex_a53_advsimd_mul): Likewise.
28052 (cortex_a53_advsimd_mul_q): Likewise.
28053 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28054 Add missing fmul, r2f_cvt and fconst cases.
28056 2017-06-14 Richard Biener <rguenther@suse.de>
28058 PR middle-end/81088
28059 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28061 (fold_binary_loc): When associating do not treat pre-existing
28062 TREE_OVERFLOW on literal constants as a reason to allow
28063 TREE_OVERFLOW on associated literal constants.
28065 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28067 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28068 (MASK_FEATURES): New macro.
28069 * config/sparc/sparc.c (sparc_option_override): Remove the special
28070 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28072 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28074 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28075 a division of 0 if non-call exceptions are enabled.
28077 2017-06-14 Andrew Pinski <apinski@cavium.com>
28078 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28081 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28082 Improve vector initialization code gen for only variable case.
28084 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28086 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28088 2017-06-14 Richard Biener <rguenther@suse.de>
28090 PR tree-optimization/81083
28091 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28094 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28096 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28097 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28098 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28099 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28100 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28101 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28103 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28105 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28106 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28108 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28110 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28112 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28114 * config/rs6000/t-rtems: Don't handle SPE.
28116 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28118 * config/rs6000/t-linux: Don't handle SPE.
28120 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28122 * config/rs6000/eabispe.h: Delete file.
28124 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28126 * config/rs6000/t-spe: Delete file.
28128 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28130 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28131 (rs6000_legitimate_offset_address_p): Return false for anything in
28132 V2SImode or V2SFmode.
28134 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28136 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28137 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28138 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28140 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28141 (rs6000_legitimate_offset_address_p): Ditto.
28142 (rs6000_emit_move): Ditto.
28143 (rs6000_init_builtins): Remove V4HI_type_node.
28145 2017-06-13 Martin Liska <mliska@suse.cz>
28148 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28149 (gate_asan): Likewise.
28150 * asan.h (asan_no_sanitize_address_p): Remove the function.
28151 (sanitize_flags_p): New function.
28152 * builtins.def: Fix coding style.
28153 * common.opt: Use renamed enum value.
28154 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28155 * doc/extend.texi: Document no_sanitize attribute.
28156 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28157 to SANITIZE_UNDEFINED_NONDEFAULT.
28158 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28159 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28160 Use sanitize_flags_p.
28161 * gimplify.c (gimplify_function_tree): Likewise.
28162 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28163 * opts.c (parse_no_sanitize_attribute): New function.
28164 (common_handle_option): Use renamed enum value.
28165 * opts.h (parse_no_sanitize_attribute): Declare.
28166 * tree.c (sanitize_flags_p): New function.
28167 * tree.h: Declared here.
28168 * tsan.c: Use sanitize_flags_p.
28169 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28170 (instrument_mem_ref): Likewise.
28171 (instrument_bool_enum_load): Likewise.
28172 (do_ubsan_in_current_function): Remove the function.
28173 (pass_ubsan::execute): Use sanitize_flags_p.
28174 * ubsan.h: Remove do_ubsan_in_current_function
28175 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28176 (dump_function_to_file): Use it here.
28178 2017-06-13 Martin Jambor <mjambor@suse.cz>
28180 PR tree-optimization/80803
28181 PR tree-optimization/81063
28182 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28183 (propagate_subaccesses_across_link): Enqueue subtree whenever
28184 necessary instead of relying on the caller.
28186 2017-06-13 Martin Jambor <mjambor@suse.cz>
28188 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28189 that have a first_link.
28190 (sort_and_splice_var_accesses): Do not check first_link before
28192 (subtree_mark_written_and_enqueue): Likewise.
28193 (propagate_all_subaccesses): Likewise and do not stop at first
28194 parent with a first_link.
28196 2017-06-13 Martin Jambor <mjambor@suse.cz>
28198 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28201 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28203 * match.pd: New pattern.
28205 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28207 * tree-vrp.c (is_masked_range_test): New function.
28208 (register_edge_assert_for): Determine ranges for
28211 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28213 PR tree-optimization/67328
28214 * fold-const.c (maskable_range_p): New function.
28215 (build_range_check): Generate bittests if possible.
28217 2017-06-13 Martin Liska <mliska@suse.cz>
28219 * gimple-pretty-print.c (dump_probability): Add new argument.
28220 (dump_edge_probability): Dump both probability and count.
28221 (dump_gimple_label): Likewise.
28222 (dump_gimple_bb_header): Likewise.
28224 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28227 * config/avr/avr-devices.c: Fix indentation.
28228 * config/avr/gen-avr-mmcu-specs.c: Dito.
28230 2017-06-13 Richard Biener <rguenther@suse.de>
28232 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28233 instead get vector type from stmt_info.
28234 (vectorizable_reduction): Adjust. Remove dead code.
28236 2017-06-13 Richard Biener <rguenther@suse.de>
28238 PR middle-end/81065
28239 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28240 case of C * (x * C2 + C3).
28241 (fold_addr_of_array_ref_difference): Properly fold index difference.
28243 2017-06-12 David S. Miller <davem@davemloft.net>
28246 * config/sparc/sparc.md (return expander): Emit frame blockage if
28247 function uses alloca.
28249 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28251 * combine.c (make_field_assignment): Check len rather than the mode
28252 precision when calling force_to_mode.
28254 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28256 Support multilibs and devices that see flash in RAM address range.
28259 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28260 (avr_mcu_t) <flash_pm_offset>: New field.
28261 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28262 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28263 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28264 (AVR_TINY_PM_OFFSET): Remove macro.
28265 * config/avr/avr.opt (-mshort-calls): New option.
28266 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28267 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28268 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28269 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28270 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28271 instead of avr_arch->have_jmp_call.
28272 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28273 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28274 avr_arch->flash_pm_offset to define.
28275 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28276 new field flash_pm_offset. Add entry for avrxmega3.
28277 (avr_texinfo): Add entry for avrxmega3.
28278 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28279 attiny212, attiny214,
28280 attiny412, attiny414, attiny416, attiny417,
28281 attiny814, attiny816, attiny817,
28282 attiny1614, attiny1616, attiny1617,
28283 attiny3214, attiny3216, attiny3217.
28284 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28285 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28286 (avr_print_operand_address) [AVR_TINY]: Same.
28287 (avr_asm_init_sections) <readonly_data_section>: Only patch
28288 callback if avr_arch->flash_pm_offset = 0.
28289 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28290 for rodata if avr_arch->flash_pm_offset != 0.
28291 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28292 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28293 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28294 (m_options): Append opt_rcall.
28295 (m_dirnames): Append dir_rcall.
28296 * config/avr/t-multilib: Regenerate.
28298 * configure.ac [target=avr]: Check whether avrxmega3 default
28299 linker description file works as needed.
28300 * configure: Regenerate.
28301 * doc/avr-mmcu.texi: Regenerate.
28302 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28303 <__AVR_ARCH__>: Document avrxmega3 and 103.
28304 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28305 <__AVR_SHORT_CALLS__>: Document it.
28306 <__AVR_PM_BASE_ADDRESS__>: Document it.
28307 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28308 (AVR Variable Attributes) <progmem>: Document this is
28309 not needed for avrxmega3.
28310 (AVR Named Address Spaces) <__flash>: Dito.
28312 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28314 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28316 2017-06-12 Doug Rupp <rupp@adacore.com>
28318 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28319 Append vxworks-stdint.h to the tm_file list.
28320 * config/vxworks-stdint.h: New file.
28322 2017-06-12 Martin Liska <mliska@suse.cz>
28324 PR tree-optimization/81041
28325 * tree-profile.c (gimple_gen_ic_func_profiler):
28326 Create an extra BB in profile-generate
28327 (gimple_gen_time_profiler): Likewise.
28329 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28331 PR tree-optimization/81003
28332 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28333 (update_range_test): Use it instead of force_gimple_operand_gsi.
28335 2017-06-12 Richard Biener <rguenther@suse.de>
28337 PR tree-optimization/81053
28338 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28339 with backedge value not defined in loop. Simplify def stmt
28342 2017-06-11 Tom de Vries <tom@codesourcery.com>
28345 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28347 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28348 nvptx_cannot_force_const_mem.
28350 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28352 * opts.c (finish_options): Move test for flag_split_stack after
28353 it has been initialized.
28355 2017-06-11 Jason Merrill <jason@redhat.com>
28357 * tree.h (id_equal): New.
28358 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28359 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28360 instead of strcmp of IDENTIFIER_POINTER.
28362 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28364 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28365 (mark_all_inlined_calls_cdtor): Fix formating.
28366 (inline_transform): Rescale profile before inlining.
28368 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28370 * cgraph.h (cgraph_edge::clone): Update prototype.
28371 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28372 (cgraph_node::create_clone): Update.
28373 (cgraph_node::create_version_clone): Update.
28374 * tree-inline.c (copy_bb): Update.
28375 (expand_call_inline): Update.
28377 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28379 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28380 factored out from ...
28381 (rs6000_emit_prologue): ... here.
28383 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28385 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28386 factored out from ...
28387 (rs6000_emit_prologue): ... here.
28389 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28391 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28393 (handle_missing_profiles): Fix computation of tp_first_run.
28394 (counts_to_freqs): Do not touch freqs when count is 0.
28396 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28398 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28401 2017-06-10 Tom de Vries <tom@codesourcery.com>
28403 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28404 attributes): Document signal effective target.
28406 2017-06-10 Tom de Vries <tom@codesourcery.com>
28408 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28409 Document effective target stack_size.
28411 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28413 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28414 to the edit_context if they can be auto-applied.
28416 2017-06-9 Ian Lance Taylor <iant@golang.org>
28418 * opts.c (finish_options): If -fsplit-stack, disable implicit
28419 -forder-blocks-and-partition.
28420 * doc/invoke.texi (Optimize Options): Document that when using
28421 -fsplit-stack -forder-blocks-and-partition is not implicitly
28424 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28426 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28427 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28428 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28429 * builtins.def (abort, trap, unreachable): Declare cold.
28430 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28431 * tree-core.h (ECF_COLD): New.
28432 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28433 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28435 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28437 * predict.c (unlikely_executed_stmt_p): Cleanup.
28439 2017-06-09 Richard Biener <rguenther@suse.de>
28441 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28442 model if the ref is always written to.
28444 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28446 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28448 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28450 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28453 2017-06-09 Tom de Vries <tom@codesourcery.com>
28456 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28457 "target cannot support label values" when encountering LABEL_REF.
28459 2017-06-09 Martin Liska <mliska@suse.cz>
28461 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28462 (gimple_gen_ic_func_profiler): Emit direct comparison
28463 of __gcov_indirect_call_callee with NULL.
28464 (gimple_gen_time_profiler): Change probability from
28465 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28467 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28469 * profile.c (edge_gcov_counts): Turn to pointer.
28470 (compute_branch_probabilities, compute_branch_probabilities): Update.
28471 (branch_prob): Do not clear edge_gcov_count.
28472 * profile.h (edge_gcov_counts): Turn to pointer.
28473 (edge_gcov_count): Update.
28475 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28477 * gimple.h (gimple_check_failed): Mark cold.
28479 2017-06-09 Richard Biener <rguenther@suse.de>
28481 PR tree-optimization/66623
28482 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28483 refactor check_reduction into two parts, properly computing
28484 whether we have to check reduction validity for outer loop
28487 2017-06-09 Richard Biener <rguenther@suse.de>
28489 PR tree-optimization/79483
28490 * graphite-scop-detection.c (order): New global.
28491 (get_order): Compute bb to order mapping that satisfies code
28492 generation constraints.
28493 (cmp_pbbs): New helper.
28494 (build_scops): Start domwalk at entry block, sort generated
28497 2017-06-09 Richard Biener <rguenther@suse.de>
28499 PR middle-end/81007
28500 * ipa-polymorphic-call.c
28501 (ipa_polymorphic_call_context::restrict_to_inner_class):
28502 Skip FIELD_DECLs with error_mark_node type.
28503 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28506 2017-06-09 Martin Liska <mliska@suse.cz>
28508 * predict.c (struct branch_predictor): New struct.
28509 (test_prediction_value_range): New test.
28510 (predict_c_tests): New function.
28511 * selftest-run-tests.c (selftest::run_tests): Run the function.
28512 * selftest.h: Declare new tests.
28514 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28517 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28518 gen_add3_insn did not fail.
28519 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28520 r0, construct that number in a temporary reg and add that reg to r0.
28521 If asked to put the result in r0 as well, fail.
28523 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28525 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28526 for early expansion of vec_eqv.
28528 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28530 PR middle-end/81005
28531 * ubsan.c (instrument_null): Avoid pointless code temporary.
28532 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28534 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28538 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28541 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28544 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28546 * predict.c (unlikely_executed_edge_p): Move ahead.
28547 (probably_never_executed_edge_p): Use it.
28549 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28551 PR middle-end/79988
28552 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28553 gimple_call_builtin_p call.
28555 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28557 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28558 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28559 rtl_check_failed_type2, rtl_check_failed_code1,
28560 rtl_check_failed_code2, rtl_check_failed_code_mode,
28561 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28562 rtvec_check_failed_bounds, rtl_check_failed_flag,
28563 _fatal_insn_not_found, _fatal_insn): Likewise.
28564 * tree.h (tree_contains_struct_check_failed,
28565 tree_check_failed, tree_not_check_failed,
28566 tree_class_check_failed, tree_range_check_failed,
28567 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28568 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28569 tree_operand_check_failed, omp_clause_check_failed,
28570 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28573 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28575 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28576 flag_branch_probabilities.
28577 * ipa-inline.c (edge_badness): Likewise.
28578 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28579 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28580 * predict.c (maybe_hot_frequency_p): Likewise.
28581 (probably_never_executed): Likewise.
28582 * sched-ebb.c (schedule_ebbs): Likewise.
28583 * sched-rgn.c (find_single_block_region): Likewise.
28584 * tracer.c (tail_duplicate): Likewise.
28586 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28588 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28589 longer requires x_flag_profile_use.
28591 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28593 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28594 instead of flag_reorder_blocks_and_partition.
28595 * dbxout.c (dbxout_function_end): Likewise.
28596 * dwarf2out.c (gen_subprogram_die): Likewise.
28597 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28598 * hw-doloop.c (reorg_loops): Likewise.
28599 * varasm.c (assemble_start_function,
28600 assemble_end_function): Likewise.
28601 (decide_function_section): Do not check for
28602 flag_reorder_blocks_and_partition.
28604 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28606 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28608 (chkp_get_hard_register_fake_addr_expr): Ditto.
28609 (chkp_build_addr_expr): Add check for hard reg case.
28610 (chkp_parse_array_and_component_ref): Ditto.
28611 (chkp_find_bounds_1): Ditto.
28612 (chkp_process_stmt): Don't generate bounds store for
28615 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28617 * predict.c (maybe_hot_bb_p): Do not check profile status.
28618 (maybe_hot_edge_p): Likewise.
28619 (probably_never_executed): Check for zero counts even if profile
28621 (unlikely_executed_edge_p): New function.
28622 (unlikely_executed_stmt_p): New function.
28623 (unlikely_executed_bb_p): New function.
28624 (set_even_probabilities): Use unlikely predicates.
28625 (combine_predictions_for_bb): Likewise.
28626 (predict_paths_for_bb): Likewise.
28627 (predict_paths_leading_to_edge): Likewise.
28628 (determine_unlikely_bbs): New function.
28629 (estimate_bb_frequencies): Use it.
28630 (compute_function_frequency): Use zero counts even if profile is
28632 * profile-count.h: Fix typo.
28634 2017-08-08 Julia Koval <julia.koval@intel.com>
28636 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28637 _mm512_mask_cvtsepi16_storeu_epi8,
28638 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28639 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28640 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28641 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28642 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28643 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28644 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28645 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28646 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28647 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28648 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28649 __builtin_ia32_pmovuswb256mem_mask,
28650 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28651 __builtin_ia32_pmovwb512mem_mask): New builtins.
28653 2017-08-08 Julia Koval <julia.koval@intel.com>
28655 PR target/73350,80862
28656 * config/i386/subst.md (round): Fix round pattern.
28657 * config/i386/i386.c (ix86_erase_embedded_rounding):
28658 Fix erasing rounding for the fixed pattern.
28660 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28662 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28664 2017-06-08 Martin Liska <mliska@suse.cz>
28666 PR gcov-profile/80911
28667 * gcov.c (block_info::block_info): New constructor.
28669 2017-06-07 Carl Love <cel@us.ibm.com>
28671 * config/rs6000/rs6000-c: The return type of the following
28672 built-in functions was implemented as int not long long. Fix sign
28673 of return value for the unsigned version of vec_mulo and vec_mule.
28674 vector unsigned long long vec_bperm (vector unsigned long long,
28675 vector unsigned char)
28676 vector signed long long vec_mule (vector signed int,
28678 vector unsigned long long vec_mule (vector unsigned int,
28679 vector unsigned int)
28680 vector signed long long vec_mulo (vector signed int,
28682 vector unsigned long long vec_mulo (vector unsigned int,
28683 vector unsigned int)
28684 * doc/extend.texi: Fix the documentation for the built-in
28687 2017-06-07 Carl Love <cel@us.ibm.com>
28690 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28693 2017-06-07 Carl Love <cel@us.ibm.com>
28695 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28696 support, Generate doublehv for signed int/float for BE case only.
28698 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28700 * doc/invoke.texi (mcx16): Rewrite.
28702 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28704 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28705 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28706 *mov<mode>_softfloat, and an anonymous splitter): Use
28707 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28709 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28711 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28713 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28714 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28715 (rs6000_debug_reg_global): Adjust.
28716 (rs6000_init_hard_regno_mode_ok): Adjust.
28717 (rs6000_dbx_register_number): Adjust.
28718 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28719 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28720 Remove SPE_ACC and SPEFSCR.
28721 (REG_ALLOC_ORDER): Ditto.
28722 (FRAME_POINTER_REGNUM): Change to 111.
28723 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28724 (REG_CLASS_NAMES): Ditto.
28725 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28726 (REGISTER_NAMES): Ditto.
28727 (ADDITIONAL_REG_NAMES): Ditto.
28728 (rs6000_reg_names): Ditto.
28729 * config/rs6000/rs6000.md: Renumber some register number
28732 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28734 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28736 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28737 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28739 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28740 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28741 Delete the SPE high registers.
28742 (REG_ALLOC_ORDER): Ditto.
28743 (enum reg_class): Remove SPE_HIGH_REGS.
28744 (REG_CLASS_NAMES): Ditto.
28745 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28746 (REGISTER_NAMES): Ditto.
28747 (rs6000_reg_names): Ditto.
28748 * doc/tm.texi.in: Remove SPE as example.
28749 * doc/tm.texi: Regenerate.
28751 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28753 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28754 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28755 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28756 * config/rs6000/rs6000.md (type): Remove "brinc".
28758 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28760 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28761 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28762 * config/rs6000/linuxspe.h: Delete file.
28763 * config/rs6000/rs6000.md: Don't include spe.md.
28764 * config/rs6000/spe.h: Delete file.
28765 * config/rs6000/spe.md: Delete file.
28766 * config/rs6000/t-rs6000: Remove spe.md.
28768 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28770 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28771 (reg_or_none500mem_operand): Delete.
28772 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28773 instead of reg_or_none500mem_operand.
28775 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28777 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28778 handling of SPE flags.
28779 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28781 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28783 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28785 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28786 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28787 paired_divv2sf3): Similar.
28788 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28789 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
28790 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
28792 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
28793 Rename the paired_* instruction patterns.
28794 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
28796 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
28797 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
28798 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
28799 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
28800 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
28801 PAIRED_VECTOR_MODE.
28802 (struct machine_function): Delete field spe_insn_chain_scanned_p.
28803 (spe_func_has_64bit_regs_p): Delete.
28804 (spe_expand_predicate_builtin): Delete.
28805 (spe_expand_evsel_builtin): Delete.
28806 (TARGET_DWARF_REGISTER_SPAN): Do not define.
28807 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
28808 (invalid_e500_subreg): Delete.
28809 (rs6000_legitimize_address): Always force_reg op2 as well, for
28810 paired single memory accesses.
28811 (rs6000_member_type_forces_blk): Delete.
28812 (rs6000_spe_function_arg): Delete.
28813 (rs6000_expand_unop_builtin): Delete SPE handling.
28814 (rs6000_expand_binop_builtin): Ditto.
28815 (spe_expand_stv_builtin): Delete.
28816 (bdesc_2arg_spe): Delete.
28817 (spe_expand_builtin): Delete.
28818 (spe_expand_predicate_builtin): Delete.
28819 (spe_expand_evsel_builtin): Delete.
28820 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28821 (spe_init_builtins): Delete.
28822 (spe_func_has_64bit_regs_p): Delete.
28823 (savres_routine_name): Delete "info" parameter. Adjust callers.
28824 (rs6000_emit_stack_reset): Ditto.
28825 (rs6000_dwarf_register_span): Delete.
28826 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28827 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28828 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28830 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28832 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28833 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28834 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28835 mulv2sf3, divv2sf3): Delete expanders.
28837 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28839 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28841 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28843 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
28844 * config/rs6000/rs6000.c: Ditto.
28846 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28848 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
28849 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
28850 comparison_operator.
28852 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28854 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28855 * config/rs6000/rs6000.opt: Ditto.
28856 * config/rs6000/t-rtems: Ditto.
28858 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28860 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
28861 TARGET_E500_SINGLE by 0, simplify.
28862 * config/rs6000/rs6000.c: Ditto.
28863 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
28864 (spe_build_register_parallel): Delete.
28865 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
28866 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
28867 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
28868 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
28869 (E500_CONVERT): Delete.
28870 * config/rs6000/spe.md: Remove many patterns and all define_constants.
28872 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28874 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28875 * config/rs6000/dfp.md: Ditto.
28876 (negdd2, *negdd2_fpr): Merge.
28877 (absdd2, *absdd2_fpr): Merge.
28878 (negtd2, *negtd2_fpr): Merge.
28879 (abstd2, *abstd2_fpr): Merge.
28880 * config/rs6000/e500.h: Delete file.
28881 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28882 TARGET_FPRS by 1 and simplify.
28883 * config/rs6000/rs6000-c.c: Ditto.
28884 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28885 TARGET_DF_SPE by 0.
28886 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28888 * config/rs6000/rs6000.md: Ditto.
28889 (floatdidf2, *floatdidf2_fpr): Merge.
28890 (move_from_CR_gt_bit): Delete.
28891 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28892 (E500_CR_IOR_COMPARE): Delete.
28893 (All patterns that require !TARGET_FPRS): Delete.
28894 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28896 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28898 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28900 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28902 * graphds.c (add_edge): Intitialize edge's attached data.
28903 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28904 pointer parameter. Call pointed function on each edge during
28905 graph traversing. Skip traversing the edge when the function
28907 (graphds_dfs, graphds_scc): Ditto.
28908 (for_each_edge): New parameter. Pass the new parameter to callback
28910 * graphds.h (skip_edge_callback): New function pointer type.
28911 (graphds_dfs, graphds_scc): New function pointer parameter.
28912 (graphds_edge_callback, for_each_edge): New parameter.
28914 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28916 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28917 out code checking if runtime alias check is possible to below ...
28918 Call the new function.
28919 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28920 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28922 2017-06-07 Marek Polacek <polacek@redhat.com>
28925 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28926 TYPE_OVERFLOW_WRAPS check.
28928 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28930 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28931 if versioning is required.
28932 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28933 peeling with the check for versioning.
28935 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28937 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28938 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28939 Set true to new parameter if new ssa variable is defined.
28940 (vect_gen_vector_loop_niters): Refactor. Set range information
28941 for the new vector loop bound variable.
28942 (vect_do_peeling): Ditto.
28944 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28946 * tree-affine.c (ssa.h): Include header file.
28947 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28948 has wrapping overflow behavior.
28950 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28952 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28954 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28956 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28957 (tree_to_aff_combination): ... here.
28959 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28961 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28962 reg_pressure model function.
28963 (ivopts_global_cost_for_size): Delete.
28964 (determine_set_costs, iv_ca_recount_cost): Call new model function
28965 ivopts_estimate_reg_pressure.
28967 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28969 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28970 expensive than udiv. Remove floating point cases from mod.
28972 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28974 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28975 Increase idiv cost.
28977 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28979 * config/aarch64/aarch64.md
28980 (copysignsf3): Fix mask generation.
28982 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28984 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28986 (class dump_manager): Add register_dumps method.
28987 * dumpfile.c: Include langhooks.h.
28988 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28989 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28990 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28991 (dump_manager::dump_register): Start with 512 entries instead of 32.
28992 (dump_manager::register_dumps): New method.
28993 * toplev.c (general_init): Instead of invoking register_dumps
28994 langhook, invoke register_dumps method on the dump manager.
28995 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28998 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29000 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29001 Say that # requires an associated define_split to exist, and that
29002 the define_split must be suitable for use after register allocation.
29004 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29006 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29007 (compute_outgoing_frequencies): Also initialize zero counts.
29008 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29009 around loops; preserve more of profile when nothing changes.
29011 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29013 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29015 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29016 * config/arm/arm-cpu-cdata.h: Regenerate.
29017 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29018 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29019 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29020 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29021 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29023 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29025 * config/arm/t-rmprofile: Likewise.
29026 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29028 2017-06-06 David S. Miller <davem@davemloft.net>
29031 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29032 blockage if function uses alloca.
29034 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29036 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29037 New "uid" fields to hold pretty-print IDs of group and ref.
29038 Memory references are now identified as <group_id>:<ref_id>
29039 instead of using [random] addresses.
29040 (dump_mem_details): Simplify, no functional change.
29041 (dump_mem_ref): Simplify and make output more concise.
29042 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29043 (find_or_create_group): Initialize group uid.
29044 (record_ref): Initialize ref uid. Improve debug output.
29045 (prune_group_by_reuse, should_issue_prefetch_p,)
29046 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29047 (mark_nontemporal_store, determine_loop_nest_reuse):
29048 Improve debug output.
29050 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29052 * dbgcnt.def (prefetch): New debug counter.
29053 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29054 (schedule_prefetches): Stop issueing prefetches if debug counter
29057 2017-06-06 Tom de Vries <tom@codesourcery.com>
29059 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29060 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29062 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29064 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29065 Use aarch64_reg_or_zero predicate for operand 4.
29066 (aarch64_compare_and_swap<mode> define_insn_and_split):
29067 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29068 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29070 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29072 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29073 (arm_compute_save_core_reg_mask): This.
29074 (thumb1_compute_save_reg_mask): Rename into ...
29075 (thumb1_compute_save_core_reg_mask): This.
29076 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29077 (arm_compute_frame_layout): Likewise.
29079 2017-06-06 Richard Biener <rguenther@suse.de>
29081 PR tree-optimization/80974
29082 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29083 keep or clear leaders SSA info.
29085 2017-06-06 Tom de Vries <tom@codesourcery.com>
29087 * config/nvptx/nvptx.c (split_mode_p): New function.
29088 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29090 2017-06-06 Tom de Vries <tom@codesourcery.com>
29092 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29094 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29097 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29100 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29102 * shrink-wrap.c (handle_simple_exit): Update profile.
29103 (try_shrink_wrapping): Upate profile.
29105 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29107 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29108 (tree_guess_outgoing_edge_probabilities): New.
29109 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29110 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29112 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29114 * ipa-split.c (split_function): Initialize return bb profile.
29116 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29118 * profile.c (compute_branch_probabilities): Also initialize
29119 EXIT_BLOCK profile.
29121 2017-06-06 Richard Biener <rguenther@suse.de>
29123 PR tree-optimization/80928
29124 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29125 (vect_analyze_loop_operations): Properly guard analysis for
29127 (vect_transform_loop): Likewise.
29128 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29129 (vect_model_induction_cost): Do not cost for pure SLP.
29130 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29131 of induction in inner loop vectorization.
29132 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29133 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29134 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29136 (vect_analyze_slp_cost_1): Cost induction.
29137 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29138 (vect_get_slp_vect_defs): Likewise.
29139 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29140 (vect_transform_stmt): Handle SLP reductions.
29141 * tree-vectorizer.h (vectorizable_induction): Adjust.
29143 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29145 * config/rs6000/rs6000.c (make_resolver_func): Update
29146 init_lowered_empty_function call.
29148 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29150 * doc/invoke.texi: Document the -fprofile-abs-path option.
29151 * common.opt (fprofile-abs-path): New option.
29152 * gcov-io.h (gcov_write_filename): Declare.
29153 * gcov-io.c (gcov_write_filename): New function.
29154 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29155 * profile.c (output_location): Likewise.
29157 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29159 * shring-wrap.c: Revert accidental commit.
29161 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29163 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29165 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29167 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29169 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29170 profile in callgraph edge.
29171 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29172 (apply_scale): Likewise.
29173 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29174 Also scale profile when inlining function with zero profile.
29175 (initialize_cfun): Update exit block profile even when it is zero.
29176 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29177 when profile is read.
29179 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29181 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29182 (CLONE_*): New constants to define the processors we can generate
29183 code for with the target_clone attribute.
29184 (rs6000_clone_map): New array to identify which clone processors
29185 the current program is running on.
29186 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29187 target_clone attribute.
29188 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29189 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29190 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29191 (cpu_expand_builtin): Add support for target_clone attribute.
29192 (rs6000_valid_attribute_p): Allow "default" attribute.
29193 (get_decl_name): New debug function to simplify printing the
29194 current function name in debugging statements.
29195 (rs6000_clone_priority): New functions to support the target_clone
29196 attribute, and be able to generate code to switch between ISA 2.05
29197 through ISA 3.0 (power6 through power9).
29198 (rs6000_compare_version_priority): Likewise.
29199 (rs6000_get_function_versions_dispatcher): Likewise.
29200 (make_resolver_func): Likewise.
29201 (add_condition_to_bb): Likewise.
29202 (dispatch_function_versions): Likewise.
29203 (rs6000_generate_version_dispatcher_body): Likewise.
29204 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29205 (fusion_gpr_load_p): Fix a spacing issue.
29206 * doc/extend.texi (Common Function Attributes): Document that the
29207 PowerPC supports the target_clone attribute.
29209 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29211 * config/arm/arm.h: explain F symbol found in description of ARM
29212 register allocation in its legend.
29214 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29216 * config/mips/frame-header-opt.c: Include profile-count.h.
29217 * config/riscv/riscv.c: Include profile-count.h
29219 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29221 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29223 (sm_set_flag_if_changed): Add bbs field.
29224 (execute_sm_if_changed_flag_set): Pass BBS.
29225 (execute_sm): Update.
29227 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29229 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29232 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29234 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29235 (peephole2): New peephole2 to emit the above.
29236 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29238 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29240 * config/aarch64/aarch64.c (define_peephole2 above
29241 *sub_<shift>_<mode>): New peephole.
29243 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29245 * config/i386/i386.c (make_resolver_func): Update.
29246 * Makefile.in: Add profile-count.h and profile-count.o
29247 * auto-profile.c (afdo_indirect_call): Update to new API.
29248 (afdo_set_bb_count): Update.
29249 (afdo_propagate_edge): Update.
29250 (afdo_propagate_circuit): Update.
29251 (afdo_calculate_branch_prob): Update.
29252 (afdo_annotate_cfg): Update.
29253 * basic-block.h: Include profile-count.h
29254 (struct edge_def): Turn count to profile_count.
29255 (struct basic_block_def): Likewie.
29256 (REG_BR_PROB_BASE): Move to profile-count.h
29257 (RDIV): Move to profile-count.h
29258 * bb-reorder.c (max_entry_count): Turn to profile_count.
29259 (find_traces): Update.
29260 (rotate_loop):Update.
29261 (connect_traces):Update.
29262 (sanitize_hot_paths):Update.
29263 * bt-load.c (migrate_btr_defs): Update.
29264 * cfg.c (RDIV): Remove.
29265 (init_flow): Use alloc_block.
29266 (alloc_block): Uninitialize count.
29267 (unchecked_make_edge): Uninitialize count.
29268 (check_bb_profile): Update.
29269 (dump_edge_info): Update.
29270 (dump_bb_info): Update.
29271 (update_bb_profile_for_threading): Update.
29272 (scale_bbs_frequencies_int): Update.
29273 (scale_bbs_frequencies_gcov_type): Update.
29274 (scale_bbs_frequencies_profile_count): New.
29275 * cfg.h (update_bb_profile_for_threading): Update.
29276 (scale_bbs_frequencies_profile_count): Declare.
29277 * cfgbuild.c (compute_outgoing_frequencies): Update.
29278 (find_many_sub_basic_blocks): Update.
29279 * cfgcleanup.c (try_forward_edges): Update.
29280 (try_crossjump_to_edge): Update.
29281 * cfgexpand.c (expand_gimple_tailcall): Update.
29282 (construct_exit_block): Update.
29283 * cfghooks.c (verify_flow_info): Update.
29284 (dump_bb_for_graph): Update.
29285 (split_edge): Update.
29286 (make_forwarder_block): Update.
29287 (duplicate_block): Update.
29288 (account_profile_record): Update.
29289 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29290 (get_estimated_loop_iterations): Update.
29291 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29292 (single_likely_exit): Update.
29293 * cfgloopmanip.c (scale_loop_profile): Update.
29295 (set_zero_probability): Update.
29296 (lv_adjust_loop_entry_edge): Update.
29297 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29298 (purge_dead_edges): Update.
29299 (rtl_account_profile_record): Update.
29300 * cgraph.c (cgraph_node::create): Uninitialize count.
29301 (symbol_table::create_edge): Uninitialize count.
29302 (cgraph_update_edges_for_call_stmt_node): Update.
29303 (cgraph_edge::dump_edge_flags): Update.
29304 (cgraph_node::dump): Update.
29305 (cgraph_edge::maybe_hot_p): Update.
29306 * cgraph.h: Include profile-count.h
29307 (create_clone), create_edge, create_indirect_edge): Update.
29308 (cgraph_node): Turn count to profile_count.
29309 (cgraph_edge0: Likewise.
29310 (make_speculative, clone): Update.
29311 (create_edge): Update.
29312 (init_lowered_empty_function): Update.
29313 * cgraphclones.c (cgraph_edge::clone): Update.
29314 (duplicate_thunk_for_node): Update.
29315 (cgraph_node::create_clone): Update.
29316 * cgraphunit.c (cgraph_node::analyze): Update.
29317 (cgraph_node::expand_thunk): Update.
29318 * final.c (dump_basic_block_info): Update.
29319 * gimple-streamer-in.c (input_bb): Update.
29320 * gimple-streamer-out.c (output_bb): Update.
29321 * graphite.c (print_global_statistics): Update.
29322 (print_graphite_scop_statistics): Update.
29323 * hsa-brig.c: Include basic-block.h.
29324 * hsa-dump.c: Include basic-block.h.
29325 * hsa-gen.c (T sum_slice): Update.
29326 (convert_switch_statements):Update.
29327 * hsa-regalloc.c: Include basic-block.h.
29328 * ipa-chkp.c (chkp_produce_thunks): Update.
29329 * ipa-cp.c (struct caller_statistics): Update.
29330 (init_caller_stats): Update.
29331 (gather_caller_stats): Update.
29332 (ipcp_cloning_candidate_p): Update.
29333 (good_cloning_opportunity_p): Update.
29334 (get_info_about_necessary_edges): Update.
29335 (dump_profile_updates): Update.
29336 (update_profiling_info): Update.
29337 (update_specialized_profile): Update.
29338 (perhaps_add_new_callers): Update.
29339 (decide_about_value): Update.
29340 (ipa_cp_c_finalize): Update.
29341 * ipa-devirt.c (struct odr_type_warn_count): Update.
29342 (struct decl_warn_count): Update.
29343 (struct final_warning_record): Update.
29344 (possible_polymorphic_call_targets): Update.
29345 (ipa_devirt): Update.
29346 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29347 * ipa-icf.c (sem_function::merge): Update.
29348 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29349 * ipa-inline.c (compute_uninlined_call_time): Update.
29350 (compute_inlined_call_time): Update.
29351 (want_inline_small_function_p): Update.
29352 (want_inline_self_recursive_call_p): Update.
29353 (edge_badness): Update.
29354 (lookup_recursive_calls): Update.
29355 (recursive_inlining): Update.
29356 (inline_small_functions): Update.
29357 (dump_overall_stats): Update.
29358 (dump_inline_stats): Update.
29359 * ipa-profile.c (ipa_profile_generate_summary): Update.
29360 (ipa_propagate_frequency): Update.
29361 (ipa_profile): Update.
29362 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29363 * ipa-utils.c (ipa_merge_profiles): Update.
29364 * loop-doloop.c (doloop_modify): Update.
29365 * loop-unroll.c (report_unroll): Update.
29366 (unroll_loop_runtime_iterations): Update.
29367 * lto-cgraph.c (lto_output_edge): Update.
29368 (lto_output_node): Update.
29369 (input_node): Update.
29370 (input_edge): Update.
29371 (merge_profile_summaries): Update.
29372 * lto-streamer-in.c (input_cfg): Update.
29373 * lto-streamer-out.c (output_cfg): Update.
29374 * mcf.c (create_fixup_graph): Update.
29375 (adjust_cfg_counts): Update.
29376 (sum_edge_counts): Update.
29377 * modulo-sched.c (sms_schedule): Update.
29378 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29379 * predict.c (maybe_hot_count_p): Update.
29380 (probably_never_executed): Update.
29381 (dump_prediction): Update.
29382 (combine_predictions_for_bb): Update.
29383 (propagate_freq): Update.
29384 (handle_missing_profiles): Update.
29385 (counts_to_freqs): Update.
29386 (rebuild_frequencies): Update.
29387 (force_edge_cold): Update.
29388 * predict.h: Include profile-count.h
29389 (maybe_hot_count_p, counts_to_freqs): UPdate.
29390 * print-rtl-function.c: Do not include cfg.h
29391 * print-rtl.c: Include basic-block.h
29392 * profile-count.c: New file.
29393 * profile-count.h: New file.
29394 * profile.c (is_edge_inconsistent): Update.
29395 (correct_negative_edge_counts): Update.
29396 (is_inconsistent): Update.
29397 (set_bb_counts): Update.
29398 (read_profile_edge_counts): Update.
29399 (compute_frequency_overlap): Update.
29400 (compute_branch_probabilities): Update; Initialize and deinitialize
29402 (branch_prob): Update.
29403 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29404 (edge_gcov_count): New.
29405 (bb_gcov_count): New.
29406 * shrink-wrap.c (try_shrink_wrapping): Update.
29407 * tracer.c (better_p): Update.
29408 * trans-mem.c (expand_transaction): Update.
29409 (ipa_tm_insert_irr_call): Update.
29410 (ipa_tm_insert_gettmclone_call): Update.
29411 * tree-call-cdce.c: Update.
29412 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29413 (gimple_duplicate_sese_tail): Update.
29414 (gimple_account_profile_record): Update.
29415 (execute_fixup_cfg): Update.
29416 * tree-inline.c (copy_bb): Update.
29417 (copy_edges_for_bb): Update.
29418 (initialize_cfun): Update.
29419 (freqs_to_counts): Update.
29420 (copy_cfg_body): Update.
29421 (expand_call_inline): Update.
29422 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29423 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29424 (try_unroll_loop_completely): Update.
29425 (try_peel_loop): Update.
29426 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29427 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29428 * tree-ssa-loop-split.c (connect_loops): Update.
29429 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29430 * tree-ssa-reassoc.c (branch_fixup): Update.
29431 * tree-ssa-tail-merge.c (replace_block_by): Update.
29432 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29433 (compute_path_counts): Update.
29434 (update_profile): Update.
29435 (recompute_probabilities): Update.
29436 (update_joiner_offpath_counts): Update.
29437 (estimated_freqs_path): Update.
29438 (freqs_to_counts_path): Update.
29439 (clear_counts_path): Update.
29440 (ssa_fix_duplicate_block_edges): Update.
29441 (duplicate_thread_path): Update.
29442 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29443 (struct switch_conv_info): Update.
29444 * tree-tailcall.c (decrease_profile): Update.
29445 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29446 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29447 * value-prof.c (check_counter): Update.
29448 (gimple_divmod_fixed_value): Update.
29449 (gimple_mod_pow2): Update.
29450 (gimple_mod_subtract): Update.
29451 (gimple_ic_transform): Update.
29452 (gimple_stringop_fixed_value): Update.
29453 * value-prof.h (gimple_ic): Update.
29455 2017-06-02 Carl Love <cel@us.ibm.com>
29457 * config/rs6000/rs6000-c: Add support for built-in functions
29458 vector double vec_doublee (vector signed int);
29459 vector double vec_doublee (vector unsigned int);
29460 vector double vec_doublee (vector float);
29461 vector double vec_doubleh (vector signed int);
29462 vector double vec_doubleh (vector unsigned int);
29463 vector double vec_doubleh (vector float);
29464 vector double vec_doublel (vector signed int);
29465 vector double vec_doublel (vector unsigned int);
29466 vector double vec_doublel (vector float);
29467 vector double vec_doubleo (vector signed int);
29468 vector double vec_doubleo (vector unsigned int);
29469 vector double vec_doubleo (vector float);.
29470 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29471 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29473 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29474 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29475 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29477 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29478 vec_doublel, vec_doubleh.
29479 * doc/extend.texi: Update the built-in documentation file for the
29480 new built-in functions.
29482 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29485 * ipa-inline-analysis.c (free_growth_caches): Set
29486 edge_removal_hook_holder to NULL after removing it.
29488 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29490 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29491 comparision with zero.
29493 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29494 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29495 for early expansion of vec_min and vec_max builtins.
29496 (builtin_function_type): Add min/max unsigned variants to those
29497 identified as having unsigned arguments.
29499 2017-06-02 Olivier Hainque <hainque@adacore.com>
29501 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29503 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29505 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29506 Use VALL_F16 iterator rather than VALL.
29508 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29510 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29511 Emit CBNZ inside loop when doing a strong exchange and comparing
29512 against zero. Generate the CC flags after the loop.
29514 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29516 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29517 (dl_section_ref): New.
29518 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29519 On AIX, append an expression to subtract the size of the
29520 section length to dl_section_ref.
29522 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29524 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29525 for early expansion of vector absolute builtins.
29527 2017-06-02 Richard Biener <rguenther@suse.de>
29529 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29530 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29532 2017-06-02 Richard Biener <rguenther@suse.de>
29534 PR tree-optimization/80948
29535 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29536 stmt order as well.
29538 2017-06-02 Richard Biener <rguenther@suse.de>
29540 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29542 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29543 for inductions as relevant.
29545 2017-06-02 Richard Biener <rguenther@suse.de>
29547 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29548 (vectorizable_induction): ... this. Remove dead code.
29550 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29552 * builtins. (expand_builtin_alloca): Remove second parameter and
29553 infer its value from the first parameter instead.
29554 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29556 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29558 PR rtl-optimization/80903
29559 * loop-doloop.c (add_test): Unshare sequence.
29561 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29563 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29565 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29567 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29569 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29570 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29571 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29572 (xlogue_layout::get_stub_rtx): Make static.
29573 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29574 (xlogue_layout::compute_stub_managed_regs): Rename to...
29575 (xlogue_layout::count_stub_managed_regs): ...this.
29576 (xlogue_layout::is_stub_managed_reg): New function.
29577 (xlogue_layout::m_stub_names): Rename to...
29578 (xlogue_layout::s_stub_names): ...this, make static.
29579 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29580 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29581 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29582 xlogue_layout::s_stub_names): Instantiate statics.
29583 (stub_managed_regs): Remove.
29584 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29585 (disable_call_ms2sysv_xlogues): Rename to...
29586 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29587 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29589 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29590 change after reload_completed.
29591 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29593 (ix86_expand_prologue): Likewise.
29594 (ix86_expand_epilogue): Likewise.
29595 (ix86_expand_split_stack_prologue): Likewise.
29596 (ix86_compute_frame_layout): Remove frame parameter ...
29597 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29598 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29600 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29601 (ix86_frame): Move from here ...
29602 * config/i386/i386.h (ix86_frame): ... to here.
29603 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29604 complete ix86_frame data structure instead. Remove some_ld_name.
29606 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29608 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29609 symbols that hold a DECL_VALUE_EXPR.
29611 2017-06-01 Martin Jambor <mjambor@suse.cz>
29613 PR tree-optimization/80898
29614 * tree-sra.c (process_subtree_disqualification): Removed.
29615 (disqualify_candidate): Do not acll
29616 process_subtree_disqualification.
29617 (subtree_mark_written_and_enqueue): New function.
29618 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29619 RHS has been disqualified and re-queue LHS if necessary. Apart
29620 from that, ignore disqualified RHS.
29622 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29624 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29625 address fetch for z10 or later.
29627 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29629 * config/arc/arc.md (tst_movb): Add guard when splitting.
29631 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29633 * config/arc/arc.c (arc_can_eliminate): Test against
29634 arc_frame_pointer_needed.
29636 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29638 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29639 to prevent store reordering.
29640 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29641 (type): Add block type.
29642 (stack_tie): Define special instruction to be used in
29645 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29647 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29648 constraint. It is not valid for the pattern.
29649 (noncommutative_binary_comparison): Likewise.
29651 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29653 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29656 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29658 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29659 be used by the reg-alloc.
29661 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29663 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29664 reg-alloc when having mul64 or mul32x16 instructions.
29665 (mulsidi3): Likewise.
29666 (umulsidi3): Likewise.
29667 (mulsi32x16): New pattern.
29668 (mulsi64): Likewise.
29669 (mulsidi64): Likewise.
29670 (umulsidi64): Likewise.
29671 (MUL32x16_REG): Define.
29672 (mul64_600): Use MUL32x16_REG.
29673 (mac64_600): Likewise.
29674 (umul64_600): Likewise.
29675 (umac64_600): Likewise.
29677 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29679 * config/arc/arc.md (mulsi3_700): Make it commutative.
29681 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29683 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29685 (*sign_extendsidi2_insn): Likewise for movstosw.
29687 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29689 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29690 the type of the input discriminant value. Convert the
29691 discriminant value of signedness vary.
29693 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29695 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29696 Add to -Wall section.
29698 2017-06-01 Richard Biener <rguenther@suse.de>
29700 PR middle-end/66313
29701 * fold-const.c (fold_plusminus_mult_expr): If the factored
29702 factor may be zero use a wrapping type for the inner operation.
29703 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29704 and handle moved defs.
29705 (process_assignment): Properly guard the unary op case. Return a
29706 tri-state indicating that moving the stmt before the call may allow
29707 to continue. Pass through to_move.
29708 (find_tail_calls): Handle moving unrelated defs before
29711 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29714 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29715 splitter result in the canonical way.
29717 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29719 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29720 also for 32bit target. Update insn attributes.
29721 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29723 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29725 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29726 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29727 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29728 (_mm_maskz_min_ss): New intrinsics.
29730 2017-05-31 Martin Liska <mliska@suse.cz>
29732 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29733 Change comment style to one we normally use.
29734 (vectorizable_reduction): Likewise.
29735 (vectorizable_induction): Likewise.
29736 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29737 (vectorizable_call): Likewise.
29738 (vectorizable_simd_clone_call): Likewise.
29739 (vectorizable_conversion): Likewise.
29740 (vectorizable_assignment): Likewise.
29741 (vectorizable_shift): Likewise.
29742 (vectorizable_operation): Likewise.
29743 (vectorizable_store): Likewise.
29744 (vectorizable_load): Likewise.
29745 * tree-vectorizer.h: Likewise.
29747 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29749 * passes.c (emergency_dump_function): New.
29750 * tree-pass.h (emergency_dump_function): Declare.
29751 * plugin.c (plugins_internal_error_function): Remove.
29752 * plugin.h (plugins_internal_error_function): Remove declaration.
29753 * toplev.c (internal_error_function): New static function. Use it...
29754 (general_init): ...here.
29756 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29758 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29759 (arc_rtx_costs): Add costs for new patterns.
29760 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29761 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29763 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29765 * tree-ssa-strlen.c (get_next_strinfo): New function.
29766 (get_stridx_plus_constant): Use it.
29767 (zero_length_string): Likewise.
29768 (adjust_related_strinfos): Likewise.
29769 (adjust_last_stmt): Likewise.
29771 2017-05-31 Richard Biener <rguenther@suse.de>
29774 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29775 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29777 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29779 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29780 loop_vinfo argument and use of dependence distance vectors.
29781 Check instead whether the two references differ only in their
29782 initial value and assume that they have the same alignment if the
29783 difference is a multiple of the vector alignment.
29784 (vect_analyze_data_refs_alignment): Update call accordingly.
29786 2017-05-31 Martin Liska <mliska@suse.cz>
29789 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
29791 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29793 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
29794 (create_intersect_range_checks): Move from ...
29795 * tree-data-ref.c (create_intersect_range_checks_index)
29796 (create_intersect_range_checks): ... to here.
29797 (create_runtime_alias_checks): New function factored from ...
29798 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
29799 here. Call above function.
29800 * tree-data-ref.h (create_runtime_alias_checks): New function.
29802 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29804 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
29805 segment length for dr_b and compute it in wide_int.
29807 2017-05-31 Richard Biener <rguenther@suse.de>
29809 PR tree-optimization/80906
29810 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
29811 and pass through iv_map.
29812 (copy_bb_and_scalar_dependences): Adjust.
29813 (translate_pending_phi_nodes): Likewise.
29814 (copy_loop_close_phi_args): Handle code-generating IVs instead
29817 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29819 * diagnostic-color.c (color_dict): Add "type-diff".
29820 (parse_gcc_colors): Update comment.
29821 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29822 -fdiagnostics-show-template-tree and -fno-elide-type.
29823 (GCC_COLORS): Add type-diff to example.
29825 (-fdiagnostics-show-template-tree): New.
29826 (-fno-elide-type): New.
29827 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29828 the pp_format_decoder callback. Call any m_format_postprocessor's
29830 (pretty_printer::pretty_printer): Initialize
29831 m_format_postprocessor.
29832 (pretty_printer::~pretty_printer): Delete any
29833 m_format_postprocessor.
29834 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29835 (class format_postprocessor): New class.
29836 (struct pretty_printer::format_decoder): Document the new parameters.
29837 (struct pretty_printer::m_format_postprocessor): New field.
29838 * tree-diagnostic.c (default_tree_printer): Update for new
29839 bool and const char ** params.
29840 * tree-diagnostic.h (default_tree_printer): Likewise.
29842 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
29844 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
29845 (lwa_operand): Delete rs6000_gen_cell_microcode test.
29846 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29847 rs6000_gen_cell_microcode code.
29848 (rs6000_final_prescan_insn): Delete.
29849 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
29850 "warn-cell-microcode" entries.
29851 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29852 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29853 throughout. Change cc_reg_not_micro_cr0_operand to
29854 cc_reg_not_cr0_operand throughout.
29855 (*extendhi<mode>2_noload): Delete.
29856 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
29857 (mwarn-cell-microcode): Delete.
29858 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
29859 -mgen-cell-microcode and -mwarn-cell-microcode.
29861 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
29864 * config/i386/constraints.md (Yd): New constraint.
29866 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
29867 and (?Yd, r) alternatives. Update insn attributes.
29868 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
29869 and (?*Yd, r) alternatives. Update insn attributes.
29870 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
29872 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
29874 * gimplify.c (gimplify_modify_expr): Don't create a
29875 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29878 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29880 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29882 2017-05-30 Richard Biener <rguenther@suse.de>
29884 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29885 and reduc_def fields.
29886 (STMT_VINFO_REDUC_TYPE): New define.
29887 (STMT_VINFO_REDUC_DEF): Likewise.
29888 (vect_force_simple_reduction): Adjust prototype.
29889 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29890 (vect_is_simple_reduction): Remove check_reduction argument.
29891 (vect_force_simple_reduction): Adjust and set
29892 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29893 (vectorizable_reduction): Do not re-do reduction analysis
29894 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29895 * tree-parloops.c (gather_scalar_reductions): Adjust.
29897 2017-05-30 Richard Biener <rguenther@suse.de>
29899 PR middle-end/80901
29900 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29903 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29905 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29906 Introduce unknown_misalignment parameter and remove vf.
29907 (vect_peeling_hash_get_lowest_cost):
29908 Pass unknown_misalignment parameter.
29909 (vect_enhance_data_refs_alignment):
29910 Fix unsupportable data ref treatment.
29912 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29914 * tree-vect-data-refs.c (vect_get_data_access_cost):
29915 Workaround for SLP handling.
29916 (vect_enhance_data_refs_alignment):
29917 Compute costs for doing no peeling at all, compare to the best
29918 peeling costs so far and avoid peeling if cheaper.
29920 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29922 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29923 Return peeling info and set costs to zero for unlimited cost
29925 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29926 with unknown misalignment. Compute and costs for unknown
29927 misalignment, compare them to the costs for known misalignment
29928 and choose the cheapest for peeling.
29930 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29932 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29933 (vect_get_peeling_costs_all_drs): Create function.
29934 (vect_peeling_hash_get_lowest_cost):
29935 Use vect_get_peeling_costs_all_drs.
29936 (vect_peeling_supportable): Create function.
29937 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29939 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29941 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29942 DR_HAS_NEGATIVE_STEP.
29943 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29944 (vect_enhance_data_refs_alignment): Use.
29945 (vect_duplicate_ssa_name_ptr_info): Use.
29946 * tree-vectorizer.h (dr_misalignment): Use.
29947 (known_alignment_for_access_p): Use.
29949 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29952 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29954 (has_section_name): New function.
29956 2017-05-30 Martin Liska <mliska@suse.cz>
29959 * auto-profile.c (get_function_decl_from_block): Fix
29962 2017-05-30 Richard Biener <rguenther@suse.de>
29964 PR middle-end/80876
29965 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29967 2017-05-30 Martin Liska <mliska@suse.cz>
29969 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29970 * dumpfile.h (struct dump_file_info): Remove ctors.
29972 2017-05-30 Martin Liska <mliska@suse.cz>
29974 * predict.def: Fix GNU coding style.
29976 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29978 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29979 Mark 'to' argument with ATTRIBUTE_UNUSED.
29981 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29983 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29984 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29985 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29988 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29990 * doc/install.texi (Options specification): Restore entry of
29991 --enable-sjlj-exceptions.
29993 2017-05-27 Michael Eager <eager@eagercon.com>
29996 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29998 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30000 * config/microblaze/microblaze.h
30001 (FIXED_REGISTERS): Update in macro.
30002 (CALL_USED_REGISTERS): Update in macro.
30004 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30006 * doc/install.texi: Add links to macOS binary distributions.
30008 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30012 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30014 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30016 2017-05-26 Martin Liska <mliska@suse.cz>
30018 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30020 2017-05-26 Martin Liska <mliska@suse.cz>
30022 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30023 always leading ';; '.
30024 (dump_bb_info): Likewise.
30025 (brief_dump_cfg): Likewise.
30026 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30027 * dumpfile.c: Remove usage of TDF_VERBOSE.
30028 * dumpfile.h (enum dump_kind): Likewise.
30029 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30030 * print-tree.c (debug_verbose): Remove.
30031 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30032 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30033 * tree-diagnostic.c (default_tree_printer): Replace
30034 TDF_DIAGNOSTIC with TDF_SLIM.
30036 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30038 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30039 in parameter loop, rather than loop_vinfo.
30040 (create_intersect_range_checks): Ditto.
30041 (vect_create_cond_for_alias_checks): Update call to above functions.
30043 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30045 PR tree-optimization/80815
30046 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30047 for merging runtime alias checks. Handle negative DR_STEPs.
30049 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30051 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30053 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30054 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30055 out code pruning runtime alias checks.
30056 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30057 factored out from above.
30058 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30060 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30062 (prune_runtime_alias_test_list): New decalaration.
30064 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30066 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30067 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30068 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30069 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30070 (operator==, comp_dr_with_seg_len_pair): Ditto.
30071 (vect_prune_runtime_alias_test_list): Ditto.
30073 2017-05-26 Martin Liska <mliska@suse.cz>
30076 * params.def: Bound partial-inlining-entry-probability param.
30078 2017-05-26 Marek Polacek <polacek@redhat.com>
30081 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30084 2017-05-26 Richard Biener <rguenther@suse.de>
30086 PR tree-optimization/80842
30087 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30090 2017-05-26 Richard Biener <rguenther@suse.de>
30092 PR tree-optimization/80844
30093 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30095 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30097 * doc/md.texi (Machine Constraints): Update x86 family
30098 machine constraints section to match 'config/i386/constraints.md'.
30100 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30102 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30104 2017-05-25 Nathan Sidwell <nathan@acm.org>
30106 * doc/invoke.texi (--enable-languages): Update documentation.
30108 2017-05-25 Martin Liska <mliska@suse.cz>
30110 * dumpfile.c: Add TDF_FOLDING.
30111 * dumpfile.h (enum dump_kind): Likewise.
30112 * genmatch.c (dt_simplify::gen_1): Use it.
30114 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30116 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30118 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30120 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30121 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30123 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30125 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30126 * match.pd (X == C): Rewrite it here.
30127 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30128 with_certain_nonzero_bits2): New predicates.
30129 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30131 2017-05-24 Nathan Sidwell <nathan@acm.org>
30133 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30136 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30139 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30141 * config/powerpcspe: New port. Files are copied from the rs6000
30142 port, with "rs6000" in filenames replaced by "powerpcspe".
30144 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30146 PR rtl-optimization/80754
30147 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30149 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30151 * config/sparc/sparc.md (length): Return the correct value for -mflat
30152 sibcalls to match output_sibcall.
30154 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30158 * config/rs6000/rs6000.c (struct machine_function): Add new field
30160 (rs6000_get_separate_components): Init that field, use it.
30161 (rs6000_components_for_bb): Use the field.
30163 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30165 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30167 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30169 PR middle-end/80823
30170 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30172 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30175 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30176 address against address_operand predicate.
30177 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30179 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30181 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30182 parameters passed indirectly.
30184 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30186 * config/i386/i386.md (*movdi_internal): Remove SSE4
30187 alternative 18 (?r, *v). Update insn attributes.
30188 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30189 Update insn attributes.
30190 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30191 Update insn attributes.
30192 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30193 alternative 1 (r, v). Remove isa attribute.
30194 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30195 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30196 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30198 2017-05-23 Tom de Vries <tom@codesourcery.com>
30200 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30203 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30205 * cgraphunit.c (symbol_table::process_new_functions): Update.
30206 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30207 (inline_generate_summary): Rename to ...
30208 (ipa_fn_summary_generate): ... this one.
30209 (inline_read_summary): Rename to ...
30210 (ipa_fn_summary_read): ... this one.
30211 (inline_write_summary): Rename to ...
30212 (ipa_fn_summary_write): ... this one.
30213 (inline_free_summary): Rename to ...
30214 (ipa_free_fn_summary): ... this one.
30215 (pass_data_local_fn_summary, pass_local_fn_summary,
30216 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30217 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30218 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30219 make_pass_ipa_fn_summary): New.
30220 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30221 inline_write_summary, inline_free_summary): Remove.
30222 (ipa_free_fn_summary) : New.
30223 * ipa-inline.c (ipa_inline): Update.
30224 (pass_ipa_inline): Do not generate summaries.
30225 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30227 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30228 and add pass_ipa_fn_summary.
30229 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30231 (make_pass_inline_parameters): Remove.
30233 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30235 * omp-low.c (struct omp_context): Remove "default_kind" member.
30238 * omp-offload.c (execute_oacc_device_lower): Remove the
30239 parallelism dimensions function attributes for unparallelized
30240 OpenACC kernels constructs.
30242 2017-05-23 Martin Liska <mliska@suse.cz>
30244 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30246 (cgraph_edge::make_speculative): Likewise.
30247 (cgraph_edge::resolve_speculation): Likewise.
30248 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30249 (cgraph_node::dump): Likewise.
30250 * cgraph.h: Likewise.
30251 * cgraphunit.c (analyze_functions): Likewise.
30252 (symbol_table::compile): Likewise.
30253 * ipa-cp.c (print_all_lattices): Likewise.
30254 (determine_versionability): Likewise.
30255 (initialize_node_lattices): Likewise.
30256 (ipcp_verify_propagated_values): Likewise.
30257 (estimate_local_effects): Likewise.
30258 (update_profiling_info): Likewise.
30259 (create_specialized_node): Likewise.
30260 (perhaps_add_new_callers): Likewise.
30261 (decide_about_value): Likewise.
30262 (decide_whether_version_node): Likewise.
30263 (identify_dead_nodes): Likewise.
30264 (ipcp_store_bits_results): Likewise.
30265 * ipa-devirt.c (dump_targets): Likewise.
30266 (ipa_devirt): Likewise.
30267 * ipa-icf.c (sem_item::dump): Likewise.
30268 (sem_function::equals): Likewise.
30269 (sem_variable::equals): Likewise.
30270 (sem_item_optimizer::read_section): Likewise.
30271 (sem_item_optimizer::execute): Likewise.
30272 (congruence_class::dump): Likewise.
30273 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30274 (dump_inline_summary): Likewise.
30275 (estimate_node_size_and_time): Likewise.
30276 (inline_analyze_function): Likewise.
30277 * ipa-inline-transform.c (inline_call): Likewise.
30278 * ipa-inline.c (report_inline_failed_reason): Likewise.
30279 (want_early_inline_function_p): Likewise.
30280 (edge_badness): Likewise.
30281 (update_edge_key): Likewise.
30282 (inline_small_functions): Likewise.
30283 * ipa-profile.c (ipa_profile): Likewise.
30284 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30285 (ipa_make_edge_direct_to_target): Likewise.
30286 (remove_described_reference): Likewise.
30287 (ipa_impossible_devirt_target): Likewise.
30288 (propagate_controlled_uses): Likewise.
30289 (ipa_print_node_params): Likewise.
30290 (ipcp_transform_function): Likewise.
30291 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30292 (propagate_pure_const): Likewise.
30293 * ipa-reference.c (generate_summary): Likewise.
30294 (read_write_all_from_decl): Likewise.
30295 (propagate): Likewise.
30296 (ipa_reference_read_optimization_summary): Likewise.
30297 * ipa-utils.c (ipa_merge_profiles): Likewise.
30298 * ipa.c (walk_polymorphic_call_targets): Likewise.
30299 (symbol_table::remove_unreachable_nodes): Likewise.
30300 (ipa_single_use): Likewise.
30301 * passes.c (execute_todo): Likewise.
30302 * predict.c (drop_profile): Likewise.
30303 * symtab.c (symtab_node::get_dump_name): New function.
30304 (symtab_node::dump_name): Likewise.
30305 (symtab_node::dump_asm_name): Likewise.
30306 (symtab_node::dump_references): Likewise.
30307 (symtab_node::dump_referring): Likewise.
30308 (symtab_node::dump_base): Likewise.
30309 (symtab_node::debug_symtab): Likewise.
30310 * tree-sra.c (convert_callers_for_node): Likewise.
30311 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30312 * value-prof.c (init_node_map): Likewise.
30314 2017-05-23 Martin Liska <mliska@suse.cz>
30316 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30317 and symtab_node::debug_symtab to symbol_table::debug.
30318 * cgraphunit.c (analyze_functions): Use the renamed function.
30319 (symbol_table::compile): Likewise.
30320 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30321 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30322 * passes.c (execute_todo): Likewise.
30323 * symtab.c (symbol_table::dump): New function.
30324 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30326 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30328 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30329 that nonconst implies exec.
30331 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30333 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30334 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30335 (inline_edge_summary_vec): Turn into ...
30336 (ipa_call_summaries): ... this one.
30337 (redirect_to_unreachable, edge_set_predicate,
30338 evaluate_properties_for_edge, inline_summary_alloc,
30339 reset_ipa_call_summary, reset_inline_summary,
30340 inline_summary_t::duplicate): Update.
30341 (inline_edge_duplication_hook): Turn to ...
30342 (ipa_call_summary_t::duplicate): ... this one.
30343 (inline_edge_removal_hook): Turn to ...
30344 (ipa_call_summary_t::remove): ... this one.
30345 (dump_inline_edge_summary): Turn to ...
30346 (dump_ipa_call_summary): ... this one.
30347 (estimate_function_body_sizes): Update.
30348 (inline_update_callee_summaries): Update.
30349 (remap_edge_change_prob): Update.
30350 (remap_edge_summaries): Update.
30351 (inline_merge_summary): Update.
30352 (do_estimate_edge_time): Update.
30353 (inline_generate_summary): Update.
30354 (inline_read_section): Update.
30355 (inline_read_summary): Update.
30356 (inline_free_summary): Update.
30357 * ipa-inline.c (can_inline_edge_p): Update.
30358 (compute_inlined_call_time): Update.
30359 (want_inline_small_function_p): Update.
30360 (edge_badness): Update.
30361 (early_inliner): Update.
30362 * ipa-inline.h (inline_edge_summary): Turn to ...
30363 (ipa_call_summary): ... this one.
30364 (ipa_call_summary_t): New class.
30365 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30366 (ipa_call_summaries): New.
30367 (inline_edge_summary): Remove.
30368 (estimate_edge_growth): Update.
30369 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30370 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30371 * ipa-split.c (execute_split_functions): Update.
30372 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30374 2017-05-23 Tom de Vries <tom@codesourcery.com>
30376 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30377 attributes): Document rdrand effective target.
30379 2017-05-23 Tom de Vries <tom@codesourcery.com>
30381 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30382 attributes): Sort alphabetically.
30384 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30386 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30388 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30391 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30392 V2DF/V2DI splat into two separate patterns, one that handles
30393 registers, and the other that only handles memory. Drop support
30394 for splatting from a GPR on ISA 2.07 and then splitting the
30395 splat into direct move and splat.
30396 (vsx_splat_<mode>_reg): Likewise.
30397 (vsx_splat_<mode>_mem): Likewise.
30399 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30401 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30403 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30405 PR middle-end/80809
30406 * omp-low.c (finish_taskreg_remap): New function.
30407 (finish_taskreg_scan): If unit size of ctx->record_type
30408 is non-constant, unshare the size expression and replace
30409 decls in it with possible outer var refs.
30411 PR middle-end/80809
30412 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30413 GOVD_SHARED rather than GOVD_PRIVATE with it.
30414 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30415 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30417 PR middle-end/80853
30418 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30419 as last argument to build_outer_var_ref for pointer bases of array
30420 section reductions.
30422 2017-05-19 Martin Sebor <msebor@redhat.com>
30424 * print-tree.c (print_node): Print DECL_READ_P flag.
30426 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30428 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30429 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30430 * cgraph.c: Likewise.
30431 * cgraphunit.c: Likewise.
30432 * gengtype.c: Likewise.
30433 * ipa-cp.c: Likewise.
30434 * ipa-devirt.c: Likewise.
30435 * ipa-icf.c: Likewise.
30436 * ipa-predicate.c: Likewise.
30437 * ipa-profile.c: Likewise.
30438 * ipa-prop.c: Likewise.
30439 * ipa-split.c: Likewise.
30441 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30442 edge_predicate_pool, dump_inline_hints,
30443 inline_summary::account_size_time, redirect_to_unreachable,
30444 edge_set_predicate, set_hint_predicate,
30445 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30446 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30447 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30448 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30449 ipa_call_summary_t::remove, initialize_growth_caches,
30450 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30451 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30452 mark_modified, unmodified_parm_1, unmodified_parm,
30453 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30454 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30455 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30456 will_be_nonconstant_predicate, record_modified_bb_info,
30457 get_minimal_bb, record_modified, param_change_prob,
30458 phi_result_unknown_predicate, predicate_for_phi_result,
30459 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30460 estimate_function_body_sizes, compute_inline_parameters,
30461 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30462 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30463 inline_update_callee_summaries, remap_edge_change_prob,
30464 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30465 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30466 inline_analyze_function, inline_summary_t::insert,
30467 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30468 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30469 inline_free_summary): Move to ipa-fnsummary.h
30470 (predicate_t): Remove.
30471 * ipa-fnsummary.c: New file.
30472 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30473 (enum inline_hints_vals, inline_hints, agg_position_info,
30474 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30475 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30476 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30477 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30478 inline_read_summary, inline_write_summary, inline_free_summary,
30479 inline_analyze_function, initialize_inline_failed,
30480 inline_merge_summary, inline_update_overall_summary,
30481 compute_inline_parameters): Move to ipa-fnsummary.h
30482 * ipa-fnsummary.h: New file.
30483 * ipa-inline-transform.h: Include ipa-inline.h.
30484 * ipa-inline.c: LIkewise.
30486 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30488 * ipa-inline.c (edge_badness): Use inlined_time instead of
30489 inline_summaries->get.
30491 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30493 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30495 2017-05-22 Nathan Sidwell <nathan@acm.org>
30497 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30498 (fdump-lang): Document 'raw' option.
30499 * dumpfile.h (TDI_tu): Delete.
30500 * dumpfile.c (dump_files): Remove translation-unit.
30501 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30503 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30505 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30506 command option from $(AWK) call.
30507 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30508 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30509 [FORMAT]: Remove handling of variable.
30510 * config/avr/t-multilib: Regenerate.
30512 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30514 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30516 (dump_inline_summary): Do not print self_time.
30517 (estimate_function_body_sizes): Do not set self_time.
30518 (compute_inline_parameters): Likewise.
30519 (inline_read_section, inline_write_summary): Do not stream self_time.
30520 * ipa-inline.h (inline_summary): Drop self_time.
30522 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30524 * ipa-inline-analysis.c (account_size_time): Rename to ...
30525 (inline_summary::account_size_time): ... this one.
30526 (reset_ipa_call_summary): Turn to ...
30527 (ipa_call_summary::reset): ... this one.
30528 (reset_inline_summary): Turn to ...
30529 (inline_summary::reset): ... this one.
30530 (inline_summary_t::remove): Update.
30531 (inline_summary_t::duplicate): Update.
30532 (ipa_call_summary_t::remove): Update.
30533 (dump_inline_summary): Update.
30534 (estimate_function_body_sizes): Update.
30535 (compute_inline_parameters): Update.
30536 (estimate_node_size_and_time): Update.
30537 (inline_merge_summary): Update.
30538 (inline_update_overall_summary): Update.
30539 (inline_read_section): Update.
30540 (inline_write_summary): Update.
30541 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30542 add account_size_time and reset member functions.
30543 (ipa_call_summary): Add reset function.
30544 * ipa-predicate.h (predicate::operator &): Constify.
30546 2017-05-22 Richard Biener <rguenther@suse.de>
30548 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30550 2017-05-19 Jason Merrill <jason@redhat.com>
30552 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30554 2017-05-19 Marek Polacek <polacek@redhat.com>
30557 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30558 TYPE_OVERFLOW_WRAPS checks.
30560 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30562 * tree-core.h (enum omp_clause_default_kind): Add
30563 "OMP_CLAUSE_DEFAULT_PRESENT".
30564 * tree-pretty-print.c (dump_omp_clause): Handle it.
30565 * gimplify.c (enum gimplify_omp_var_data): Add
30566 "GOVD_MAP_FORCE_PRESENT".
30567 (gimplify_adjust_omp_clauses_1): Map it to
30568 "GOMP_MAP_FORCE_PRESENT".
30569 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30571 * gimplify.c (oacc_default_clause): Clarify.
30573 2017-05-19 Nathan Sidwell <nathan@acm.org>
30575 LANG_HOOK_REGISTER_DUMPS
30576 * toplev.c (general_init): Call register dump lang hook.
30577 * doc/invoke.texi: Document -fdump-lang option family.
30578 * dumpfile.c (dump_files): Remove class dump here.
30579 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30580 * dumpfile.h (tree_dump_index): Remove TDI_class.
30581 * langhooks-def.h (lhd_register_dumps): Declare.
30582 (LANG_HOOKS_REGISTER_DUMPS): Define.
30583 (LANG_HOOKS_INITIALIZER): Add it.
30584 * langhooks.c (lhd_register_dumps): Define.
30585 * langhooks.h (struct lang_hooks): Add register_dumps.
30587 2017-05-19 Nathan Sidwell <nathan@acm.org>
30589 * context.h (context::set_passes): New.
30590 * context.c (context::context): Do not create pass manager.
30591 * toplev.c (general_init): Create pass manager here.
30593 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30595 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30596 use this splitter if two add or or instructions would also work for
30597 the constant we want to generate.
30599 2017-05-19 Richard Biener <rguenther@suse.de>
30602 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30603 predicate evaluation.
30605 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30607 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30609 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30612 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30614 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30615 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30616 (inline_edge_summary_vec): Turn into ...
30617 (ipa_call_summaries): ... this one.
30618 (redirect_to_unreachable, edge_set_predicate,
30619 evaluate_properties_for_edge, inline_summary_alloc,
30620 reset_ipa_call_summary, reset_inline_summary,
30621 inline_summary_t::duplicate): Update.
30622 (inline_edge_duplication_hook): Turn to ...
30623 (ipa_call_summary_t::duplicate): ... this one.
30624 (inline_edge_removal_hook): Turn to ...
30625 (ipa_call_summary_t::remove): ... this one.
30626 (dump_inline_edge_summary): Turn to ...
30627 (dump_ipa_call_summary): ... this one.
30628 (estimate_function_body_sizes): Update.
30629 (inline_update_callee_summaries): Update.
30630 (remap_edge_change_prob): Update.
30631 (remap_edge_summaries): Update.
30632 (inline_merge_summary): Update.
30633 (do_estimate_edge_time): Update.
30634 (inline_generate_summary): Update.
30635 (inline_read_section): Update.
30636 (inline_read_summary): Update.
30637 (inline_free_summary): Update.
30638 * ipa-inline.c (can_inline_edge_p): Update.
30639 (compute_inlined_call_time): Update.
30640 (want_inline_small_function_p): Update.
30641 (edge_badness): Update.
30642 (early_inliner): Update.
30643 * ipa-inline.h (inline_edge_summary): Turn to ...
30644 (ipa_call_summary): ... this one.
30645 (ipa_call_summary_t): New class.
30646 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30647 (ipa_call_summaries): New.
30648 (inline_edge_summary): Remove.
30649 (estimate_edge_growth): Update.
30650 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30651 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30652 * ipa-split.c (execute_split_functions): Update.
30653 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30655 2017-05-19 Richard Biener <rguenther@suse.de>
30657 PR middle-end/80764
30658 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30660 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30662 * config/rs6000/rs6000.c (struct machine_function): Add field
30663 fpr_is_wrapped_separately.
30664 (rs6000_get_separate_components): Use 64 components. Handle the
30665 new FPR components.
30666 (rs6000_components_for_bb): Handle the FPR components.
30667 (rs6000_emit_prologue_components): Handle the FPR components.
30668 (rs6000_emit_epilogue_components): Handle the FPR components.
30669 (rs6000_set_handled_components): Handle the FPR components.
30670 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30671 that are already separately shrink-wrapped.
30672 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30673 that are already separately shrink-wrapped.
30675 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30678 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30681 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30682 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30683 cases where the register allocator uses a move and an offsettable
30684 memory operation to/from a FPR register on ISA 2.06/2.07.
30685 (define_peephole2 for Altivec d-form store): Likewise.
30687 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30690 * config/i386/mmx.md (*mov<mode>_internal): Enable
30691 alternatives 11, 12, 13 and 14 also for 32bit targets.
30692 Remove alternatives 15, 16, 17 and 18.
30693 * config/i386/sse.md (vec_concatv2di): Change
30694 alternative (!x, *y) to (x, ?!*Yn).
30696 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30698 * dumpfile.h (enum dump_kind): Remove stray comma.
30700 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30702 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30703 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30704 predicate::num_conditions
30705 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30706 (CHANGED): turn into predicate::changed.
30707 (agg_position_info): Move to ipa-predicate.h
30708 (add_condition, predicate::add_clause, predicate::operator &=,
30709 predicate::or_with, predicate::evaluate, predicate::probability,
30710 dump_condition, dump_clause, predicate::dump,
30711 predicate::remap_after_duplication, predicate::remap_after_inlining,
30712 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30713 (evaluate_conditions_for_known_args): Update.
30714 (set_cond_stmt_execution_predicate): Update.
30715 * ipa-inline.h: Include ipa-predicate.h
30716 (condition, inline_param_summary, conditions, agg_position_info,
30717 predicate): Move to ipa-predicate.h
30718 * ipa-predicate.c: New file.
30719 * ipa-predicate.h: New file.
30721 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30723 * final.c (leaf_function_p): Check we are not in a sequence.
30725 2017-05-18 Martin Liska <mliska@suse.cz>
30727 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30728 * dumpfile.c (dump_register): Use new enum dump_kind.
30729 (get_dump_file_name): Likewise.
30730 (dump_enable_all): Likewise.
30731 (dump_switch_p_1): Likewise.
30732 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30733 * dumpfile.h (enum dump_kind): New enum type.
30734 (struct dump_file_info): Create constructor and
30735 format fields and comments.
30736 * passes.c (pass_manager::register_one_dump_file):
30738 * statistics.c (statistics_early_init): Likewise.
30739 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30740 TDF_TREE with TDF_SLIM.
30741 (gather_memory_references_ref): Likewise.
30743 2017-05-18 Martin Liska <mliska@suse.cz>
30745 * vec.h (struct vnull): Use it.
30747 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30749 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30750 (true_predicate, false_predicate, true_predicate_p,
30751 false_predicate_p): Remove.
30752 (single_cond_predicate, not_inlined_predicate): Turn to member function
30754 (add_condition): Update.
30755 (add_clause): Turn to...
30756 (predicate::add_clause): ... this one; update; allow passing NULL
30758 (and_predicates): Turn to ...
30759 (predicate::operator &=): ... this one.
30760 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30761 (or_predicates): Turn to ...
30762 (predicate::or_with): ... this one.
30763 (evaluate_predicate): Turn to ...
30764 (predicate::evaluate): ... this one.
30765 (predicate_probability): Turn to ...
30766 (predicate::probability): ... this one.
30767 (dump_condition): Update.
30768 (dump_predicate): Turn to ...
30769 (predicate::dump): ... this one.
30770 (account_size_time): Update.
30771 (edge_set_predicate): Update.
30772 (set_hint_predicate): UPdate.
30773 (evaluate_conditions_for_known_args): Update.
30774 (evaluate_properties_for_edge): Update.
30775 (remap_predicate_after_duplication): Turn to...
30776 (predicate::remap_after_duplication): ... this one.
30777 (remap_hint_predicate_after_duplication): Update.
30778 (inline_summary_t::duplicate): UPdate.
30779 (dump_inline_edge_summary): Update.
30780 (dump_inline_summary): Update.
30781 (set_cond_stmt_execution_predicate): Update.
30782 (set_switch_stmt_execution_predicate): Update.
30783 (compute_bb_predicates): Update.
30784 (will_be_nonconstant_expr_predicate): Update.
30785 (will_be_nonconstant_predicate): Update.
30786 (phi_result_unknown_predicate): Update.
30787 (predicate_for_phi_result): Update.
30788 (array_index_predicate): Update.
30789 (estimate_function_body_sizes): Update.
30790 (estimate_node_size_and_time): Update.
30791 (estimate_ipcp_clone_size_and_time): Update.
30792 (remap_predicate): Rename to ...
30793 (predicate::remap_after_inlining): ... this one.
30794 (remap_hint_predicate): Update.
30795 (inline_merge_summary): Update.
30796 (inline_update_overall_summary): Update.
30797 (estimate_size_after_inlining): Update.
30798 (read_predicate): Rename to ...
30799 (predicate::stream_in): ... this one.
30800 (read_inline_edge_summary): Update.
30801 (write_predicate): Rename to ...
30802 (predicate::stream_out): ... this one.
30803 (write_inline_edge_summary): Update.
30804 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
30805 (clause_t): Turn to uint32_t
30806 (predicate): Turn to class; implement constructor and operators
30808 (size_time_entry): Update.
30809 (inline_summary): Update.
30810 (inline_edge_summary): Update.
30812 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
30814 * fold-const.c (fold_binary_loc): Move transformation...
30815 * match.pd (C - X CMP X): ... here.
30817 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30819 * config/sparc/sparc.c (sparc_option_override): Set function
30820 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30821 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30823 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30825 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30827 2017-05-18 Marek Polacek <polacek@redhat.com>
30830 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30831 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30832 gimple_assign_load_p.
30834 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30836 PR middle-end/80692
30837 * real.c (do_compare): Give decimal_do_compare preference over
30838 comparing just the signs.
30840 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30842 * doc/md.texi (Canonicalization of Instructions): Describe the
30843 canonical form of instructions that inherently set a condition
30846 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
30848 PR middle-end/80775
30849 * tree-cfg.c: Move deletion of unreachable case statements to after
30850 the merging of consecutive case labels.
30852 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30854 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30855 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
30856 restoring of callee-saved registers.
30858 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
30860 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
30861 * config/visium/visium.c (single_set_and_flags): Likewise.
30862 * config/visium/visium.md (Substitutions): Likewise.
30864 2017-05-17 Martin Liska <mliska@suse.cz>
30866 * cfg.c: Introduce dump_flags_t type and
30867 use it instead of int type.
30869 * cfghooks.c: Likewise.
30870 * cfghooks.h (struct cfg_hooks): Likewise.
30871 * cfgrtl.c: Likewise.
30872 * cfgrtl.h: Likewise.
30873 * cgraph.c (cgraph_node::get_body): Likewise.
30874 * coretypes.h: Likewise.
30875 * domwalk.c: Likewise.
30876 * domwalk.h: Likewise.
30877 * dumpfile.c (struct dump_option_value_info): Likewise.
30878 (dump_enable_all): Likewise.
30879 (dump_switch_p_1): Likewise.
30880 (opt_info_switch_p): Likewise.
30881 * dumpfile.h (enum tree_dump_index): Likewise.
30882 (struct dump_file_info): Likewise.
30883 * genemit.c: Likewise.
30884 * generic-match-head.c: Likewise.
30885 * gengtype.c (open_base_files): Likewise.
30886 * gimple-pretty-print.c: Likewise.
30887 * gimple-pretty-print.h: Likewise.
30888 * graph.c (print_graph_cfg): Likewise.
30889 * graphite-scop-detection.c (dot_all_sese): Likewise.
30890 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30891 * loop-unroll.c (report_unroll): Likewise.
30892 * passes.c (pass_manager::register_one_dump_file): Likewise.
30893 * print-tree.c: Likewise.
30894 * statistics.c: Likewise.
30895 * tree-cfg.c: Likewise.
30896 * tree-cfg.h: Likewise.
30897 * tree-dfa.c: Likewise.
30898 * tree-dfa.h: Likewise.
30899 * tree-dump.c (dump_function): Likewise.
30900 * tree-dump.h (struct dump_info): Likewise.
30901 * tree-pretty-print.c: Likewise.
30902 * tree-pretty-print.h: Likewise.
30903 * tree-ssa-live.c: Likewise.
30904 * tree-ssa-live.h: Likewise.
30905 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30906 * tree-vect-loop.c: Likewise.
30907 * tree-vect-slp.c: Likewise.
30909 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30910 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30912 PR tree-optimization/80457
30913 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30914 of all arguments to a statement as scalar_to_vec operations.
30915 (vectorizable_call): Adjust call to vect_model_simple_cost for
30917 (vectorizable_conversion): Likewise.
30918 (vectorizable_assignment): Likewise.
30919 (vectorizable_shift): Likewise.
30920 (vectorizable_operation): Likewise.
30921 (vectorizable_comparison): Likewise.
30922 (vect_is_simple_cond): Record the def types for operands.
30923 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30924 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30925 for statement argument count.
30927 2017-05-16 Carl Love <cel@us.ibm.com>
30929 * config/rs6000/rs6000-c: Add support for built-in functions
30930 vector unsigned long long vec_bperm (vector unsigned long long,
30931 vector unsigned char)
30932 vector signed long long vec_mule (vector signed int,
30934 vector unsigned long long vec_mule (vector unsigned int,
30935 vector unsigned int)
30936 vector signed long long vec_mulo (vector signed int,
30938 vector unsigned long long vec_mulo (vector unsigned int,
30939 vector unsigned int)
30940 vector signed char vec_sldw (vector signed char,
30941 vector signed char,
30943 vector unsigned char vec_sldw (vector unsigned char,
30944 vector unsigned char,
30946 vector signed short vec_sldw (vector signed short,
30947 vector signed short,
30949 vector unsigned short vec_sldw (vector unsigned short,
30950 vector unsigned short,
30952 vector signed int vec_sldw (vector signed int,
30955 vector unsigned int vec_sldw (vector unsigned int,
30956 vector unsigned int,
30958 vector signed long long vec_sldw (vector signed long long,
30959 vector signed long long,
30961 vector unsigned long long vec_sldw (vector unsigned long long,
30962 vector unsigned long long,
30964 * config/rs6000/rs6000-c: Add support for built-in functions
30965 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30966 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30967 * doc/extend.texi: Update the built-in documentation for the
30968 new built-in functions.
30970 2017-05-16 Marek Polacek <polacek@redhat.com>
30974 * tree.c (save_expr): Don't fold the expression.
30976 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30978 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30979 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30980 and (?*y,m). Update insn attributes.
30982 2017-05-16 Martin Liska <mliska@suse.cz>
30984 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30985 flags argument of print_gimple_stmt, print_gimple_expr,
30986 print_generic_stmt and print_generic_expr.
30987 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30988 * coretypes.h: Likewise.
30989 * except.c (dump_eh_tree): Likewise.
30990 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30991 * gimple-pretty-print.h: Likewise.
30992 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30993 (backprop::push_to_worklist): Likewise.
30994 (backprop::pop_from_worklist): Likewise.
30995 (backprop::process_use): Likewise.
30996 (backprop::intersect_uses): Likewise.
30997 (note_replacement): Likewise.
30998 * gimple-ssa-store-merging.c
30999 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31000 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31001 (pass_store_merging::execute): Likewise.
31002 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31003 (ssa_base_cand_dump_callback): Likewise.
31004 (dump_incr_vec): Likewise.
31005 (replace_refs): Likewise.
31006 (replace_mult_candidate): Likewise.
31007 (create_add_on_incoming_edge): Likewise.
31008 (create_phi_basis): Likewise.
31009 (insert_initializers): Likewise.
31010 (all_phi_incrs_profitable): Likewise.
31011 (introduce_cast_before_cand): Likewise.
31012 (replace_one_candidate): Likewise.
31013 * gimplify.c (gimplify_expr): Likewise.
31014 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31015 (set_rename): Likewise.
31016 (rename_uses): Likewise.
31017 (copy_loop_phi_nodes): Likewise.
31018 (add_close_phis_to_merge_points): Likewise.
31019 (copy_loop_close_phi_args): Likewise.
31020 (copy_cond_phi_args): Likewise.
31021 (graphite_copy_stmts_from_block): Likewise.
31022 (translate_pending_phi_nodes): Likewise.
31023 * graphite-poly.c (print_pdr): Likewise.
31024 (dump_gbb_cases): Likewise.
31025 (dump_gbb_conditions): Likewise.
31026 (print_scop_params): Likewise.
31027 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31028 (build_cross_bb_scalars_use): Likewise.
31029 (gather_bbs::before_dom_children): Likewise.
31030 * hsa-dump.c (dump_hsa_immed): Likewise.
31031 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31032 (get_replacement_map): Likewise.
31033 * ipa-inline-analysis.c (dump_condition): Likewise.
31034 (estimate_function_body_sizes): Likewise.
31035 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31036 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31037 * ipa-prop.c (ipa_dump_param): Likewise.
31038 (ipa_print_node_jump_functions_for_edge): Likewise.
31039 (ipa_modify_call_arguments): Likewise.
31040 (ipa_modify_expr): Likewise.
31041 (ipa_dump_param_adjustments): Likewise.
31042 (ipa_dump_agg_replacement_values): Likewise.
31043 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31044 * ipa-pure-const.c (check_stmt): Likewise.
31045 (pass_nothrow::execute): Likewise.
31046 * ipa-split.c (execute_split_functions): Likewise.
31047 * omp-offload.c (dump_oacc_loop_part): Likewise.
31048 (dump_oacc_loop): Likewise.
31049 * trans-mem.c (tm_log_emit): Likewise.
31050 (tm_memopt_accumulate_memops): Likewise.
31051 (dump_tm_memopt_set): Likewise.
31052 (dump_tm_memopt_transform): Likewise.
31053 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31054 (print_loop): Likewise.
31055 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31056 (chkp_gather_checks_info): Likewise.
31057 (chkp_get_check_result): Likewise.
31058 (chkp_remove_check_if_pass): Likewise.
31059 (chkp_use_outer_bounds_if_possible): Likewise.
31060 (chkp_reduce_bounds_lifetime): Likewise.
31061 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31062 (chkp_mark_completed_bounds): Likewise.
31063 (chkp_register_incomplete_bounds): Likewise.
31064 (chkp_mark_invalid_bounds): Likewise.
31065 (chkp_maybe_copy_and_register_bounds): Likewise.
31066 (chkp_build_returned_bound): Likewise.
31067 (chkp_get_bound_for_parm): Likewise.
31068 (chkp_build_bndldx): Likewise.
31069 (chkp_get_bounds_by_definition): Likewise.
31070 (chkp_generate_extern_var_bounds): Likewise.
31071 (chkp_get_bounds_for_decl_addr): Likewise.
31072 * tree-chrec.c (chrec_apply): Likewise.
31073 * tree-data-ref.c (dump_data_reference): Likewise.
31074 (dump_subscript): Likewise.
31075 (dump_data_dependence_relation): Likewise.
31076 (analyze_overlapping_iterations): Likewise.
31077 * tree-inline.c (expand_call_inline): Likewise.
31078 (tree_function_versioning): Likewise.
31079 * tree-into-ssa.c (dump_defs_stack): Likewise.
31080 (dump_currdefs): Likewise.
31081 (dump_names_replaced_by): Likewise.
31082 (dump_update_ssa): Likewise.
31083 (update_ssa): Likewise.
31084 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31085 * tree-parloops.c (build_new_reduction): Likewise.
31086 (try_create_reduction_list): Likewise.
31087 (ref_conflicts_with_region): Likewise.
31088 (oacc_entry_exit_ok_1): Likewise.
31089 (oacc_entry_exit_single_gang): Likewise.
31090 * tree-pretty-print.h: Likewise.
31091 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31092 (get_scalar_evolution): Likewise.
31093 (add_to_evolution): Likewise.
31094 (get_loop_exit_condition): Likewise.
31095 (analyze_evolution_in_loop): Likewise.
31096 (analyze_initial_condition): Likewise.
31097 (analyze_scalar_evolution): Likewise.
31098 (instantiate_scev): Likewise.
31099 (number_of_latch_executions): Likewise.
31100 (gather_chrec_stats): Likewise.
31101 (final_value_replacement_loop): Likewise.
31102 (scev_const_prop): Likewise.
31103 * tree-sra.c (dump_access): Likewise.
31104 (disqualify_candidate): Likewise.
31105 (create_access): Likewise.
31106 (reject): Likewise.
31107 (maybe_add_sra_candidate): Likewise.
31108 (create_access_replacement): Likewise.
31109 (analyze_access_subtree): Likewise.
31110 (analyze_all_variable_accesses): Likewise.
31111 (sra_modify_assign): Likewise.
31112 (initialize_constant_pool_replacements): Likewise.
31113 (find_param_candidates): Likewise.
31114 (decide_one_param_reduction): Likewise.
31115 (replace_removed_params_ssa_names): Likewise.
31116 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31117 * tree-ssa-copy.c (dump_copy_of): Likewise.
31118 (copy_prop_visit_cond_stmt): Likewise.
31119 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31120 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31121 (record_equivalences_from_stmt): Likewise.
31122 * tree-ssa-dse.c (compute_trims): Likewise.
31123 (delete_dead_call): Likewise.
31124 (delete_dead_assignment): Likewise.
31125 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31126 (forward_propagate_into_cond): Likewise.
31127 (pass_forwprop::execute): Likewise.
31128 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31129 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31131 (move_computations_worker): Likewise.
31132 (execute_sm): Likewise.
31133 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31134 (remove_exits_and_undefined_stmts): Likewise.
31135 (remove_redundant_iv_tests): Likewise.
31136 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31137 (adjust_iv_update_pos): Likewise.
31138 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31139 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31140 (value_replacement): Likewise.
31141 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31142 * tree-ssa-pre.c (print_pre_expr): Likewise.
31143 (get_representative_for): Likewise.
31144 (create_expression_by_pieces): Likewise.
31145 (insert_into_preds_of_block): Likewise.
31146 (eliminate_insert): Likewise.
31147 (eliminate_dom_walker::before_dom_children): Likewise.
31148 (eliminate): Likewise.
31149 (remove_dead_inserted_code): Likewise.
31150 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31151 * tree-ssa-reassoc.c (get_rank): Likewise.
31152 (eliminate_duplicate_pair): Likewise.
31153 (eliminate_plus_minus_pair): Likewise.
31154 (eliminate_not_pairs): Likewise.
31155 (undistribute_ops_list): Likewise.
31156 (eliminate_redundant_comparison): Likewise.
31157 (update_range_test): Likewise.
31158 (optimize_range_tests_var_bound): Likewise.
31159 (optimize_vec_cond_expr): Likewise.
31160 (rewrite_expr_tree): Likewise.
31161 (rewrite_expr_tree_parallel): Likewise.
31162 (linearize_expr): Likewise.
31163 (break_up_subtract): Likewise.
31164 (linearize_expr_tree): Likewise.
31165 (attempt_builtin_powi): Likewise.
31166 (attempt_builtin_copysign): Likewise.
31167 (transform_stmt_to_copy): Likewise.
31168 (transform_stmt_to_multiply): Likewise.
31169 (dump_ops_vector): Likewise.
31170 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31171 (print_scc): Likewise.
31172 (set_ssa_val_to): Likewise.
31173 (visit_reference_op_store): Likewise.
31174 (visit_use): Likewise.
31175 (sccvn_dom_walker::before_dom_children): Likewise.
31176 (run_scc_vn): Likewise.
31177 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31179 (expr_hash_elt::print): Likewise.
31180 (const_and_copies::pop_to_marker): Likewise.
31181 (const_and_copies::record_const_or_copy_raw): Likewise.
31182 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31183 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31184 (dump_predicates): Likewise.
31185 (find_uninit_use): Likewise.
31186 (warn_uninitialized_phi): Likewise.
31187 (pass_late_warn_uninitialized::execute): Likewise.
31188 * tree-ssa.c (verify_vssa): Likewise.
31189 (verify_ssa): Likewise.
31190 (maybe_optimize_var): Likewise.
31191 * tree-vrp.c (dump_value_range): Likewise.
31192 (dump_all_value_ranges): Likewise.
31193 (dump_asserts_for): Likewise.
31194 (register_edge_assert_for_2): Likewise.
31195 (vrp_visit_cond_stmt): Likewise.
31196 (vrp_visit_switch_stmt): Likewise.
31197 (vrp_visit_stmt): Likewise.
31198 (vrp_visit_phi_node): Likewise.
31199 (simplify_cond_using_ranges_1): Likewise.
31200 (fold_predicate_in): Likewise.
31201 (evrp_dom_walker::before_dom_children): Likewise.
31202 (evrp_dom_walker::push_value_range): Likewise.
31203 (evrp_dom_walker::pop_value_range): Likewise.
31204 (execute_early_vrp): Likewise.
31206 2017-05-16 Richard Biener <rguenther@suse.de>
31208 * dwarf2out.c (loc_list_from_tree_1): Do not create
31209 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31211 2017-05-16 Richard Biener <rguenther@suse.de>
31213 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31215 (note_variable_value_in_expr): If we resolved the decl ref
31216 do not push to the stack.
31218 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31220 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31221 operations in fast-math mode.
31222 (vaddq_f16): Likewise.
31223 (vmul_f16): Likewise.
31224 (vmulq_f16): Likewise.
31225 (vsub_f16): Likewise.
31226 (vsubq_f16): Likewise.
31227 * config/arm/neon.md (add<mode>3): New.
31229 (fma:<VH:mode>3): New. Also remove outdated comment.
31232 2017-05-16 Martin Liska <mliska@suse.cz>
31236 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31237 (odr_types_equivalent_p): Likewise.
31239 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31241 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31243 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31246 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31247 non-interunit SSE move alternatives with '?'.
31248 (zero-extendsidi peephole2): New peephole to skip intermediate
31249 general register in SSE zero-extend sequence.
31251 2017-05-15 Jeff Law <law@redhat.com>
31253 * reorg.c (relax_delay_slots): Create a new variable to hold
31254 the temporary target rather than clobbering TARGET_LABEL.
31256 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31257 missing argument to extract_bit_field call.
31258 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31260 2017-05-15 Martin Liska <mliska@suse.cz>
31263 * gcc.c (process_command): Do not allow empty argument of -o option.
31265 2017-05-15 Renlin Li <renlin.li@arm.com>
31267 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31268 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31269 * config/aarch64/constraints.md (Usf): Add long call check.
31270 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31271 (call_value): Likewise.
31272 (sibcall): Likewise.
31273 (sibcall_value): Likewise.
31275 (call_value_insn): New.
31276 (sibcall_insn): Update rtx pattern.
31277 (sibcall_value_insn): Likewise.
31278 (call_internal): Remove.
31279 (call_value_internal): Likewise.
31280 (sibcall_internal): Likewise.
31281 (sibcall_value_internal): Likewise.
31282 (call_reg): Likewise.
31283 (call_symbol): Likewise.
31284 (call_value_reg): Likewise.
31285 (call_value_symbol): Likewise.
31287 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31290 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31292 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31294 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31295 compatible with CCGOCmode and with CCZmode.
31297 2017-05-14 Martin Sebor <msebor@redhat.com>
31299 PR middle-end/77671
31300 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31301 (gimple_fold_builtin_snprintf): Same.
31302 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31303 (gimple_fold_builtin_snprintf): Same.
31304 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31305 of character types.
31306 (is_call_safe): New function.
31307 (try_substitute_return_value): Call it.
31308 (try_simplify_call): New function.
31309 (pass_sprintf_length::handle_gimple_call): Call it.
31311 2017-05-14 Martin Sebor <msebor@redhat.com>
31313 PR middle-end/80669
31314 * builtins.c (expand_builtin_stpncpy): Simplify.
31316 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31318 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31319 * config/i386/i386.h
31320 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31321 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31322 (struct machine_function): Add new members call_ms2sysv,
31323 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31324 (struct machine_frame_state): New fields sp_realigned and
31325 sp_realigned_offset.
31326 * config/i386/i386.c
31327 (enum xlogue_stub): New enum.
31328 (enum xlogue_stub_sets): New enum.
31329 (class xlogue_layout): New class.
31330 (struct ix86_frame): New fields stack_realign_allocate_offset,
31331 stack_realign_offset and outlined_save_offset. Modify comments to
31332 detail stack layout when using out-of-line stubs.
31333 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31334 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31335 -mcall-ms2sysv-xlogues.
31336 (stub_managed_regs): New static variable.
31337 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31338 registers managed by out-of-line stub.
31339 (disable_call_ms2sysv_xlogues): New function.
31340 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31341 m->call_ms2sysv when appropriate and compute frame layout for
31343 (sp_valid_at, fp_valid_at): New inline functions.
31344 (choose_basereg): New function.
31345 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31347 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31348 Use align parameter of choose_baseaddr to generated aligned SSE movs
31350 (pro_epilogue_adjust_stack): Modify to track
31351 machine_frame_state::sp_realigned.
31352 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31353 (ix86_nsaved_sseregs): Likewise.
31354 (ix86_emit_save_regs): Likewise.
31355 (ix86_emit_save_regs_using_mov): Likewise.
31356 (ix86_emit_save_sse_regs_using_mov): Likewise.
31357 (get_scratch_register_on_entry): Likewise.
31358 (gen_frame_set): New function.
31359 (gen_frame_load): Likewise.
31360 (gen_frame_store): Likewise.
31361 (emit_outlined_ms2sysv_save): Likewise.
31362 (emit_outlined_ms2sysv_restore): Likewise.
31363 (ix86_expand_prologue): Modify stack re-alignment code and call
31364 emit_outlined_ms2sysv_save when appropriate.
31365 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31366 parameter rtx_insn *insn, which allows the function to be used to only
31367 generate the notes.
31368 (ix86_expand_epilogue): Modify validity checks of frame and stack
31369 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31370 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31371 * config/i386/predicates.md
31372 (save_multiple): New predicate.
31373 (restore_multiple): Likewise.
31374 * config/i386/sse.md
31375 (save_multiple<mode>): New pattern.
31376 (save_multiple_realign<mode>): Likewise.
31377 (restore_multiple<mode>): Likewise.
31378 (restore_multiple_and_return<mode>): Likewise.
31379 (restore_multiple_leave_return<mode>): Likewise.
31380 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31382 2017-05-14 Julia Koval <julia.koval@intel.com>
31384 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31385 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31386 (__builtin_ia32_xsetbv): New builtins.
31387 * config/i386/i386.c (ix86_expand_special_args_builtin):
31389 (ix86_expand_builtin): Special expand for new intrinsics.
31390 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31391 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31392 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31394 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31396 * cfganal.c (inverted_post_order_compute): Change argument type
31398 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31399 * df-core.c (rest_of_handle_df_initialize): Adjust.
31400 (rest_of_handle_df_finish): Likewise.
31401 (df_analyze_1): Likewise.
31402 (df_analyze): Likewise.
31403 (loop_inverted_post_order_compute): Change argument to be a vec *.
31404 (df_analyze_loop): Adjust.
31405 (df_get_n_blocks): Likewise.
31406 (df_get_postorder): Likewise.
31407 * df.h (struct df_d): Change field to be a vec.
31408 * lcm.c (compute_laterin): Adjust.
31409 (compute_available): Likewise.
31410 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31411 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31412 * tree-ssa-pre.c (compute_antic): Likewise.
31414 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31416 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31417 (depth_first_search::depth_first_search): Change structure init
31418 function to this constructor.
31419 (depth_first_search::add_bb): Rename function to this member.
31420 (depth_first_search::execute): Likewise.
31421 (flow_dfs_compute_reverse_finish): Adjust.
31423 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31425 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31426 (longest_simple_path): Likewise.
31427 * shrink-wrap.c (spread_components): Likewise.
31428 (disqualify_problematic_components): Likewise.
31429 (emit_common_heads_for_components): Likewise.
31430 (emit_common_tails_for_components): Likewise.
31431 (insert_prologue_epilogue_for_components): Likewise.
31433 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31435 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31438 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31440 * df-core.c (df_set_blocks): Start using auto_bitmap.
31441 (df_compact_blocks): Likewise.
31442 * df-problems.c (df_rd_confluence_n): Likewise.
31443 * df-scan.c (df_insn_rescan_all): Likewise.
31444 (df_process_deferred_rescans): Likewise.
31445 (df_update_entry_block_defs): Likewise.
31446 (df_update_exit_block_uses): Likewise.
31447 (df_entry_block_bitmap_verify): Likewise.
31448 (df_exit_block_bitmap_verify): Likewise.
31449 (df_scan_verify): Likewise.
31450 * lra-constraints.c (lra_constraints): Likewise.
31451 (undo_optional_reloads): Likewise.
31452 (lra_undo_inheritance): Likewise.
31453 * lra-remat.c (calculate_gen_cands): Likewise.
31454 (do_remat): Likewise.
31455 * lra-spills.c (assign_spill_hard_regs): Likewise.
31456 (spill_pseudos): Likewise.
31457 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31458 (bitmap_set_subtract_values): Likewise.
31460 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31462 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31463 management with auto_bitmap.
31464 (fix_inter_tick): Likewise.
31465 (fix_recovery_deps): Likewise.
31466 * ira.c (add_store_equivs): Likewise.
31467 (find_moveable_pseudos): Likewise.
31468 (split_live_ranges_for_shrink_wrap): Likewise.
31469 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31470 (rtx_reuse_manager::seen_def_p): Likewise.
31471 (rtx_reuse_manager::set_seen_def): Likewise.
31472 * print-rtl.h (class rtx_reuse_manager): Likewise.
31474 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31476 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31478 (migrate_btr_def): Likewise.
31479 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31480 * df-core.c (loop_post_order_compute): Likewise.
31481 (loop_inverted_post_order_compute): Likewise.
31482 * hsa-common.h: Likewise.
31483 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31484 * init-regs.c (initialize_uninitialized_regs): Likewise.
31485 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31486 (inline_small_functions): Likewise.
31487 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31488 * ira.c (combine_and_move_insns): Likewise.
31489 (build_insn_chain): Likewise.
31490 * loop-invariant.c (find_invariants): Likewise.
31491 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31492 * predict.c (tree_predict_by_opcode): Likewise.
31493 (predict_paths_leading_to): Likewise.
31494 (predict_paths_leading_to_edge): Likewise.
31495 (estimate_loops_at_level): Likewise.
31496 (estimate_loops): Likewise.
31497 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31498 (spread_components): Likewise.
31499 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31500 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31501 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31502 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31503 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31504 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31505 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31506 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31507 (mark_threaded_blocks): Likewise.
31508 (thread_through_all_blocks): Likewise.
31509 * tree-ssa.c (verify_ssa): Likewise.
31510 (execute_update_addresses_taken): Likewise.
31511 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31513 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31515 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31517 (post_order_compute): Likewise.
31518 (inverted_post_order_compute): Likewise.
31519 (pre_and_rev_post_order_compute_fn): Likewise.
31521 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31523 * genrecog.c (int_set::int_set): Explicitly construct our
31524 auto_vec base class.
31525 * vec.h (auto_vec::auto_vec): New constructor.
31527 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31529 * bitmap.h (class auto_bitmap): New constructor taking
31530 bitmap_obstack * argument.
31532 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31534 * bitmap.h (class auto_bitmap): Change type of m_bits to
31535 bitmap_head, and adjust ctor / dtor and member operators.
31537 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31539 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31540 when returned register mode doesn't match original mode.
31542 2017-05-12 Jeff Law <law@redhat.com>
31543 Jakub Jelinek <jakub@redhat.com>
31545 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31546 we look for cc setter after the compare-elim changes.
31547 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31548 within the vector to match what compare-elim now expects.
31549 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31550 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31552 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31553 after the compare-elim changes.
31554 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31555 the vector to match what compare-elim now expects.
31556 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31557 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31558 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31559 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31560 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31562 * config/visium/visium.c (single_set_and_flags): Fix where
31563 we look for cc setter after the compare-elim changes.
31564 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31565 with the vector to match what compare-elim now expects.
31566 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31567 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31568 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31569 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31570 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31571 (neg<mode>2_insn_set_overflow): Likewise.
31573 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31575 PR middle-end/79794
31576 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31577 maybe_expand_insn call, set ops[0].target. If still set after call,
31578 set alt_rtl. Add extra arg to recursive calls.
31579 (extract_bit_field): Add alt_rtl argument. Pass to
31581 * expmed.h (extract_bit_field): Fix prototype.
31582 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31583 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31584 to extract_bit_field_calls.
31585 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31586 Pass alt_rtl to extract_bit_field calls.
31587 * calls.c (store_unaligned_arguments_into_psuedos)
31588 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31589 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31591 * optabs.h (struct expand_operand): Add target bitfield.
31593 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31595 * compare-elim.c (try_eliminate_compare): Canonicalize
31596 operation with embedded compare to
31597 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31598 (set (reg) (operation)].
31600 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31602 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31605 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31606 cost of adding a carry flag for ADC instruction.
31607 [case MINUS]: Ignore the cost of subtracting a carry flag
31608 for SBB instruction.
31610 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31612 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31614 * config/rs6000/bmiintrin.h: New file.
31615 * config/rs6000/bmi2intrin.h: New file.
31616 * config/rs6000/x86intrin.h: New file.
31618 2017-05-12 Jeff Law <law@redhat.com>
31620 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31623 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31625 PR middle-end/80707
31626 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31628 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31630 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31631 early expansion of vector divide builtins.
31632 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31633 builtins identified as having unsigned arguments.
31635 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31637 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31638 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31639 expansion of vector logical operations (and, andc, or, xor,
31642 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31644 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31645 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31647 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31649 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31650 early GIMPLE expansion of vector multiplies.
31652 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31654 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31655 TARGET_HAVE_MOVT conditional.
31656 (movt splitter): Likewise.
31658 2017-05-12 Richard Biener <rguenther@suse.de>
31660 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31661 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31662 Fold all stmts not inplace.
31664 2017-05-12 Richard Biener <rguenther@suse.de>
31666 PR tree-optimization/80713
31667 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31668 inserted_exprs bit for not removed stmts.
31670 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31672 PR middle-end/69921
31673 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31674 parallelized" attribute for parallelized OpenACC kernels.
31675 * omp-offload.c (execute_oacc_device_lower): Use it.
31677 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31678 Set "oacc kernels" attribute.
31679 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31680 parameter. Adjust all users.
31681 (oacc_fn_attrib_kernels_p): Remove function.
31682 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31683 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31684 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31685 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31686 assert "oacc kernels" attribute is set.
31688 2017-05-11 Carl Love <cel@us.ibm.com>
31690 * config/rs6000/rs6000-c: Add support for built-in functions
31691 vector unsigned char vec_popcnt (vector signed char)
31692 vector unsigned char vec_popcnt (vector unsigned char)
31693 vector unsigned short vec_popcnt (vector signed short)
31694 vector unsigned short vec_popcnt (vector unsigned short)
31695 vector unsigned int vec_popcnt (vector signed int)
31696 vector unsigned int vec_popcnt (vector unsigned int)
31697 vector unsigned long long vec_popcnt (vector signed long long)
31698 vector unsigned long long vec_popcnt (vector unsigned long long)
31699 vector signed long long vec_slo (vector signed long long,
31700 vector signed char)
31701 vector signed long long vec_slo (vector signed long long,
31702 vector unsigned char)
31703 vector unsigned long long vec_slo (vector unsigned long long,
31704 vector signed char)
31705 vector unsigned long long vec_slo (vector unsigned long long,
31706 vector unsigned char)
31707 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31708 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31709 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31710 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31711 * doc/extend.texi: Update the built-in documentation file for the
31712 new built-in functions.
31714 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31716 * attribs.h (sorted_attr_string): Move machine independent
31717 functions for target clone support from the i386 port to common
31718 code. Rename ix86_function_versions to common_function_versions.
31719 Rename make_name to make_unique_name.
31720 (common_function_versions): Likewise.
31721 (make_unique_name): Likewise.
31722 (make_dispatcher_decl): Likewise.
31723 (is_function_default_version): Likewise.
31724 * attribs.c (attr_strcmp): Likewise.
31725 (sorted_attr_string): Likewise.
31726 (common_function_versions): Likewise.
31727 (make_unique_name): Likewise.
31728 (make_dispatcher_decl): Likewise.
31729 (is_function_default_version): Likewise.
31730 * config/i386/i386.c (attr_strcmp): Likewise.
31731 (sorted_attr_string): Likewise.
31732 (ix86_function_versions): Likewise.
31733 (make_name): Likewise.
31734 (make_dispatcher_decl): Likewise.
31735 (is_function_default_version): Likewise.
31736 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31738 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31741 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31742 Account for direct move costs for vec_construct of integer
31745 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31748 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31749 (UNSPEC_STX_ATOMIC): Ditto.
31750 (loaddi_via_sse): New insn.
31751 (storedi_via_sse): Ditto.
31752 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31753 Update corresponding peephole2 patterns.
31754 (atomic_storedi_fpu): Ditto.
31756 2017-05-11 Julia Koval <julia.koval@intel.com>
31758 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31759 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31761 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31762 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31763 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31765 2017-05-11 Nathan Sidwell <nathan@acm.org>
31767 * graphite-poly.c: Include dumpfile.h.
31769 * dumpfle.h (dump_function): Declare here ...
31770 * tree-dump.h (dump_function): ... not here.
31771 * dumpfile.c: #include tree-cfg.h.
31772 (dump_function): Move here from ...
31773 * tree-dump.c (dump_function): ... here.
31774 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31775 * graphite-poly.c: Don't include tree-dump.h.
31776 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31777 * print-tree.c: Likewise.
31778 * stor-layout.c: Likewise.
31779 * tree-nested.c: Likewise.
31781 * dumpfile.c (dump_start): Use TDF_FLAGS.
31782 (dump_enable_all): Fix TDF_KIND check thinko.
31784 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31786 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31787 array entries to represent two legal parameterizations of the
31788 overloaded __builtin_cmpb function, as represented by the
31789 P6_OV_BUILTIN_CMPB constant.
31790 (altivec_resolve_overloaded_builtin): Add special case handling
31791 for the __builtin_cmpb function, as represented by the
31792 P6_OV_BUILTIN_CMPB constant.
31793 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
31794 (BU_P6_64BIT_2): New macro.
31795 (BU_P6_OVERLOAD_2): New macro
31796 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
31797 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
31798 (CMPB): Add overload support to represent both 32-bit and 64-bit
31799 compare-bytes function.
31800 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
31801 support for TARGET_CMPB.
31802 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
31803 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31804 documentation of the __builtin_cmpb overloaded built-in function.
31806 2017-05-11 Richard Biener <rguenther@suse.de>
31808 PR tree-optimization/80705
31809 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
31810 bases are not vectorizable.
31812 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31814 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
31815 when counting register pressure.
31817 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31819 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31820 (iv_ca_more_deps): Renamed to ...
31821 (iv_ca_compare_deps): ... this.
31822 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31824 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31826 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31828 (determine_group_iv_costs): ... here.
31829 (find_inv_vars_cb): Record inv var if it's not recorded before.
31831 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31833 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31834 (get_shiftadd_cost): Ditto.
31836 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31838 * tree-ssa-address.c: Include header file.
31839 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31841 (add_to_parts): Refactor.
31842 (addr_to_parts): New parameter. Update use of move_hint_to_base.
31843 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
31846 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31848 PR tree-optimization/53090
31849 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31851 (extract_cond_operands): Detect condition with IV on both sides
31852 and return COMP_IV_EXPR_2.
31853 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31854 (rewrite_use_compare): Simplify by removing call to function
31855 extract_cond_operands.
31857 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31859 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
31860 (extract_cond_operands): Detect condition comparing against non-
31861 invariant bound and return appropriate enum value.
31862 (find_interesting_uses_cond): Update use of extract_cond_operands.
31863 Handle its return value accordingly.
31864 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
31866 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31868 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
31869 nonlinear iv_use computation in loop invariant sensitive way.
31871 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31873 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
31874 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31876 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31878 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31879 (dump_cand): Support iv_cand.inv_exprs.
31880 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31882 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31885 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31887 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31889 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31890 as local function. Include necessary header files.
31891 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31893 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31895 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31897 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31899 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31900 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31901 RSHIFT_EXPR and BIT_NOT_EXPR.
31903 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31905 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31906 (adjust_setup_cost): New parameter supporting round up adjustment.
31907 (struct address_cost_data): Delete.
31908 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31909 (split_address_cost, ptr_difference_cost): Delete.
31910 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31911 (struct ainc_cost_data): New struct.
31912 (get_address_cost_ainc): New function.
31913 (get_address_cost, get_computation_cost): Reimplement.
31914 (determine_group_iv_cost_address): Record inv_expr for all uses of
31916 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31917 (iv_ca_has_deps): Reimplemented to ...
31918 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31920 (iv_ca_extend): Call iv_ca_more_deps.
31922 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31924 * tree-ssa-address.c (struct mem_address): Move to header file.
31925 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31926 * tree-ssa-address.h (struct mem_address): Move from C file.
31927 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31929 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31931 * tree-affine.h (aff_combination_type): New interface.
31932 (aff_combination_zero_p): Remove static.
31933 (aff_combination_const_p): New interface.
31934 (aff_combination_singleton_var_p): New interfaces.
31936 2017-05-11 Richard Biener <rguenther@suse.de>
31938 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31939 Skip unreachable blocks and destinations.
31940 (eliminate): Move stmt removal and fixup ...
31941 (fini_eliminate): ... here. Skip inserted exprs.
31942 (pass_pre::execute): Move fini_pre after fini_eliminate.
31943 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31944 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31945 PRE to get rid of dead code that has invalid SSA form and
31946 split critical edges again.
31948 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31950 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31952 2017-05-11 Richard Biener <rguenther@suse.de>
31954 * passes.c (execute_function_todo): Verify loops if they are
31955 said to be up-to-date.
31956 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31957 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31959 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31962 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31963 handle calling assemble_external ourself.
31966 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31967 modes with zero size. Enhance comment.
31969 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31971 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31972 built-ins for vec_xl and vec_xst with short and char pointer
31975 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31977 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31978 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31979 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31980 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31981 (_mm_maskz_min_round_ss): New intrinsics.
31982 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31983 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31984 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31985 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31986 (__builtin_ia32_minss_mask_round): New builtins.
31987 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31988 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31989 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31991 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31992 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31994 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31997 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31999 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32000 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32001 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32002 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32003 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32004 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32005 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32006 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32007 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32008 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32009 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32010 (__builtin_ia32_mulss_mask_round): New builtins.
32011 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32012 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32013 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32015 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32016 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32018 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32021 2017-05-10 Julia Koval <julia.koval@intel.com>
32023 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32024 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32025 (_mm256_setr_m128i): New intrinsics.
32027 2017-05-10 Julia Koval <julia.koval@intel.com>
32029 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32030 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32031 (_mm_maskz_rcp14_ss): New intrinsics.
32032 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32033 (__builtin_ia32_rcp14ss_mask): New builtins.
32034 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32036 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32038 PR tree-optimization/51513
32039 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32040 (assert_unreachable_fallthru_edge_p): Use it.
32041 (group_case_labels_stmt): Likewise.
32042 * tree-cfg.h: Prototype it.
32043 * stmt.c: Include cfghooks.h and tree-cfg.h.
32044 (emit_case_dispatch_table) <gap_label>: New local variable.
32045 Use it to fill dispatch table gaps.
32046 Test for default_label before updating probabilities.
32047 (expand_case) <default_label>: Remove unneeded initialization.
32048 Test for unreachable default case statement and remove its edge.
32049 Set default_label accordingly.
32050 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32052 2017-05-10 Carl Love <cel@us.ibm.com>
32054 * config/rs6000/rs6000-c: Add support for built-in functions
32055 vector signed char vec_neg (vector signed char)
32056 vector signed short int vec_neg (vector short int)
32057 vector signed int vec_neg (vector signed int)
32058 vector signed long long vec_neg (vector signed long long)
32059 vector float vec_neg (vector float)
32060 vector double vec_neg (vector double)
32061 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32063 * config/rs6000/altivec.h: Add define for vec_neg
32064 * doc/extend.texi: Update the built-in documentation for the
32065 new built-in functions.
32067 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32069 PR tree-optimization/77644
32070 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32072 2017-05-10 Nathan Sidwell <nathan@acm.org>
32074 * dumpfile.h (TDI_lang_all): New.
32075 (TDF_KIND): New. Renumber others
32076 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32078 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32080 (get_dump_file_name): Adjust suffix generation.
32081 (dump_enable_all): Use TDF_KIND.
32082 * doc/invoke.texi (-fdump-lang-all): Document.
32084 * dumpfile.h: Tabify.
32086 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32089 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32090 Move member access before delete.
32092 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32094 * tree-inline.c (expand_call_inline): Split block at stmt
32097 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32100 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32101 are now unused after splitting mov{sf,sd}_hardfloat.
32102 (f32_lr2): Likewise.
32103 (f32_lm): Likewise.
32104 (f32_lm2): Likewise.
32105 (f32_li): Likewise.
32106 (f32_li2): Likewise.
32107 (f32_lv): Likewise.
32108 (f32_sr): Likewise.
32109 (f32_sr2): Likewise.
32110 (f32_sm): Likewise.
32111 (f32_sm2): Likewise.
32112 (f32_si): Likewise.
32113 (f32_si2): Likewise.
32114 (f32_sv): Likewise.
32115 (f32_dm): Likewise.
32116 (f32_vsx): Likewise.
32117 (f32_av): Likewise.
32118 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32119 For movsf, order stores so the VSX stores occur before the GPR
32120 store which encourages the register allocator to use a traditional
32121 FPR instead of a GPR. For movsd, order the stores so that the GPR
32122 store comes before the VSX stores to allow the power6 to work.
32123 This is due to the power6 not having a 32-bit integer store
32124 instruction from a FPR.
32125 (movsf_hardfloat): Likewise.
32126 (movsd_hardfloat): Likewise.
32128 2017-05-09 Martin Sebor <msebor@redhat.com>
32130 PR translation/80280
32131 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32134 PR translation/80280
32135 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32136 data member added in r247778.
32137 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32139 2017-05-09 Nathan Sidwell <nathan@acm.org>
32141 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32143 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32146 2017-05-09 Marek Polacek <polacek@redhat.com>
32148 * doc/invoke.texi: Fix typo.
32150 2017-05-09 Richard Biener <rguenther@suse.de>
32152 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32153 (vrp_val_is_min): Likewise.
32154 (set_value_range_to_value): Likewise.
32155 (set_value_range_to_nonnegative): Likewise.
32156 (gimple_assign_nonzero_p): Likewise.
32157 (gimple_stmt_nonzero_p): Likewise.
32158 (vrp_int_const_binop): Likewise. Remove unreachable case.
32159 (adjust_range_with_scev): Adjust comments.
32160 (compare_range_with_value): Likewise.
32161 (extract_range_from_phi_node): Likewise.
32162 (test_for_singularity): Likewise.
32164 2017-05-09 Richard Biener <rguenther@suse.de>
32166 * tree-vrp.c (get_single_symbol): Add assert that we don't
32167 get overflowed constants as invariant part.
32168 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32169 checks. Use wi::cmp instead of recursing for integer constants.
32170 (compare_values): Just ignore whether we assumed undefined
32171 overflow instead of failing the compare.
32172 (extract_range_for_var_from_comparison_expr): Add comment before the
32173 TREE_NO_WARNING sets.
32174 (test_for_singularity): Likewise.
32175 (extract_range_from_comparison): Do not disable optimization
32176 when we assumed undefined overflow.
32177 (extract_range_basic): Remove init of unused var.
32179 2017-05-09 Richard Biener <rguenther@suse.de>
32181 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32182 (extract_range_from_multiplicative_op_1): Adjust.
32183 (extract_range_from_binary_expr_1): Use int_const_binop.
32185 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32188 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32189 rs6000_store_data_bypass_p in seven define_bypass directives and
32190 in several comments.
32191 * config/rs6000/rs6000-protos.h: Add prototype for
32192 rs6000_store_data_bypass_p function.
32193 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32194 function implements slightly different (rs6000-specific) semantics
32195 than store_data_bypass_p, returning false rather than aborting
32196 with assertion error when arguments do not satisfy the
32197 requirements of store data bypass.
32198 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32199 rs6000_store_data_bypass_p.
32201 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32203 * config/xtensa/xtensa-protos.h
32204 (xtensa_initial_elimination_offset): New declaration.
32205 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32206 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32207 macro definition, add case for FRAME_POINTER_REGNUM when
32208 FRAME_GROWS_DOWNWARD.
32209 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32210 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32211 xtensa_initial_elimination_offset.
32213 2017-05-08 Nathan Sidwell <nathan@acm.org>
32215 * doc/invoke.texi: Alphabetize -fdump options.
32217 2017-05-08 Martin Sebor <msebor@redhat.com>
32219 PR translation/80280
32220 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32222 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32224 * target.def (compute_frame_layout): New optional target hook.
32225 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32226 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32227 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32229 * reload1.c (verify_initial_elim_offsets): Likewise.
32230 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32231 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32232 (arm_get_frame_offsets): Split up into this ...
32233 (arm_compute_frame_layout): ... and this function.
32235 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32237 * config/aarch64/constraints.md (Usa): New constraint.
32238 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32240 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32242 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32243 with_multilib_list after it has been checked.
32245 2017-05-08 Richard Biener <rguenther@suse.de>
32247 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32248 (bitmap_set_subtract_values): Likewise.
32250 2017-05-08 Richard Biener <rguenther@suse.de>
32252 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32253 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32255 (gimple_stmt_nonzero_warnv_p): Rename to ...
32256 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32258 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32259 (extract_range_basic): Adjust, do not disable propagation on
32260 strict overflow sensitive simplification.
32261 (vrp_visit_cond_stmt): Likewise.
32263 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32265 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32266 body size unconditionally.
32268 2017-05-07 Jeff Law <law@redhat.com>
32271 2017-05-06 Jeff Law <law@redhat.com>
32272 PR tree-optimization/78496
32273 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32276 PR tree-optimization/78496
32277 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32278 (simplify_stmt_using_ranges): Call it.
32279 (vrp_dom_walker::before_dom_children): Extract equivalences
32280 from an ASSERT_EXPR with an equality comparison against a
32283 2017-05-06 Jeff Law <law@redhat.com>
32285 PR tree-optimization/78496
32286 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32289 PR tree-optimization/78496
32290 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32291 (simplify_stmt_using_ranges): Call it.
32292 (vrp_dom_walker::before_dom_children): Extract equivalences
32293 from an ASSERT_EXPR with an equality comparison against a
32296 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32298 * lra-constraints.c (lra_copy_reg_equiv): New function.
32299 (split_reg): Use it to copy equivalence information from the
32300 original register to the spill register.
32302 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32304 PR rtl-optimization/75964
32305 * simplify-rtx.c (simplify_const_relational_operation): Remove
32306 invalid handling of comparisons of integer ABS.
32308 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32310 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32311 initialize to zero.
32312 (init_regs): Remove declaration.
32313 (function_arg_advance_32): Initialize error_p as boolean variable.
32315 2017-05-05 Nathan Sidwell <nathan@acm.org>
32317 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32318 lines. Use for (;;).
32320 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32322 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32323 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32324 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32325 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32326 VF=2 that require versioning.
32328 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32330 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32333 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32335 * diagnostic.h (diagnostic_override_option_index): Convert from
32336 macro to inline function.
32338 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32340 * diagnostic.c (last_module_changed_p): New function.
32341 (set_last_module): New function.
32342 (diagnostic_report_current_module): Convert macro usage to
32343 the above functions.
32344 * diagnostic.h (diagnostic_context::last_module): Strengthen
32345 from const line_map * to const line_map_ordinary *.
32346 (diagnostic_last_module_changed): Delete macro.
32347 (diagnostic_set_last_module): Delete macro.
32349 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32351 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32352 with diagnostic_report_diagnostic.
32353 (diagnostic_n_impl_richloc): Likewise.
32354 * diagnostic.h (report_diagnostic): Delete macro.
32355 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32356 with diagnostic_report_diagnostic.
32357 * substring-locations.c (format_warning_va): Likewise.
32359 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32361 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32362 save/restor of format_spec. Move option-printing code to...
32363 (print_option_information): ...this new function, and
32364 reimplement by simply printing to the pretty_printer,
32365 rather than appending to the format string.
32367 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32369 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32370 handling logic into...
32371 (update_effective_level_from_pragmas): ...this new function.
32373 2017-05-04 Andrew Waterman <andrew@sifive.com>
32375 * config/riscv/riscv.opt (mstrict-align): New option.
32376 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32377 (SLOW_UNALIGNED_ACCESS): Define.
32378 (riscv_slow_unaligned_access): Declare.
32379 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32381 (riscv_slow_unaligned_access): New variable.
32382 (rocket_tune_info): Set slow_unaligned_access to true.
32383 (optimize_size_tune_info): Set slow_unaligned_access to false.
32384 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32385 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32386 (riscv_option_override): Set riscv_slow_unaligned_access.
32387 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32389 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32391 * config/riscv/riscv.md: Unify indentation.
32393 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32398 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32400 (extendsi<mode>2): Add support for doing sign extension via
32401 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32402 don't have ISA 3.0 instructions.
32403 (extendsi<mode>2 splitter): Likewise.
32404 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32405 generate the normal insns since SImode can now go in vector
32406 registers. Disallow the special UNSPECs needed for previous
32407 machines to hide SImode being used. Add new insns
32408 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32409 (fix_trunc<mode>si2_stfiwx): Likewise.
32410 (fix_trunc<mode>si2_internal): Likewise.
32411 (fixuns_trunc<mode>si2): Likewise.
32412 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32413 (fctiw<u>z_<mode>_smallint): Likewise.
32414 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32415 of floating point to 32-bit integer from doing a direct move to
32416 the GPR registers to do a store.
32417 (fctiwz_<mode>): Break long line.
32419 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32421 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32422 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32423 (addr_list, addr_offset_valid_p): New.
32424 (split_address_groups): Check offset validity with above function.
32425 (gt-tree-ssa-loop-ivopts.h): Include header file.
32427 2017-05-05 Nathan Sidwell <nathan@acm.org>
32429 * config.gcc (arm*-*-*): Add missing 'fi'.
32431 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32433 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32434 included in -fopt-info does not matter.
32435 * doc/optinfo.texi (-fopt-info): Fix description of default
32436 behavour. Explicitly say order of options included in -fopt-info
32439 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32441 * config.gcc: Allow combinations of aprofile and rmprofile values for
32442 --with-multilib-list.
32443 * config/arm/t-multilib: New file.
32444 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32445 variables. Remove setting of ISA and floating-point ABI in
32446 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32447 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32448 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32449 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32451 * config/arm/t-rmprofile: Likewise except for the matches changes.
32452 * doc/install.texi (--with-multilib-list): Document the combination of
32453 aprofile and rmprofile values and warn about pitfalls in doing that.
32455 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32457 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32458 (movdi_aarch64): Likewise.
32460 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32462 PR tree-optimization/80632
32463 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32465 (build_arrays): Initialize it for virtual phis.
32466 (fix_phi_nodes): Use it for virtual phis.
32468 PR tree-optimization/80558
32469 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32470 [x, y] op z into [x op, y op z] for op & or | if conditions
32473 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32474 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32477 * config/arm/arm.md (use_literal_pool): Remove.
32478 (64-bit immediate split): No longer takes cost into consideration
32479 if arm_disable_literal_pool is enabled.
32480 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32481 used when arm_disable_literal_pool is enabled.
32482 (arm_max_const_double_inline_cost): Remove use of
32483 arm_disable_literal_pool.
32484 (push_minipool_fix): Add assert.
32485 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32486 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32487 (no_literal_pool_sf_immediate): New.
32489 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32491 PR tree-optimization/80613
32492 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32493 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32495 2017-05-05 Richard Biener <rguenther@suse.de>
32497 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32499 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32501 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32502 of this flag from insn conditions due to removal from r247495.
32504 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32506 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32508 (arm_early_store_addr_dep_ptr): Likewise.
32509 * config/arm/aarch-common-protos.h
32510 (arm_early_load_addr_dep_ptr): Add prototype.
32511 (arm_early_store_addr_dep_ptr): Likewise.
32512 * config/arm/cortex-a53.md: Add new bypasses.
32514 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32516 * tree.c (next_type_uid): Change type to unsigned.
32517 (type_hash_canon): Decrement back next_type_uid if
32518 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32519 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32522 2017-05-04 Martin Sebor <msebor@redhat.com>
32524 * builtins.c: Fix a trivial typo in a comment.
32526 PR middle-end/79234
32527 * builtins.c (check_sizes): Adjust to handle reading past the end.
32528 Avoid printing excessive upper bound of ranges. Use %E to print
32529 tree nodes instead of converting them to %wu.
32530 (expand_builtin_memchr): New function.
32531 (compute_dest_size): Rename...
32532 (compute_objsize): ...to this.
32533 (expand_builtin_memcpy): Adjust.
32534 (expand_builtin_mempcpy): Adjust.
32535 (expand_builtin_strcat): Adjust.
32536 (expand_builtin_strcpy): Adjust.
32537 (check_strncat_sizes): Adjust.
32538 (expand_builtin_strncat): Adjust.
32539 (expand_builtin_strncpy): Adjust and simplify.
32540 (expand_builtin_memset): Adjust.
32541 (expand_builtin_bzero): Adjust.
32542 (expand_builtin_memcmp): Adjust.
32543 (expand_builtin): Handle memcmp.
32544 (maybe_emit_chk_warning): Check strncat just once.
32546 2017-05-04 Martin Sebor <msebor@redhat.com>
32548 PR preprocessor/79214
32549 PR middle-end/79222
32550 PR middle-end/79223
32551 * builtins.c (check_sizes): Add inlining context and issue
32552 warnings even when -Wno-system-headers is set.
32553 (check_strncat_sizes): Same.
32554 (expand_builtin_strncat): Same.
32555 (expand_builtin_memmove): New function.
32556 (expand_builtin_stpncpy): Same.
32557 (expand_builtin): Handle memmove and stpncpy.
32559 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32561 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32562 which is not used any more.
32564 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32566 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32568 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32570 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32571 (cortexa53_tunings): Likewise.
32572 (cortexa57_tunings): Likewise.
32573 (cortexa72_tunings): Likewise.
32574 (cortexa73_tunings): Likewise.
32576 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32578 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32579 Set loop alignment to 8.
32581 2017-05-04 Martin Sebor <msebor@redhat.com>
32583 PR translation/80280
32584 * builtins.c (expand_builtin_object_size): Add missing quoting to
32585 %D and like directives.
32586 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32587 (hsa_type_for_tree_type): Same.
32588 (verify_function_arguments): Same.
32589 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32590 * varasm.c (get_section): Same.
32593 2017-05-04 Martin Sebor <msebor@redhat.com>
32595 PR translation/80280
32596 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32598 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32600 * config/aarch64/aarch64.c (generic_addrcost_table):
32601 Change HI/TI mode setting.
32603 2017-05-04 Martin Jambor <mjambor@suse.cz>
32605 PR tree-optimization/80622
32606 * tree-sra.c (comes_initialized_p): New function.
32607 (build_accesses_from_assign): Only set write lazily when
32608 comes_initialized_p is false.
32609 (analyze_access_subtree): Use comes_initialized_p.
32610 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32611 instead of testing for PARM_DECL.
32613 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32615 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32616 constraint on operand 0 to allow more general addressing modes.
32617 Adjust output template.
32618 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32620 * config/aarch64/aarch64-protos.h
32621 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32622 * config/aarch64/constraints.md (Dp): New address constraint.
32623 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32626 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32628 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32629 update use of estimate_ipcp_clone_size_and_time.
32630 (estimate_local_effects): Update use of
32631 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32632 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32633 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32634 Return nonspecialized time.
32636 2017-05-04 Richard Biener <rguenther@suse.de>
32638 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32639 for the last VUSE which def dominates the PHI. Directly call
32641 (get_continuation_for_phi_1): Remove.
32643 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32645 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32646 to explain the use of truncating division. Cap the number of
32647 iterations to the maximum given by nb_iterations_upper_bound,
32650 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32652 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32653 * configure: Regenerate.
32654 * config.in: Regenerate.
32655 * config/i386/driver-mingw32.c: new file.
32656 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32657 * config.host: Link driver-mingw32.o on MinGW host.
32658 * doc/install.texi: Document new --enable-mingw-wildcard configure
32661 2017-05-04 Marek Polacek <polacek@redhat.com>
32663 PR tree-optimization/80612
32664 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32666 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32667 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32669 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32670 (movt splitter): Likewise.
32671 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32672 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32673 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32674 block for Thumb-1 with MOVT.
32675 (thumb2_legitimate_address_p): Move code block ...
32676 (can_avoid_literal_pool_for_label_p): ... into this new function.
32677 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32679 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32680 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32681 "M-profile targets with the MOVT instruction".
32683 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32685 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32686 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32687 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32689 2017-05-04 Martin Liska <mliska@suse.cz>
32691 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32692 variable cond_code.
32694 2017-05-04 Richard Biener <rguenther@suse.de>
32696 * tree.c (array_at_struct_end_p): Handle arrays at struct
32697 end with flexarrays more conservatively. Refactor and treat
32698 arrays of arrays or aggregates more strict. Fix
32699 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32700 * tree.c (array_at_struct_end_p): Adjust prototype.
32701 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32702 * gimple-fold.c (get_range_strlen): Likewise.
32703 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32705 2017-05-04 Richard Biener <rguenther@suse.de>
32707 PR tree-optimization/31130
32708 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32710 (supports_overflow_infinity): Likewise.
32711 (is_negative_overflow_infinity): Likewise.
32712 (is_positive_overflow_infinity): Likewise.
32713 (is_overflow_infinity): Likewise.
32714 (stmt_overflow_infinity): Likewise.
32715 (overflow_infinity_range_p): Likewise.
32716 (usable_range_p): Remove as always returning true.
32717 (make_overflow_infinity): Remove.
32718 (negative_overflow_infinity): Likewise.
32719 (positive_overflow_infinity): Likewise.
32720 (avoid_overflow_infinity): Likewise.
32721 (set_value_range): Adjust accordingly.
32722 (set_value_range_to_nonnegative): Likewise, remove now unused
32723 overflow_infinity arg.
32724 (vrp_operand_equal_p): Adjust.
32725 (update_value_range): Likewise.
32726 (range_int_cst_singleton_p): Likewise.
32727 (operand_less_p): Likewise.
32728 (compare_values_warnv): Likewise.
32729 (extract_range_for_var_from_comparison_expr): Likewise.
32730 (vrp_int_const_binop): Likewise.
32731 (zero_nonzero_bits_from_vr): Likewise.
32732 (extract_range_from_multiplicative_op_1): Likewise.
32733 (extract_range_from_binary_expr_1): Likewise.
32734 (extract_range_from_unary_expr): Likewise.
32735 (extract_range_from_comparison): Likewise.
32736 (extract_range_basic): Likewise.
32737 (adjust_range_with_scev): Likewise.
32738 (compare_ranges): Likewise.
32739 (compare_range_with_value): Likewise.
32740 (dump_value_range): Likewise.
32741 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32743 (simplify_cond_using_ranges): Adjust.
32745 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32747 * brig-builtins.def: Added a builtin for class_f64.
32748 * builtin-types.def: Added a builtin type needed by class_f64.
32750 2017-05-03 Jason Merrill <jason@redhat.com>
32752 * timevar.def: Add TV_CONSTEXPR.
32754 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32756 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32758 2017-05-03 Martin Jambor <mjambor@suse.cz>
32760 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32761 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32762 * ipa-cp.c (ipcp_propagate_stage): Do not call
32763 ipa_update_after_lto_read.
32764 * ipa-inline.c (ipa_inline): Likewise.
32766 2017-05-03 Martin Jambor <mjambor@suse.cz>
32768 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32769 tag. Added a default constructor and a destructor.
32770 (ipa_edge_args_sum_t): New class;
32771 (ipa_edge_args_sum): Declare.
32772 (ipa_edge_args_vector): Remove declaration.
32773 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32774 (ipa_free_edge_args_substructures): Remove declaration.
32775 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32776 (ipa_edge_args_info_available_for_edge_p): Likewise.
32777 * ipa-prop.c (ipa_edge_args_vector): Removed.
32778 (edge_removal_hook_holder): Likewise.
32779 (edge_duplication_hook_holder): Likewise.
32780 (ipa_edge_args_sum): New variable.
32781 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32782 ipa_edge_args_vector.
32783 (ipa_free_edge_args_substructures): Likewise.
32784 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32785 ipa_edge_args_vector.
32786 (ipa_edge_removal_hook): Turned into method
32787 ipa_edge_args_sum_t::remove.
32788 (ipa_edge_duplication_hook): Turned into method
32789 ipa_edge_args_sum_t::duplicate.
32790 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
32791 registering edge hooks.
32792 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
32793 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
32794 ipa_edge_args_sum instead of ipa_edge_args_vector.
32795 * ipa-profile.c (ipa_profile): Likewise.
32797 2017-05-03 Martin Jambor <mjambor@suse.cz>
32799 * symbol-summary.h (function_summary): New method exists.
32800 (function_summary::symtab_removal): Deallocate through release.
32801 (call_summary): New class.
32802 (gt_ggc_mx): New overload.
32803 (gt_pch_nx): Likewise.
32804 (gt_pch_nx): Likewise.
32806 2017-05-03 Jeff Law <law@redhat.com>
32808 PR tree-optimization/78496
32809 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
32810 from simplify_cond_using_ranges. Split off code to walk
32811 backwards through casts into ...
32812 (simplify_cond_using_ranges_2): New function.
32813 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
32814 (execute_vrp): After identifying jump threads, call
32815 simplify_cond_using_ranges_2.
32817 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32820 * ipa-inline.h (inline_summary): Add ctor.
32821 (create_ggc): Do not use ggc_cleared_alloc.
32823 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32824 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32826 * gcc.c (handle_braces): Support escaping in switch matching
32828 * doc/invoke.texi (Spec Files): Document it.
32829 Remove superfluous @code markup in items.
32831 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32833 * diagnostic-show-locus.c (struct column_range): New struct.
32834 (get_affected_columns): New function.
32835 (get_printed_columns): New function.
32836 (struct correction): New struct.
32837 (correction::ensure_capacity): New function.
32838 (correction::ensure_terminated): New function.
32839 (struct line_corrections): New struct.
32840 (line_corrections::~line_corrections): New dtor.
32841 (line_corrections::add_hint): New function.
32842 (layout::print_trailing_fixits): Reimplement in terms of the new
32844 (selftest::test_overlapped_fixit_printing): New function.
32845 (selftest::diagnostic_show_locus_c_tests): Call it.
32847 2017-05-03 Nathan Sidwell <nathan@acm.org>
32849 Canonicalize canonical type hashing
32850 * tree.h (type_hash_canon_hash): Declare.
32851 * tree.c (type_hash_list, attribute_hash_list): Move into
32852 type_hash_canon_hash.
32853 (build_type_attribute_qual_variant): Break out hash code calc into
32854 type_hash_canon_hash.
32855 (type_hash_canon_hash): New. Generic type hash computation.
32856 (build_range_type_1, build_array_type_1, build_function_type,
32857 build_method_type_directly, build_offset_type, build_complex_type,
32858 make_vector_type): Call it.
32860 2017-05-03 Richard Biener <rguenther@suse.de>
32862 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
32863 When all DRs have unknown misaligned do not always peel
32864 when there is a store but apply the same costing model as if
32865 there were only loads.
32867 2017-05-03 Richard Biener <rguenther@suse.de>
32870 PR tree-optimization/80492
32871 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32872 compare_base_decls returning dont-know properly.
32874 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32876 * config/arm/iterators.md (CCSI): New mode iterator.
32877 (arch): New mode attribute.
32878 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32879 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32880 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32881 code iterator for success result mode.
32882 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32883 the corresponding new insn generators.
32885 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32888 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32889 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32891 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32893 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32894 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32895 (DDR_A): Wrap DDR argument in brackets.
32896 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32897 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32898 (DDR_REVERSED_P): Likewise.
32900 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32902 PR tree-optimization/79472
32903 * tree-switch-conversion.c (struct switch_conv_info): Add
32904 contiguous_range and default_case_nonstandard fields.
32905 (collect_switch_conv_info): Compute contiguous_range and
32906 default_case_nonstandard fields, don't clear final_bb if
32907 contiguous_range and only the default case doesn't have the required
32909 (check_all_empty_except_final): Set default_case_nonstandard instead
32910 of failing if contiguous_range and the default case doesn't have empty
32912 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32913 and only the default case doesn't have the required constants. Skip
32915 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32916 if default_case_nonstandard.
32917 (build_constructors): Build constant 1 just once. Assert that default
32918 values aren't inserted in between cases if contiguous_range. Skip
32920 (build_arrays): Skip virtual phis.
32921 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32922 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32923 Handle virtual phis.
32924 (gen_inbound_check): Handle default_case_nonstandard case.
32925 (process_switch): Adjust check_final_bb caller. Call
32926 gather_default_values with the first non-default case instead of
32927 default case if default_case_nonstandard.
32929 2017-05-02 Nathan Sidwell <nathan@acm.org>
32931 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32932 check. Fix formatting.
32934 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32936 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32937 errors when comparing specialized and unspecialized times.
32939 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32941 * diagnostic-show-locus.c
32942 (layout::should_print_annotation_line_p): Make private.
32943 (layout::print_annotation_line): Make private.
32944 (layout::annotation_line_showed_range_p): Make private.
32945 (layout::show_ruler): Make private.
32946 (layout::print_source_line): Make private. Pass in line and
32947 line_width, rather than calling location_get_source_line. Drop
32949 (layout::print_leading_fixits): New method.
32950 (layout::print_any_fixits): Rename to...
32951 (layout::print_trailing_fixits): ...this, and make private.
32952 Don't print newline fixits.
32953 (diagnostic_show_locus): Move logic for printing one row into...
32954 (layout::print_line): ...this new function. Move the
32955 location_get_source_line call and error-handling from
32956 print_source_line to here. Call print_leading_fixits, and rename
32957 print_any_fixits to print_trailing_fixits.
32958 (selftest::test_fixit_insert_containing_newline): Update now that
32959 newlines are partially supported.
32960 (selftest::test_fixit_insert_containing_newline_2): New test.
32961 (selftest::test_fixit_replace_containing_newline): Update comments.
32962 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32963 * edit-context.c (class added_line): New class.
32964 (class edited_line): Describe newline handling in comment.
32965 (edited_line::actually_edited_p): New method.
32966 (edited_line::print_content): Delete redundant decl.
32967 (edited_line::m_predecessors): New field.
32968 (edited_file::print_content): Call edited_line::print_content.
32969 (edited_file::print_diff): Update to support newlines.
32970 (edited_file::print_diff_hunk): Likewise.
32971 (edited_file::print_run_of_changed_lines): New function.
32972 (edited_file::print_diff_line): Convert to...
32973 (print_diff_line): ...this.
32974 (edited_file::get_effective_line_count): New function.
32975 (edited_line::edited_line): Initialize new field m_predecessors.
32976 (edited_line::~edited_line): Clean up m_predecessors.
32977 (edited_line::apply_fixit): Handle newlines.
32978 (edited_line::get_effective_line_count): New function.
32979 (edited_line::print_content): New function.
32980 (edited_line::print_diff_lines): New function.
32981 (selftest::test_applying_fixits_insert_containing_newline): New
32983 (selftest::test_applying_fixits_replace_containing_newline): New
32985 (selftest::insert_line): New function.
32986 (selftest::test_applying_fixits_multiple_lines): Add example of
32988 (selftest::edit_context_c_tests): Call the new tests.
32990 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32992 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32993 parameter cand. Update dump information.
32994 (get_computation_cost): Update uses.
32996 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32998 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32999 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33000 (get_computation_at, rewrite_use_address): Update use of
33001 get_computation_aff.
33003 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33005 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33006 (get_computation): Delete.
33007 (get_computation_cost): Implement like get_computation_cost_at.
33008 Use get_computation_at.
33009 (get_computation_cost_at): Delete.
33010 (rewrite_use_nonlinear_expr): Use get_computation_at.
33011 (rewrite_use_compare, remove_unused_ivs): Ditto.
33013 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33015 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33017 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33019 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33020 (ivopts_global_cost_for_size): Rename parameter and update uses.
33021 (iv_ca_recount_cost): Update uses.
33022 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33023 candidates seperately in n_invs and n_cands.
33024 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33026 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33028 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33029 (find_inv_vars_cb): New.
33030 (find_depends): Renamed to ...
33031 (find_inv_vars): ... this.
33032 (add_candidate_1, force_var_cost): Call find_inv_vars.
33033 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33035 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33037 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33038 inv_vars. Add inv_exprs.
33039 (struct iv_cand): Rename depends_on to inv_vars.
33040 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33041 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33042 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33043 (dump_cand): Dump inv_vars.
33044 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33045 (record_invariant, find_depends, add_candidate_1): Ditto.
33046 (set_group_iv_cost, force_var_cost): Ditto.
33047 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33048 (get_computation_cost_at, get_computation_cost): Ditto.
33049 (determine_group_iv_cost_generic): Ditto.
33050 (determine_group_iv_cost_address): Ditto.
33051 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33052 (determine_group_iv_costs): Ditto.
33053 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33054 (iv_ca_set_remove_invariants): Renamed to ...
33055 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33056 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33057 (iv_ca_set_add_invariants): Renamed to ...
33058 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33059 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33060 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33061 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33062 (create_new_ivs): Remove useless dump.
33064 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33066 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33068 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33069 (iv_ca_set_no_cp, create_new_iv): Ditto.
33071 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33073 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33075 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33077 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33078 function tree_check2.
33080 2017-05-02 Martin Liska <mliska@suse.cz>
33082 * doc/gcov.texi: Add missing preposition.
33083 * gcov.c (function_info::function_info): Properly fill up
33084 all member variables.
33086 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33088 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33090 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33092 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33094 2017-05-02 Martin Liska <mliska@suse.cz>
33097 * lto-streamer-in.c (lto_read_tree_1): Remove
33098 LTO_STREAMER_DEBUG.
33099 * lto-streamer.c (struct tree_hash_entry): Likewise.
33100 (struct tree_entry_hasher): Likewise.
33101 (tree_entry_hasher::hash): Likewise.
33102 (tree_entry_hasher::equal): Likewise.
33103 (lto_streamer_init): Likewise.
33104 (lto_orig_address_map): Likewise.
33105 (lto_orig_address_get): Likewise.
33106 (lto_orig_address_remove): Likewise.
33107 * lto-streamer.h: Likewise.
33108 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33109 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33111 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33113 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33114 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33115 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33116 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33117 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33118 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33119 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33120 (mm_maskz_sub_ss): New intrinsics.
33121 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33122 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33123 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33124 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33125 (__builtin_ia32_subss_mask_round): New builtins.
33126 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33127 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33128 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33130 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33131 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33133 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33136 2017-05-02 Martin Jambor <mjambor@suse.cz>
33138 PR tree-optimization/78687
33139 * tree-sra.c (access): New field parent.
33140 (process_subtree_disqualification): New function.
33141 (disqualify_candidate): Call it.
33142 (build_accesses_from_assign): Reset write flag if creating an
33144 (build_access_subtree): Fill in parent field and also prpagate
33145 down grp_write flag.
33146 (create_artificial_child_access): New parameter set_grp_write, set
33147 grp_write to its value.
33148 (propagate_subaccesses_across_link): Also propagate grp_write flag
33150 (propagate_all_subaccesses): Push the closest parent back to work
33151 queue if add_access_to_work_queue returned true.
33153 2017-05-02 Richard Biener <rguenther@suse.de>
33155 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33156 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33157 -fstrict-overflow documentation.
33158 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33159 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33160 flag_strict_overflow.
33161 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33162 * lto-opts.c (lto_write_options): Do not stream it.
33163 * lto-wrapper.c (merge_and_complain): Do not handle it.
33164 * opts.c (default_options_table): Do not set -fstrict-overflow.
33165 (finish_options): Likewise do not clear it when sanitizing.
33166 * simplify-rtx.c (simplify_const_relational_operation): Do not
33167 test flag_strict_overflow.
33169 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33171 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33172 using enabled attribute.
33173 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33174 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33175 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33176 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33177 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33178 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33179 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33180 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33181 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33182 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33184 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33186 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33188 2017-05-02 Richard Biener <rguenther@suse.de>
33190 PR tree-optimization/80591
33192 2017-04-10 Richard Biener <rguenther@suse.de>
33194 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33197 2017-05-02 Richard Biener <rguenther@suse.de>
33199 PR tree-optimization/80549
33200 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33201 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33202 headers if they do not have a preheader.
33204 2017-05-02 Martin Liska <mliska@suse.cz>
33207 * common.opt: Fix typo.
33208 * doc/invoke.texi: Likewise.
33210 2017-05-01 Jan Beulich <jbeulich@suse.com>
33212 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33213 swapping, add (x,x,m,x,n) alternative.
33215 2017-05-01 Nathan Sidwell <nathan@acm.org>
33217 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33218 unnecessary unadjusted_alignment check.
33220 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33223 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33225 * gimplify.c (gimplify_cilk_detach): New function.
33226 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33227 * tree-core.h: Document EXPR_CILK_SPAWN.
33228 * tree.h (EXPR_CILK_SPAWN): Define.
33230 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33232 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33233 to use new fixit_hint representation, using the "replace" logic.
33234 (get_line_span_for_fixit_hint): Likewise.
33235 (layout::print_any_fixits): Likewise.
33236 (selftest::test_one_liner_many_fixits): Rename to...
33237 (selftest::test_one_liner_many_fixits_1): ...this, and update
33238 comment and expected output to reflect that the multiple fix-it
33239 hints are now consolidated into one insertion.
33240 (selftest::test_one_liner_many_fixits_2): New test.
33241 (selftest::test_diagnostic_show_locus_one_liner): Update for
33243 (selftest::test_fixit_consolidation): Update for fix-it API
33245 * diagnostic.c (print_parseable_fixits): Likewise.
33246 * edit-context.c (edited_line::m_line_events): Convert from
33247 auto_vec <line_event *> to auto_vec <line_event>.
33248 (class line_event): Convert from abstract base class to a concrete
33249 class, taking over the role of replace_event.
33250 (class insert_event): Delete.
33251 (class replace_event): Rename to class line_event. Convert to
33253 (edit_context::add_fixits): Reimplement.
33254 (edit_context::apply_insert): Delete.
33255 (edit_context::apply_replace): Rename to...
33256 (edit_context::apply_fixit): ...this. Convert to half-open range.
33257 (edited_file::apply_insert): Delete.
33258 (edited_file::apply_replace): Rename to...
33259 (edited_file::apply_fixit): ...this.
33260 (edited_line::~edited_line): Drop deletion of events.
33261 (edited_line::apply_insert): Delete.
33262 (edited_line::apply_replace): Rename to...
33263 (edited_line::apply_fixit): ...this. Convert to half-open range.
33264 Update for change to type of m_line_events.
33265 * edit-context.h (edit_context::apply_insert): Delete.
33266 (edit_context::apply_replace): Rename to...
33267 (edit_context::apply_fixit): ...this.
33269 2017-05-01 Martin Sebor <msebor@redhat.com>
33271 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33274 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33277 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33278 __get_cpuid_max returns 0.
33279 (__get_cpuid_count): Ditto.
33281 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33283 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33284 replacement expression is another instance of one of its arguments.
33286 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33289 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33290 check for stack push/pop autoinc.
33291 * config/i386/i386.c (ix86_agi_dependent): Return false
33292 if the only reason why modified_in_p returned true is that
33293 addr is SP based and set_insn is a push or pop.
33295 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33297 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33300 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33303 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33304 (account_size_time): Use two predicates - exec_pred and
33306 (evaluate_conditions_for_known_args): Compute both clause and
33308 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33309 (inline_summary_t::duplicate): Update.
33310 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33312 (compute_inline_parameters): Likewise.
33313 (estimate_edge_size_and_time): Update caluclation of time.
33314 (estimate_node_size_and_time): Compute both time and nonspecialized
33316 (estimate_ipcp_clone_size_and_time): Update.
33317 (inline_merge_summary): Update.
33318 (do_estimate_edge_time): Update.
33319 (do_estimate_edge_size): Update.
33320 (do_estimate_edge_hints): Update.
33321 (inline_read_section, inline_write_summary): Stream both new predicates.
33322 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33324 (compute_inlined_call_time): Cleanup.
33325 (big_speedup_p): Update.
33326 (edge_badness): Update.
33327 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33328 (size_time_entry): Replace predicate by exec_predicate and
33329 nonconst_predicate.
33330 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33331 (estimate_edge_time): Return also nonspec_time.
33332 (reset_edge_growth_cache): Update.
33334 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33336 PR rtl-optimization/80491
33337 * ifcvt.c (noce_process_if_block): When looking for x setter
33338 with missing else_bb, don't check only the insn right before
33339 cond_earliest, but look for the last insn that x is modified in
33340 within the same bb.
33342 PR rtl-optimization/80491
33343 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33345 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33347 PR tree-optimization/80487
33348 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33350 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33352 PR tree-optimization/79697
33353 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33354 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33355 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33357 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33358 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33360 2017-04-28 Martin Sebor <msebor@redhat.com>
33362 PR tree-optimization/80523
33363 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33364 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33366 (maybe_warn, format_directive, parse_directive): Use new functions.
33367 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33369 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33371 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33373 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33375 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33376 target_header_dir): Set correctly.
33377 * configure: Regenerated.
33378 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33379 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33380 instead of SYSTEM_HEADER_DIR.
33382 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33384 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33385 (estimate_local_effects): Likewise.
33386 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33387 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33388 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33389 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33390 do_estimate_edge_time, estimate_edge_time): Likewise.
33391 * ipa-inline-analysis.c (estimate_node_size_and_time,
33392 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33393 (estimate_time_after_inlining): Remove.
33395 2017-04-28 Martin Liska <mliska@suse.cz>
33397 * doc/gcov.texi: Enhance documentation of gcov.
33399 2017-04-28 Martin Liska <mliska@suse.cz>
33401 * doc/gcov.texi: Sort options in alphabetic order.
33402 * doc/gcov-dump.texi: Likewise.
33403 * doc/gcov-tool.texi: Likewise.
33404 * gcov.c (print_usage): Likewise.
33405 * gcov-dump.c (print_usage): Likewise.
33406 * gcov-tool.c (print_merge_usage_message): Likewise.
33407 (print_rewrite_usage_message): Likewise.
33408 (print_overlap_usage_message): Likewise.
33410 2017-04-28 Martin Liska <mliska@suse.cz>
33412 PR gcov-profile/53915
33413 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33415 2017-04-28 Martin Liska <mliska@suse.cz>
33417 PR gcov-profile/79891
33418 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33419 is marked by compiler as living on a line.
33420 (get_cycles_count): Remove usage of the union.
33421 (output_intermediate_file): Likewise.
33422 (find_source): Fix GNU coding style.
33423 (accumulate_line_counts): Remove old non-all block mode.
33424 (output_lines): Remove usage of the union.
33425 * profile.c (output_location): Include all BBs, even if
33426 belonging to a same line (and file) as a previous BB.
33428 2017-04-28 Martin Liska <mliska@suse.cz>
33430 * gcov.c (process_args): Handle new argument 'w'.
33431 (read_graph_file): Assign ID to BBs.
33432 (output_branch_count): Display BB # if verbose flag is set.
33433 (output_lines): Likewise for arcs.
33434 (print_usage): Add '--verbose' option help.
33435 * doc/gcov.texi: Document --verbose (-w) option.
33437 2017-04-28 Martin Liska <mliska@suse.cz>
33439 * gcov.c (struct block_location_info): New struct.
33440 (process_file): Fill up the new structure.
33441 (read_graph_file): Replace usage of encoding by the newly added
33443 (add_line_counts): Likewise.
33444 (accumulate_line_counts): Remove usage of the union.
33445 (function_info::function_info): New function.
33446 (function_info::~function_info): Likewise.
33447 (process_file): Call delete instead of release_function.
33448 (release_function): Release the function.
33449 (release_structures): Call delete instead of release_function.
33450 (solve_flow_graph): Replace usage of num_blocks.
33451 (find_exception_blocks): Likewise.
33452 (output_lines): Fix GNU coding style.
33454 2017-04-28 Martin Liska <mliska@suse.cz>
33457 * coverage.c (coverage_remove_note_file): New function.
33458 * coverage.h: Declare the function.
33459 * toplev.c (finalize): Clean if an error has been seen.
33461 2017-04-28 Martin Liska <mliska@suse.cz>
33463 PR gcov-profile/80031
33464 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33465 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33466 * gcov.c (read_graph_file): Read just number of blocks.
33467 * profile.c (branch_prob): Do not stream 0 flags per a basic
33470 2017-04-28 Martin Liska <mliska@suse.cz>
33472 * gcov-dump.c (tag_*): Add new argument to declarations.
33473 (dump_gcov_file): Likewise.
33474 (tag_blocks): Add and use new argument depth.
33475 (tag_arcs): Likewise.
33476 (tag_lines): Likewise.
33477 (tag_counters): Likewise.
33478 (tag_summary): Likewise.
33479 (dump_working_sets): Use depth to do a proper indentation.
33481 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33484 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33485 * symtab.c (symtab_node::debug_symtab): Move definition here.
33487 2017-04-28 Richard Biener <rguenther@suse.de>
33489 * lto-streamer.h (LTO_major_version): Bump to 7.
33491 2017-04-28 Richard Biener <rguenther@suse.de>
33493 * tree-vrp.c (assert_info): New struct.
33494 (add_assert_info): New helper.
33495 (register_edge_assert_for_2): Refactor to add asserts to a vector
33497 (register_edge_assert_for_1): Likewise.
33498 (register_edge_assert_for): Likewise.
33499 (finish_register_edge_assert_for): New helper actually registering
33500 asserts where live on edge.
33501 (find_conditional_asserts): Adjust.
33502 (find_switch_asserts): Likewise.
33503 (evrp_dom_walker::try_find_new_range): Generalize.
33504 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33506 2017-04-27 Marek Polacek <polacek@redhat.com>
33509 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33510 arg10 and arg11 to itype.
33512 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33514 * doc/extend.texi (Object Size Checking): Improve grammar.
33516 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33519 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33520 that the logic for permitting reciprocal estimates matches that
33523 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33526 * tree.c (type_cache_hasher::equal): Only compare
33527 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33528 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33529 non-aggregate element types.
33530 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33531 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33533 2017-04-27 Richard Biener <rguenther@suse.de>
33535 PR middle-end/80533
33536 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33537 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33538 keeping a reference to a trailing array.
33540 2017-04-27 Richard Biener <rguenther@suse.de>
33542 PR middle-end/80539
33543 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33544 being in loop-closed SSA form conservatively.
33545 (chrec_fold_multiply_poly_poly): Likewise.
33547 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33549 PR middle-end/79665
33550 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33551 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33553 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33556 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33557 (aarch64_function_arg_alignment): Return unsigned int again, but still
33558 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33559 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33560 Don't emit -Wpsabi note.
33561 (aarch64_function_arg_boundary): Likewise.
33562 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33565 2017-04-26 Nathan Sidwell <nathan@acm.org>
33567 * tree.h (crc32_unsigned_n): Declare.
33568 (crc32_unsigned, crc32_unsigned): Make inline.
33569 * tree.c (crc32_unsigned_bits): Replace with ...
33570 (crc32_unsigned_n): ... this.
33571 (crc32_unsigned, crc32_byte): Remove.
33572 (crc32_string): Remove unnecessary braces.
33574 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33576 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33577 * ipa-inline-analysis.c (MAX_TIME): Remove.
33578 (account_size_time): Use sreal for time.
33579 (dump_inline_summary): Update.
33580 (estimate_function_body_sizes): Update.
33581 (estimate_edge_size_and_time): Update.
33582 (estimate_calls_size_and_time): Update.
33583 (estimate_node_size_and_time): Update.
33584 (inline_merge_summary): Update.
33585 (inline_update_overall_summary): Update.
33586 (estimate_time_after_inlining): Update.
33587 (inline_read_section): Update.
33588 (inline_write_summary): Update.
33589 * ipa-inline.c (compute_uninlined_call_time): Update.
33590 (compute_inlined_call_time): Update.
33591 (recursive_inlining): Update.
33592 (inline_small_functions): Update.
33593 (dump_overall_stats): Update.
33594 * ipa-inline.h: Include sreal.h.
33595 (size_time_entry): Turn time to sreal.
33596 (inline_summary): Turn self_time nad time to sreal.
33598 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33600 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33602 (sreal::stream_out, sreal::stream_in): New.
33603 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33605 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33607 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33610 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33613 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33615 (dimode_scalar_chain::compute_convert_gain): Ditto.
33616 (dimode_scalar_chain::make_vector_copies): Ditto.
33617 (dimode_scalar_chain::convert_reg): Ditto.
33618 (dimode_scalar_chain::convert_insn): Ditto.
33619 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33620 (VI248_AVX512BW_1): New mode iterator.
33621 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33622 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33625 2017-04-25 Martin Sebor <msebor@redhat.com>
33627 PR tree-optimization/80497
33628 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33629 constants are representable in HOST_WIDE_INT.
33630 (parse_directive): Ditto.
33632 2017-04-25 Martin Sebor <msebor@redhat.com>
33635 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33636 (new_zero_array): Adjust signature.
33637 (dom_info::dom_init): Used unsigned rather that size_t.
33638 (dom_info::dom_info): Same.
33640 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33641 Jakub Jelinek <jakub@redhat.com>
33644 * config/arm/arm.c: Include gimple.h.
33645 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33646 returns negative, increment ncrn only if it returned positive.
33647 (arm_needs_doubleword_align): Return int instead of bool,
33648 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33649 members, but if there is any such non-FIELD_DECL
33650 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33651 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33652 returns negative, increment nregs only if it returned positive.
33653 (arm_setup_incoming_varargs): Likewise.
33654 (arm_function_arg_boundary): Emit -Wpsabi note if
33655 arm_needs_doubleword_align returns negative, return
33656 DOUBLEWORD_ALIGNMENT only if it returned positive.
33658 2017-04-25 Marek Polacek <polacek@redhat.com>
33661 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33662 first argument to type.
33664 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33667 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33668 type checks to test for compatibility instead of equality.
33670 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33671 Jakub Jelinek <jakub@redhat.com>
33674 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33676 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33677 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33678 the alignment computation, but return their maximum in warn_alignment.
33679 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33680 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33682 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33683 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33686 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33688 * config/arc/simdext.md (dmpyh): Fix typo.
33690 2017-04-25 Richard Biener <rguenther@suse.de>
33692 PR tree-optimization/80492
33693 * alias.c (compare_base_decls): Handle registers with asm
33694 specification conservatively.
33695 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33696 compare_base_decls returning dont-know properly.
33698 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33700 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33701 (legitimate_offset_address_p): New function.
33702 (arc_legitimate_address_p): Use above function.
33704 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33706 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33708 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33710 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33711 ACCH registers whenever they are available.
33713 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33715 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33716 double regs fix when not used.
33718 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33720 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33722 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33723 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33725 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33727 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33728 when emitting short ADD instructions.
33730 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33732 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33734 (cmpsi_cc_c_insn): Likewise.
33735 (cbranchsi4_scratch): Compute proper instruction length using
33736 compact_hreg_operand.
33737 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33739 2017-04-25 Richard Biener <rguenther@suse.de>
33741 PR middle-end/80509
33742 * passes.c (pass_manager::pass_manager): Initialize
33743 m_name_to_pass_map.
33745 2017-04-25 Richard Biener <rguenther@suse.de>
33747 PR tree-optimization/79201
33748 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33750 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33753 * config/s390/vector.md: Split MEM->GPR vector moves for
33754 non-s_operand addresses.
33756 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33759 * config/s390/predicates.md (reload_const_wide_int_operand): New
33761 * config/s390/s390.md ("movti"): Remove d/P alternative.
33762 ("movti_bigconst"): New pattern definition.
33764 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33767 * s390-protos.h (s390_expand_cs_hqi): Removed.
33768 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33769 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33770 modes as well as CCZ1mode and CCZmode.
33771 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33772 signature of s390_emit_compare_and_swap.
33773 (s390_expand_cs_hqi): Likewise, make static.
33774 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33775 compare-and-swap, in some cases.
33776 (s390_expand_cs): Wrapper function.
33777 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33779 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33780 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33781 patterns for small and large integers. Forbid symref memory operands.
33782 Move expander to s390.c. Require cc register.
33783 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33784 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33785 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33786 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33787 symref memory operands. Remove CC mode and call s390_match_ccmode
33789 ("atomic_exchange<mode>"): Allow and implement all integer modes.
33791 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33793 * config/s390/s390.md (define_peephole2): New peephole to help
33794 combining the load-and-test pattern with volatile memory.
33796 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33798 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
33799 with CCZmode for TARGET_Z196.
33801 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33803 PR rtl-optimization/80501
33804 * combine.c (make_compound_operation_int): Set subreg_code to SET
33805 even for AND with mask of the sign bit of mode.
33807 PR rtl-optimization/80500
33808 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
33809 sum's initial value.
33811 2017-04-25 Julian Brown <julian@codesourcery.com>
33812 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33814 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
33816 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33818 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33820 2017-04-25 Julian Brown <julian@codesourcery.com>
33821 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33823 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33824 (thunderx2t99_sha): New Reservation.
33826 2017-04-25 Julian Brown <julian@codesourcery.com>
33827 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33829 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33830 type for 1-element load.
33832 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33834 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33836 2017-04-24 Martin Jambor <mjambor@suse.cz>
33838 PR tree-optimization/80293
33839 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33840 char arrays not totally scalarizable if it is false.
33841 (analyze_all_variable_accesses): Pass correct value in the new
33842 parameter. Add a statistics counter.
33844 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
33846 PR middle-end/79931
33847 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
33849 2017-04-24 Richard Biener <rguenther@suse.de>
33851 PR tree-optimization/80494
33852 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33853 out for complex types.
33855 2017-04-24 Richard Biener <rguenther@suse.de>
33857 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
33858 * tree-ssa-sccvn.c (print_scc): Print SCC size.
33859 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
33860 (DFS): Adjust and never fail.
33861 (sccvn_dom_walker::fail): Remove.
33862 (sccvn_dom_walker::before_dom_children): Adjust.
33863 (run_scc_vn): Likewise and never fail.
33864 * tree-ssa-pre.c (pass_pre::execute): Adjust.
33865 (pass_fre::execute): Likewise.
33867 2017-04-24 Richard Biener <rguenther@suse.de>
33869 PR tree-optimization/79725
33870 * tree-ssa-sink.c (statement_sink_location): Return whether
33871 failure reason was zero uses. Move that check later.
33872 (sink_code_in_bb): Deal with zero uses by removing the stmt
33875 2017-04-24 Richard Biener <rguenther@suse.de>
33878 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33879 pointer-based references.
33881 2017-04-24 Richard Biener <rguenther@suse.de>
33884 * pass_manager.h (pass_manager::operator new): Remove.
33885 (pass_manager::operator delete): Likewise.
33886 * passes.c (pass_manager::operator new): Remove.
33887 (pass_manager::operator delete): Likewise.
33888 (pass_manager::pass_manager): Zero individual pass members.
33890 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33893 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33894 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33895 Check "XEXP (src, 1)" operand here.
33896 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33897 Check "XEXP (src, 1)" operand here.
33898 (dimode_scalar_chain::make_vector_copies): Detect count register
33899 of a shift instruction. Zero extend count register from QImode
33900 to DImode to satisfy vector shift pattern count operand predicate.
33901 Substitute vector shift count operand with a DImode copy.
33902 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33905 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33907 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33908 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33909 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33910 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33911 (UNSPEC_NOREX_MEM): Remove definition.
33913 2017-04-21 Richard Biener <rguenther@suse.de>
33915 PR tree-optimization/79547
33916 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33917 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33918 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33919 without any constraints.
33921 2017-04-21 Richard Biener <rguenther@suse.de>
33923 PR tree-optimization/78847
33924 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33926 2017-04-21 Richard Biener <rguenther@suse.de>
33928 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33929 (build_distinct_type_copy): Likewise.
33930 (build_variant_type_copy): Likewise.
33931 * tree.c (build_qualified_type): Pass down mem-stat info.
33932 (build_distinct_type_copy): Likewise.
33933 (build_variant_type_copy): Likewise.
33935 2017-04-21 Richard Biener <rguenther@suse.de>
33937 PR tree-optimization/80237
33938 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33940 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33941 for a simplified result.
33943 2016-04-21 Richard Biener <rguenther@suse.de>
33945 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33946 sth as strict as a simple_iv but a chrec without symbols and an
33947 operand defined in the loop we are peeling (and not some subloop).
33948 (propagate_constants_for_unrolling): Propagate all constants.
33950 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33953 * config/i386/i386.c (print_reg): Remove assert for disalowed
33954 regno values, call output_operand_lossage instead.
33956 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33959 * config/i386/constraints.md (Yc): New register constraint.
33960 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33961 Use Yc constraint for alternative 2 of operand 0. Remove
33962 preferred_for_speed attribute.
33964 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33966 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33967 lastprivate clauses in SIMT case.
33969 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33971 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33973 2017-04-20 Richard Biener <rguenther@suse.de>
33975 PR tree-optimization/57796
33976 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33977 as N scalar stores.
33978 (vect_model_load_cost): Cost gathers as N scalar loads.
33980 2017-04-20 Richard Biener <rguenther@suse.de>
33982 * ggc-page.c (ggc_allocated_p): Rename to ...
33983 (safe_lookup_page_table_entry): ... this and return the lookup
33985 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33987 2017-04-20 Richard Biener <rguenther@suse.de>
33989 PR tree-optimization/80453
33990 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33991 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33992 from the conditions.
33993 (vn_phi_eq): Pass them down.
33994 (vn_phi_lookup): Record them.
33995 (vn_phi_insert): Likewise.
33997 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33999 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34000 uninitialized variable warning to avoid buffer overrun.
34002 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34005 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34006 is suppressed for '{ 0 }' in C.
34008 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34010 * BASE-VER: Set to 8.0.0.
34012 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34014 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34015 priority .init_array and .fini_array section with SECTION_NOTYPE
34018 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34020 PR middle-end/80423
34021 * tree.h (build_array_type): Add typeless_storage default argument.
34022 * tree.c (type_cache_hasher::equal): Also compare
34023 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34024 (build_array_type): Add typeless_storage argument, set
34025 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34027 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34028 (build_array_type): Likewise. Add typeless_storage argument.
34030 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34031 Jakub Jelinek <jakub@redhat.com>
34033 PR tree-optimization/80426
34034 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34035 operation on symbolic operands, also compute the overflow for the
34036 invariant part when the operation degenerates into a negation.
34038 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34041 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34042 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34045 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34047 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34050 * config/avr/avr.c (tree.h): Include it.
34051 (cgraph.h): Include it.
34052 (avr_encode_section_info): Don't warn for uninitialized progmem
34053 variable if it's just an alias.
34055 2017-04-19 Richard Biener <rguenther@suse.de>
34058 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34059 when needed by AutoPGO.
34061 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34064 * doc/lto.texi: Remove an extra 'that'.
34066 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34068 PR rtl-optimization/80429
34069 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34070 are only used in debug insns.
34072 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34073 Vladimir Makarov <vmakarov@redhat.com>
34075 * config/sparc/predicates.md (input_operand): Add comment. Return
34076 true for any memory operand when LRA is in progress.
34077 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34079 2017-04-18 Jeff Law <law@redhat.com>
34082 * mips.md ({return,simple_return}_internal): Do not overwrite
34085 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34087 PR tree-optimization/80443
34088 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34089 instead of adding 1, subtract -1 and similarly instead of subtracting
34092 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34094 PR rtl-optimization/80357
34095 * haifa-sched.c (tmp_bitmap): New variable.
34096 (model_recompute): Handle duplicate use records.
34097 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34098 (free_global_sched_pressure_data): Free it.
34100 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34103 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34104 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34105 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34106 instead of SYSTEM_HEADER_DIR.
34108 2017-04-18 Jeff Law <law@redhat.com>
34110 PR middle-end/80422
34111 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34112 predecessors after walking up the insn chain.
34114 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34117 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34118 sizetype type into debug info.
34120 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34123 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34124 unneeded test for TARGET_UPPER_REGS_SF.
34125 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34127 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34130 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34131 instead of gsi_after_labels.
34133 2017-04-18 Jeff Law <law@redhat.com>
34135 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34139 2017-04-13 Jeff Law <law@redhat.com>
34140 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34141 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34143 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34146 * config/avr/avr.c (intl.h): Include it.
34147 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34149 2017-04-18 Martin Liska <mliska@suse.cz>
34151 PR gcov-profile/78783
34152 * gcov-tool.c (gcov_output_files): Validate that destination
34153 file is either removed by the tool or by a user.
34155 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34156 Guy Benyei <guybe@mellanox.com>
34158 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34159 block, and do not negate it, the stored id is already negative.
34161 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34163 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34165 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34168 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34169 masks of options that should be turned off if the VSX vector
34170 options are turned off.
34171 (OTHER_P8_VECTOR_MASKS): Likewise.
34172 (OTHER_VSX_VECTOR_MASKS): Likewise.
34173 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34174 rs6000_disable_incompatible_switches to validate no type switches
34176 (rs6000_incompatible_switch): New function to disallow turning on
34177 other vector options if -mno-vsx, -mno-power8-vector, or
34178 -mno-power9-vector are specified.
34180 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34182 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34184 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34186 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34187 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34188 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34189 (ARG_POINTER_CFA_OFFSET): Likewise.
34191 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34193 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34194 conditions to take advantage of various optimizations.
34196 2017-04-13 Jeff Law <law@redhat.com>
34198 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34199 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34200 (zero_extendsidi2_dext): Likewise.
34202 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34205 * fold-const.c (fold_ternary_loc): Revert
34206 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34209 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34211 PR rtl-optimization/80343
34212 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34213 new scratch pseudo.
34215 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34218 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34219 to ubsan_encode_value.
34221 2017-04-13 Jeff Law <law@redhat.com>
34223 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34224 appearing in DEBUG_INSNs.
34226 2017-04-13 Martin Liska <mliska@suse.cz>
34228 PR gcov-profile/80413
34229 * gcov-io.c (gcov_write_string): Copy to buffer just when
34230 allocated size is greater than zero.
34232 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34235 * dwarf2out.c (decls_for_scope): Ignore declarations of
34236 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34238 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34241 * ipa-visibility.c (non_local_p): Fix typos.
34242 (localize_node): When localizing symbol in same comdat group,
34243 dissolve the group only when we know external symbols are going
34245 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34247 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34249 PR tree-optimization/79390
34250 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34251 order does not result in usable sequence, retry with reversed operand
34257 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34258 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34259 op0 instead of fold_convert_loc (loc, type, arg0).
34261 2017-04-12 Jeff Law <law@redhat.com>
34263 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34264 has a delay slot in the generated code.
34266 * config/cris/cris.md (cris_preferred_reload_class): Return
34267 GENNONACR_REGS rather than GENERAL_REGS.
34269 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34272 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34273 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34274 signedness of the result type.
34276 2017-04-12 Richard Biener <rguenther@suse.de>
34277 Jeff Law <law@redhat.com>
34279 PR tree-optimization/80359
34280 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34281 trim stores to TARGET_MEM_REFs.
34283 2017-04-12 Richard Biener <rguenther@suse.de>
34285 PR tree-optimization/79390
34286 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34287 threading case even more.
34289 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34292 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34293 for quad_address_p for TImode, instead of just not indexed_address.
34295 2017-04-12 Richard Biener <rguenther@suse.de>
34296 Bernd Edlinger <bernd.edlinger@hotmail.de>
34298 PR middle-end/79671
34299 * alias.c (component_uses_parent_alias_set_from): Handle
34300 TYPE_TYPELESS_STORAGE.
34301 (get_alias_set): Likewise.
34302 * tree-core.h (tree_type_common): Add typeless_storage flag.
34303 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34304 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34305 for types containing members with TYPE_TYPELESS_STORAGE.
34306 (place_field): Likewise.
34307 (layout_type): Likewise for ARRAY_TYPE.
34308 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34309 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34310 TYPE_TYPELESS_STORAGE.
34311 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34313 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34316 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34317 first argument to type.
34319 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34323 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34324 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34325 (rs6000_expand_binop_builtin): Likewise.
34326 (rs6000_expand_ternop_builtin): Likewise; also add missing
34327 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34328 vshasigma built-ins.
34329 * doc/extend.texi: Document that vec_xxpermdi's third argument
34330 must be a constant.
34332 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34334 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34335 Use shift_const cost parameter when calculating gain of STV shifts.
34337 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34339 PR rtl-optimization/70478
34340 * lra-constraints.c (process_alt_operands): Check memory for
34341 disfavoring memory insn operand.
34343 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34345 PR middle-end/80100
34346 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34347 left shift in unsigned HOST_WIDE_INT type.
34349 PR rtl-optimization/80385
34350 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34351 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34354 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34355 if they have any depend clauses.
34357 2017-04-11 Martin Liska <mliska@suse.cz>
34360 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34361 * ipa-split.c (split_function): Create a local comdat symbol
34362 if caller is in a comdat group.
34364 2017-04-11 Martin Liska <mliska@suse.cz>
34367 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34370 2017-04-11 Martin Sebor <msebor@redhat.com>
34372 PR middle-end/80364
34373 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34374 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34376 (directive::set_width, directive::set_precision, format_character):
34378 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34381 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34384 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34385 conflict, set target->arch_name instead of target->cpu_name.
34387 2017-04-11 Richard Biener <rguenther@suse.de>
34389 PR tree-optimization/80374
34390 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34391 build_zero_cst, remove fold_convertible_p check again.
34393 2017-04-11 Martin Liska <mliska@suse.cz>
34396 * ubsan.c (instrument_object_size): Do not instrument register
34399 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34402 * config/i386/i386-builtin-types.def
34403 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34404 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34405 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34406 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34407 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34408 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34409 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34410 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34411 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34412 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34413 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34414 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34415 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34416 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34417 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34418 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34419 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34420 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34421 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34422 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34423 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34424 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34425 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34426 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34427 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34428 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34429 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34430 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34431 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34432 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34433 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34434 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34435 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34436 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34437 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34438 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34439 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34440 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34441 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34442 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34443 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34444 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34445 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34446 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34447 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34448 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34450 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34451 flag to second_arg_count, handle 4 argument function type _COUNT
34452 aliases, handle second_arg_count on second argument rather than last.
34454 2017-04-10 Jeff Law <law@redhat.com>
34456 PR tree-optimization/80374
34457 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34458 record anything if we can not convert integer_zero_node to the
34461 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34464 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34465 Enhance special handling given to the TARGET_P9_MINMAX option in
34466 relation to certain other options.
34468 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34470 PR tree-optimization/80153
34471 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34472 remove POINTER_PLUS_EXPR's base part directly, rather than through
34475 2017-04-10 Richard Biener <rguenther@suse.de>
34476 Bin Cheng <bin.cheng@arm.com>
34478 PR tree-optimization/80153
34479 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34480 the first element of pointer type aff_tree. Build result expr in
34482 (add_elt_to_tree): Convert to type unconditionally. Remove other
34483 fold_convert calls.
34484 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34485 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34487 2017-04-10 Richard Biener <rguenther@suse.de>
34489 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34492 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34494 PR rtl-optimization/70478
34495 * lra-constraints.c (curr_small_class_check): New.
34496 (update_and_check_small_class_inputs): New.
34497 (process_alt_operands): Update curr_small_class_check. Disfavor
34498 alternative insn memory operands. Check available regs for small
34501 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34504 * config/mips/mips.opt (-mvirt): Update description.
34505 * doc/invoke.texi (-mvirt): Likewise.
34507 2017-04-10 Richard Biener <rguenther@suse.de>
34509 PR middle-end/80362
34510 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34511 looking for NEGATE_EXPR in -A / -B to A / B folding.
34513 2017-04-10 Martin Liska <mliska@suse.cz>
34515 PR gcov-profile/80224
34516 * gcov.c (print_usage): Fix usage string.
34517 (get_gcov_intermediate_filename): Remove.
34518 (output_gcov_file): Use both for normal and intermediate format.
34519 (generate_results): Do not initialize special file for
34520 intermediate format.
34522 2017-04-10 Richard Biener <rguenther@suse.de>
34524 PR tree-optimization/80304
34525 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34528 2017-04-10 Nathan Sidwell <nathan@acm.org>
34531 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34532 (rs6000_init_builtins): Use it.
34534 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34536 * config/arm/arm.md (<mrc>): Add mode to SET source.
34537 (<mrrc>): Likewise.
34539 2017-04-10 Richard Biener <rguenther@suse.de>
34541 PR middle-end/80344
34542 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34544 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34547 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34548 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34549 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34550 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34551 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34552 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34553 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34554 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34555 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34556 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34557 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34558 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34559 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34560 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34561 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34562 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34563 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34564 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34565 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34566 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34567 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34568 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34569 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34571 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34573 PR rtl-optimization/70478
34574 * lra-constraints.c: Reverse the last patch.
34576 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34578 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34579 Add comment for WCHAR_T.
34581 2017-04-08 Martin Liska <mliska@suse.cz>
34584 2017-04-07 Martin Liska <mliska@suse.cz>
34587 * ipa-split.c (split_function): Add function part to a same comdat
34590 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34593 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34595 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34597 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34598 (vec_pairing): ...to this.
34599 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34600 (rs6000_sched_init): Adjust for name change.
34601 (struct rs6000_sched_context): Likewise.
34602 (rs6000_init_sched_context): Likewise.
34603 (rs6000_set_sched_context): Likewise.
34605 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34611 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34613 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34614 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34615 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34617 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34619 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34621 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34623 PR rtl-optimization/70703
34624 * ira-color.c (update_conflict_hard_regno_costs): Use
34625 int64_t instead of HOST_WIDE_INT.
34627 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34629 PR rtl-optimization/70478
34630 * lra-constraints.c (process_alt_operands): Disfavor alternative
34631 insn memory operands.
34633 2017-04-07 Jeff Law <law@redhat.com>
34635 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34636 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34638 2017-04-07 Martin Liska <mliska@suse.cz>
34641 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34642 Show error message instead of an ICE.
34644 2017-04-07 Martin Liska <mliska@suse.cz>
34647 * ipa-split.c (split_function): Add function part to a same comdat
34650 2017-04-07 Richard Biener <rguenther@suse.de>
34652 PR middle-end/80341
34653 * tree.c (get_unwidened): Also handle ! for_type case for
34655 * convert.c (do_narrow): Split out from ...
34656 (convert_to_integer_1): ... here. Do not pass final truncation
34657 type to get_unwidened for TRUNC_DIV_EXPR.
34659 2017-04-07 Richard Biener <rguenther@suse.de>
34661 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34663 (aff_combination_const): Adjust.
34664 (aff_combination_scale): Likewise.
34665 (aff_combination_add_elt): Likewise.
34666 (aff_combination_add_cst): Likewise.
34667 (aff_combination_convert): Likewise.
34668 (add_elt_to_tree): Likewise. Remove unused argument.
34669 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34671 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34673 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34675 * config/arm/arm.c (arm_default_short_enums): Use
34676 ARM_DEFAULT_SHORT_ENUMS.
34677 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34679 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34682 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34683 members with redundant out-of-class redeclaration.
34685 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34688 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34689 * config/i386/i386.md (*zero_extendsidi2):
34690 Add (?*x,*x) and (?*v,*v) alternatives.
34692 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34695 * config/i386/i386.c (ix86_expand_builtin)
34696 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34697 mode from insn data. Convert operands to insn operand mode.
34698 Copy operands that don't satisfy insn predicate to a register.
34700 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34702 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34705 2017-04-06 Richard Biener <rguenther@suse.de>
34707 PR tree-optimization/80334
34708 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34709 preserve alignment of accesses.
34711 2017-04-06 Richard Biener <rguenther@suse.de>
34713 PR tree-optimization/80262
34714 * tree-sra.c (build_ref_for_offset): Preserve address-space
34716 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34717 Drop useless address-space information on MEM_REF offsets.
34719 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34721 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34723 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34725 PR rtl-optimization/70703
34726 * ira-color.c (update_conflict_hard_regno_costs): Use
34727 HOST_WIDE_INT instead of long.
34729 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34732 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34733 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34735 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34736 for x86_64 target. Handle -m3dnowa option.
34738 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34740 PR rtl-optimization/70703
34741 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34742 (update_conflict_hard_regno_costs): Use long instead of unsigned
34743 arithmetic for cost calculation.
34745 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34746 Bernd Edlinger <bernd.edlinger@hotmail.de>
34749 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34752 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34755 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34756 ptr_mode with Pmode throughout.
34757 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34758 into probe_stack_range and use DImode.
34760 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34763 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34764 call_eh_return is true.
34766 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34768 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34769 Initialize last_match_fntype_index.
34771 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34774 * tree-nvr.c: Include internal-fn.h.
34775 (pass_return_slot::execute): Ignore internal calls without
34778 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34779 Richard Biener <rguenther@suse.de>
34782 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34783 captures used multiple times, except for the last use.
34784 * generic-match-head.c: Include gimplify.h.
34786 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34788 PR tree-optimization/79390
34789 * target.h (struct noce_if_info): Declare.
34790 * targhooks.h (default_noce_conversion_profitable_p): Declare.
34791 * target.def (noce_conversion_profitable_p): New target hook.
34792 * ifcvt.h (struct noce_if_info): New type, moved from ...
34793 * ifcvt.c (struct noce_if_info): ... here.
34794 (noce_conversion_profitable_p): Renamed to ...
34795 (default_noce_conversion_profitable_p): ... this. No longer
34797 (noce_try_store_flag_constants, noce_try_addcc,
34798 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
34799 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
34800 instead of noce_conversion_profitable_p.
34801 * config/i386/i386.c: Include ifcvt.h.
34802 (ix86_option_override_internal): Don't override
34803 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
34804 (ix86_noce_conversion_profitable_p): New function.
34805 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
34806 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
34807 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
34808 * doc/tm.texi: Regenerated.
34810 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34812 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
34815 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34818 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34819 instructions for small multiply cores.
34821 2017-04-04 Jeff Law <law@redhat.com>
34823 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34825 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34826 that are not set by the loop over the elements.
34828 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34831 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34832 int mode, convert_modes it to mode as unsigned, otherwise use
34833 lowpart_subreg to mode rather than SImode.
34834 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34835 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34836 Use DImode instead of SImode for the shift count operand.
34837 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34840 2017-04-04 Richard Biener <rguenther@suse.de>
34842 PR middle-end/80281
34843 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
34844 arithmetic done for the negate or the plus. Simplify.
34845 (A - (-B) -> A + B): Likewise.
34846 * fold-const.c (split_tree): Make sure to not negate pointers.
34848 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
34850 PR rtl-optimization/60818
34851 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34852 a compare of comparisons with the thing compared if this results
34853 in a different machine mode.
34855 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
34857 * alias.c (base_alias_check): Fix typo in comment.
34858 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
34859 * cgraphunit.c (symbol_table::compile): Likewise.
34860 * collect2.c (maybe_run_lto_and_relink): Likewise.
34861 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
34862 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
34863 * config/avr/avr.c (avr_map_op_t): Likewise.
34864 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
34865 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
34866 * config/epiphany/epiphany.md (movcc): Likewise.
34867 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
34868 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
34870 * config/mips/mips.c (mips_save_restore_reg): Likewise.
34871 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
34872 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
34873 * config/sh/sh.c (sh_rtx_costs): Likewise.
34874 * fold-const.c (fold_truth_andor): Likewise.
34875 * genautomata.c (collapse_flag): Likewise.
34876 * gengtype.h (struct type::u::s): Likewise.
34877 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34878 * input.c (FORMAT_AMOUNT): Likewise.
34879 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34880 (known_aggs_to_agg_replacement_list): Likewise.
34881 * ipa-inline-analysis.c: Likewise.
34882 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34883 * ipa-polymorphic-call.c
34884 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34885 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34886 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34888 * modulo-sched.c (apply_reg_moves): Likewise.
34889 * omp-expand.c (build_omp_regions_1): Likewise.
34890 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34891 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34892 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34893 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34894 * value-prof.c: Likewise.
34895 * var-tracking.c (val_reset): Likewise.
34897 2017-04-03 Richard Biener <rguenther@suse.de>
34899 PR tree-optimization/80275
34900 * fold-const.c (split_address_to_core_and_offset): Handle
34903 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34905 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34906 descriptors is at least equal to that of functions.
34908 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34910 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34912 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34915 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34916 (mov<IMOD4:mode>): New expander.
34917 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34919 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34921 PR rtl-optimization/79405
34922 * fwprop.c (propagations_left): New variable.
34923 (forward_propagate_into): Decrement it.
34924 (fwprop_init): Initialize it.
34925 (fw_prop): If the variable has reached zero, stop propagating.
34926 (fwprop_addr): Ditto.
34928 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34931 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34932 a FUNCTION_DECL, pass it as decl instead of origin to
34935 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34937 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34940 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34943 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34944 TARGET_VSX_SMALL_INTEGER.
34946 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34948 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34949 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34951 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34953 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34954 extraction from odd-numbered MSA register.
34956 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34958 PR middle-end/80173
34959 * expmed.c (store_bit_field_1): Don't attempt to create
34960 a word subreg out of hard registers wider than word if they
34961 have HARD_REGNO_NREGS of 1 for their mode.
34963 PR middle-end/80163
34964 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34965 conversions to integer types wider than word and pointer.
34968 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34969 (rtx_equal_for_cselib_p): Pass 0 to it.
34970 * cselib.c (cselib_hasher::equal): Likewise.
34971 (rtx_equal_for_cselib_1): Add depth argument. If depth
34972 is 128, don't look up VALUE locs and punt. Increment
34973 depth in recursive calls when walking VALUE locs.
34975 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34977 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34978 (make_gcov_file_name): Use the canonical path name for generating
34980 (read_line): Fix handling of files with ascii null bytes.
34982 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34984 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34985 to initialise a vector register instead
34986 of using a const_int.
34988 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34990 PR translation/80189
34991 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34992 diagnostic messages.
34994 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34997 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34998 (dfp_diex_<mode>): Update mode of operand 1.
34999 * doc/extend.texi (dxex, dxexq): Document change to return type.
35000 (diex, diexq): Document change to argument type.
35002 2017-03-30 Martin Jambor <mjambor@suse.cz>
35005 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35006 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35007 it reflects the signature changes performed at the callee side.
35008 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35009 to cgraph_build_function_type_skip_args.
35010 (build_function_decl_skip_args): Adjust call to the above function.
35012 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35015 * config/i386/sse.md
35016 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35017 register as dest whenever it is a MEM not rtx_equal_p to the
35018 corresponding dup operand, and when forcing into reg move the
35019 reg into the memory afterwards.
35020 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35021 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35022 for the force_reg mode.
35023 (avx512vl_vextractf128<mode>): Use register as dest either
35024 always when a MEM, or when it is a MEM not rtx_equal_p to the
35025 corresponding dup operand, or even not when it is a CONST_VECTOR
35026 depending on the mode and lo vs. hi.
35027 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35029 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35030 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35031 Likewise. Require that operands[2] is even.
35032 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35033 Remove extraneous parens. Require that operands[2] is a multiple
35035 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35036 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35037 disallow memory then.
35039 2017-03-30 Richard Biener <rguenther@suse.de>
35041 PR tree-optimization/77498
35042 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35043 to non-constants over backedges.
35045 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35047 PR rtl-optimization/80233
35048 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35049 as last_combined_insn. Do not test for BARRIER_P separately.
35051 2017-03-29 Andreas Schwab <schwab@suse.de>
35054 * calls.c (prepare_call_address): Convert funexp to Pmode before
35055 copying to temp reg.
35057 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35059 PR tree-optimization/80158
35060 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35061 Handle possible future case of more than one alternate
35063 (replace_rhs_if_not_dup): Likewise.
35064 (replace_one_candidate): Likewise.
35066 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35068 PR rtl-optimization/80193
35069 * ira.c (ira): Do not check allocation for LRA.
35071 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35073 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35074 (nvptx_output_simt_exit): Declare.
35075 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35076 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35077 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35078 (nvptx_declare_function_name): ...here. Emit declaration of local
35079 memory space buffer for omp_simt_enter insn.
35080 (nvptx_output_unisimt_switch): New.
35081 (nvptx_output_softstack_switch): New.
35082 (nvptx_output_simt_enter): New.
35083 (nvptx_output_simt_exit): New.
35084 * config/nvptx/nvptx.h (struct machine_function): New fields
35085 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35086 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35087 (UNSPECV_SIMT_EXIT): Ditto.
35088 (omp_simt_enter_insn): New insn.
35089 (omp_simt_enter): New expansion.
35090 (omp_simt_exit): New insn.
35091 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35093 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35094 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35095 (expand_GOMP_SIMT_EXIT): New.
35096 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35097 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35098 (GOMP_SIMT_EXIT): Ditto.
35099 * target-insns.def (omp_simt_enter): New insn.
35100 (omp_simt_exit): Ditto.
35101 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35103 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35104 (lower_rec_input_clauses): Likewise.
35105 (lower_lastprivate_clauses): Handle SIMT privatization.
35107 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35108 (ompdevlow_adjust_simt_enter): New.
35109 (find_simtpriv_var_op): New.
35110 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35111 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35113 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35114 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35115 (copy_decl_for_dup_finish): Ditto.
35117 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35119 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35122 * config/i386/i386.c (ix86_option_override_internal): Always
35123 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35125 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35127 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35129 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35131 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35132 mark new edge's irreducible flag accordign to it.
35133 (vect_do_peeling): Check loop preheader edge's irreducible flag
35134 and pass it to function slpeel_add_loop_guard.
35136 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35138 PR tree-optimization/80218
35139 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35140 Update block frequencies and counts.
35142 2017-03-28 Richard Biener <rguenther@suse.de>
35144 PR tree-optimization/78644
35145 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35146 of a simplification result we may not use it at all.
35148 2017-03-28 Richard Biener <rguenther@suse.de>
35151 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35152 without arguments, generate default definition of a SSA name.
35154 2017-03-28 Richard Biener <rguenther@suse.de>
35156 PR middle-end/80222
35157 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35158 TYPE_REF_CAN_ALIAS_ALL references.
35159 * fold-const.c (fold_indirect_ref_1): Likewise.
35161 2017-03-28 Martin Liska <mliska@suse.cz>
35164 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35165 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35167 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35168 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35170 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35171 (EXTRA_SPECS): Define.
35172 (SUBTARGET_EXTRA_SPECS): Likewise.
35173 (SUBTARGET_CPP_SPEC): Likewise.
35174 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35175 SUBTARGET_EXTRA_SPECS.
35176 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35178 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35180 * config/arc/simdext.md (vst64_insn): Update pattern.
35181 (vld32wh_insn): Likewise.
35182 (vld32wl_insn): Likewise.
35183 (vld64_insn): Likewise.
35184 (vld32_insn): Likewise.
35186 2017-03-28 Marek Polacek <polacek@redhat.com>
35189 * fold-const.c (fold_comparison): Use protected_set_expr_location
35190 instead of SET_EXPR_LOCATION.
35192 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35194 * tree.c (add_expr): Avoid name lookup warning.
35196 2017-03-27 Jeff Law <law@redhat.com>
35198 PR tree-optimization/80216
35199 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35200 function name. Limit recursion depth.
35201 (record_temporary_equivalences): Corresponding changes.
35203 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35205 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35208 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35211 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35213 * cfgcleanup.c (reg_note_cfa_p): New array.
35214 (insns_have_identical_cfa_notes): New function.
35215 (old_insns_match_p): Don't cross-jump in between /f
35216 and non-/f instructions. If both i1 and i2 are frame related,
35217 verify all CFA notes, their order and content.
35219 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35222 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35223 HImode and SImode with zero extend to DImode to one insn.
35224 (bswap<mode>2_extenddi): Likewise.
35225 (bswapsi2_extenddi): Likewise.
35226 (bswaphi2_extendsi): Likewise.
35227 (bswaphi2): Combine bswap HImode and SImode into one insn.
35228 Separate memory insns from swapping register.
35229 (bswapsi2): Likewise.
35230 (bswap<mode>2): Likewise.
35231 (bswaphi2_internal): Delete, no longer used.
35232 (bswapsi2_internal): Likewise.
35233 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35234 store, and gpr<-gpr swap insns.
35235 (bswap<mode>2_store): Likewise.
35236 (bswaphi2_reg): Register only splitter, combine with the splitter.
35237 (bswaphi2 splitter): Likewise.
35238 (bswapsi2_reg): Likewise.
35239 (bswapsi2 splitter): Likewise.
35240 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35241 the insns into load, store, and register/register insns.
35242 (bswapdi2_ldbrx): Likewise.
35243 (bswapdi2_load): Likewise.
35244 (bswapdi2_store): Likewise.
35245 (bswapdi2_reg): Likewise.
35247 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35249 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35250 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35252 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35255 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35257 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35258 special handling for target option conflicts between dform
35259 options (-mpower9-dform, -mpower9-dform-vector,
35260 -mpower9-dform-scalar) and -mno-direct-move.
35262 2017-03-27 Richard Biener <rguenther@suse.de>
35264 PR tree-optimization/80181
35265 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35267 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35269 * config/arc/predicates.md (move_double_src_operand): Replace the
35270 call to move_double_src_operand with a call to address_operand.
35272 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35274 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35275 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35276 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35278 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35280 * config/arc/predicates.md (long_immediate_loadstore_operand):
35281 Consider scaled addresses cases.
35283 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35285 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35286 restored when in interrupt.
35287 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35288 doesn't have delay slot.
35290 2017-03-27 Richard Biener <rguenther@suse.de>
35293 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35294 inlined thunk clones.
35296 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35299 * asan.c (instrument_derefs): Copy over last operand from
35300 original COMPONENT_REF to the new COMPONENT_REF with
35301 DECL_BIT_FIELD_REPRESENTATIVE.
35302 * ubsan.c (instrument_object_size): Likewise.
35304 2017-03-27 Richard Biener <rguenther@suse.de>
35306 PR tree-optimization/80170
35307 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35308 sure DR/SCEV didnt fold in constants we do not see when looking
35309 at the reference base alignment.
35311 2017-03-27 Richard Biener <rguenther@suse.de>
35313 PR middle-end/80171
35314 * gimple-fold.c (fold_ctor_reference): Properly guard against
35315 NULL return value from canonicalize_constructor_val.
35317 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35320 * config/i386/i386.c (ix86_expand_builtin)
35321 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35322 flags reg setting and flags reg using instructions.
35323 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35324 clobbering instructions to zero extend op2.
35326 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35328 * doc/install.texi (Configuration) <--with-aix-soname>:
35329 Update link to AIX ld.
35331 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35333 PR rtl-optimization/80160
35334 PR rtl-optimization/80159
35335 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35336 reg_alternate_class into account.
35338 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35341 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35342 to consider in curr_insn_transform.
35344 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35346 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35347 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35348 and emit_mode_inner.
35350 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35352 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35353 argument to the overloaded builtin variants. Use the new flag to
35354 deprecate certain builtin variants.
35355 * config/s390/s390-builtin-types.def: Add new builtin types.
35356 * config/s390/s390-builtins.h: Support new flags field for
35357 overloaded builtins.
35358 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35359 (s390_macro_to_expand): Enable vector float data type.
35360 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35361 builtins by incrementing the __VEC__ version number.
35362 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35364 (s390_resolve_overloaded_builtin): Emit error messages depending
35365 on the builtin flags.
35366 * config/s390/s390.c (s390_expand_builtin): Support additional
35367 flags argument. Change error message to match the messages
35368 emitted in s390-c.c.
35369 * config/s390/s390.md: New UNSPEC_* constants.
35370 (op_type): Add new instruction types.
35371 * config/s390/vecintrin.h: Add new builtins and test data class
35373 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35374 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35375 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35376 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35377 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35378 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35380 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35381 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35382 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35383 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35385 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35386 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35387 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35388 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35390 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35391 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35392 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35393 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35394 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35395 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35396 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35398 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35399 ("vec_scatter_element<V_HW_4:mode>_DI")
35400 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35401 ("vec_fpint<mode>", "vflls")
35402 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35403 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35404 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35405 ("*vec_cmphe<mode>_cc"): ... these.
35407 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35408 mode constant instead of magic value.
35410 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35412 * config/s390/s390.c (s390_expand_vec_compare): Support other
35413 vector floating point modes than just V2DF.
35414 (s390_expand_vcond): Likewise.
35415 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35416 (s390_cannot_change_mode_class): Prevent mode changes between TF
35417 and V1TF in vector registers.
35418 * config/s390/s390.md (DF, SF): New mode attributes.
35419 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35420 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35421 SFmode support for VRs.
35422 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35424 (VFT, VF_HW): New mode iterators.
35425 (vw, sdx): New mode attributes.
35426 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35427 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35428 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35429 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35430 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35431 also the new vector floating point modes. Renaming to ...
35433 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35434 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35435 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35436 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35437 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35438 ("vec_unordered<mode>"): ... these.
35440 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35441 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35442 ("*vec_extendv2df"): New insn definitions.
35444 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35446 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35447 ("mulditi3_2", "*muldi3_sign"): New patterns.
35448 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35449 rename the pattern definition.
35451 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35453 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35455 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35457 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35459 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35460 instruction if possible.
35461 * config/s390/vector.md (vec_halfnumelts): New mode
35463 ("*vec_vllezlf<mode>"): New pattern.
35465 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35467 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35468 ("popcountv4si2", "popcountv2di2"): Rename to ...
35469 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35470 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35472 ("popcount<mode>2_vxe"): New pattern.
35474 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35476 * common/config/s390/s390-common.c (processor_flags_table): Add
35478 * config.gcc: Add arch12.
35479 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35480 Default to arch12 for unknown CPU model numbers.
35481 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35482 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35483 PROCESSOR_max sanity check.
35484 * config/s390/s390-opts.h (enum processor_type): Add
35486 * config/s390/s390.c (processor_table): Add arch12.
35487 (s390_expand_builtin): Add check for B_VXE flag.
35488 (s390_issue_rate): Add PROCESSOR_ARCH12.
35489 (s390_get_sched_attrmask): Likewise.
35490 (s390_get_unit_mask): Likewise.
35491 (s390_sched_score): Enable z13 scheduling for arch12.
35492 (s390_sched_reorder): Likewise.
35493 (s390_sched_variable_issue): Likewise.
35494 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35496 (s390_tune_attr): Use z13 scheduling also for arch12.
35497 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35498 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35499 (TARGET_VXE_P): New macros.
35500 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35501 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35502 * config/s390/s390.opt: Add arch12 as processor_type.
35504 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35506 * config/s390/s390.md
35507 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35508 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35509 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35511 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35512 Rename expanders to ...
35514 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35515 ("fixuns_truncdddi2_emu"): ... these.
35517 ("fixuns_trunc<mode>si2_emu"): New expander.
35519 ("*fixuns_truncdfdi2_z13"): Rename to ...
35520 ("*fixuns_truncdfdi2_vx"): ... this.
35522 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35524 * config/s390/2964.md: Remove the single element vector compare
35525 instructions which are no longer used.
35526 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35528 (s390_canonicalize_comparison): Remove handling of DFmode
35530 (s390_expand_vec_compare_scalar): Remove function.
35531 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35532 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35534 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35536 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35538 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35540 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35541 will anyway by matched by mov<mode>_64dfp.
35543 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35545 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35546 vlef/vstef. Add missing operand to vleif.
35548 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35550 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35551 pair for all vector types with 64 bit elements.
35552 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35553 * config/s390/vector.md (V_HW_64): ... here.
35554 (V_128_NOSINGLE): New mode iterator.
35555 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35556 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35557 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35558 ("*vec_load_pairv2di"): Change to ...
35559 ("*vec_load_pair<mode>"): ... this one.
35561 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35563 * config/s390/constraints.md: Add comments.
35564 (jKK): Reject element sizes > 8 bytes.
35565 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35567 * config/s390/s390.md: Add the s_operand checks formerly in
35568 s390_split_ok_p to various splitters where they are still
35570 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35571 for 128 bit vectors. Plus two splitters.
35573 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35575 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35578 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35581 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35582 error if the boundary argument is not constant.
35584 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35586 PR rtl-optimization/80112
35587 * loop-doloop.c (doloop_condition_get): Don't check condition
35588 if cmp isn't SET with IF_THEN_ELSE src.
35590 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35592 PR tree-optimization/80158
35593 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35594 replacing a candidate statement, also replace it for the
35595 candidate's alternate interpretation.
35596 (replace_rhs_if_not_dup): Likewise.
35597 (replace_one_candidate): Likewise.
35599 2017-03-24 Richard Biener <rguenther@suse.de>
35601 PR tree-optimization/80167
35602 * graphite-isl-ast-to-gimple.c
35603 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35605 (translate_isl_ast_to_gimple::get_rename): Likewise.
35607 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35609 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35610 handling of certain combinations of target options, including the
35611 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35612 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35614 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35617 * config/arm/arm.md (*load_multiple): Add reload_completed to
35618 matching condition.
35620 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35621 Richard Biener <rguenth@suse.de>
35623 PR tree-optimization/79908
35624 PR tree-optimization/80136
35625 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35626 been cast away, gimplify_and_add suffices.
35628 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35630 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35632 2017-03-23 Richard Biener <rguenther@suse.de>
35634 PR tree-optimization/80032
35635 * gimplify.c (gimple_push_cleanup): Forced unconditional
35636 cleanups still have to go to the conditional_cleanups
35639 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35641 PR tree-optimization/80072
35642 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35644 (next_operand_entry_id): Change type to unsigned int.
35645 (sort_by_operand_rank): Make sure to return the right return value
35646 even if unsigned fields are bigger than INT_MAX.
35647 (struct oecount): Change cnt and id type to unsigned int.
35648 (oecount_hasher::equal): Formatting fix.
35649 (oecount_cmp): Make sure to return the right return value
35650 even if unsigned fields are bigger than INT_MAX.
35651 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35654 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35655 TREE_READONLY on result if writing it more than once.
35658 * doc/invoke.texi (-fsanitize=thread): Document that with
35659 -fnon-call-exceptions atomics are not able to throw
35663 * tsan.c: Include tree-eh.h.
35664 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35665 maybe_clean_or_replace_eh_stmt where needed.
35666 (instrument_memory_accesses): Add cfg_changed argument.
35667 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35668 if it returned true.
35669 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35671 PR rtl-optimization/63191
35672 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35673 wrapper function, moved the whole old content into ...
35674 (ix86_delegitimize_address_1): ... this. New inline function.
35675 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35676 true as last argument instead of ix86_delegitimize_address.
35678 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35680 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35681 cortexa57_branch_cost.
35683 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35685 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35687 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35690 * doc/md.texi (Constraints): Document wA constraint.
35691 * config/rs6000/constraints.md (wA): New.
35692 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35693 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35694 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35695 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35697 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35700 * gimplify.c (is_oacc_declared): New function.
35701 (oacc_default_clause): Use it to set default flags for acc declared
35702 variables inside parallel regions.
35703 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35704 declared variables.
35705 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35706 declare attribute to any decl as necessary.
35708 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35711 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35712 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35713 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35714 (arm_arch_lpae): This.
35715 * config/arm/arm.c (arm_arch7ve): Rename into ...
35716 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35717 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35720 2017-03-22 Martin Liska <mliska@suse.cz>
35723 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35724 error message instead of an ICE.
35726 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35728 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35730 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35732 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35734 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35737 2017-03-21 Martin Sebor <msebor@redhat.com>
35739 * doc/extend.texi: Use "cannot" instead of "can't."
35740 * doc/hostconfig.texi: Same.
35741 * doc/install.texi: Same.
35742 * doc/invoke.texi: Same.
35743 * doc/loop.texi: Same.
35744 * doc/md.texi: Same.
35745 * doc/objc.texi: Same.
35746 * doc/rtl.texi: Same.
35747 * doc/tm.texi: Same.
35748 * doc/tm.texi.in: Same.
35749 * doc/trouble.texi: Same.
35751 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35754 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35755 (collect_checksum_attributes): Set it.
35756 (die_checksum_ordered): Use it.
35758 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35760 PR tree-optimization/79908
35761 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35762 change: For a VA_ARG whose LHS has been cast away, use
35763 force_gimple_operand to construct the side effects.
35765 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35767 PR translation/80001
35768 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35769 more amenable to translation.
35770 (oacc_loop_auto_partitions): Likewise.
35772 2017-03-21 Marek Polacek <polacek@redhat.com>
35773 Martin Sebor <msebor@redhat.com>
35775 PR tree-optimization/80109
35776 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35777 on INTEGRAL_TYPE_P.
35779 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35780 Segher Boessenkool <segher@kernel.crashing.org>
35783 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35784 check reg_used_between_p between insn and one of succ or succ2
35785 depending on if succ is artificial insn not inserted into insn
35788 2017-03-21 Martin Liska <mliska@suse.cz>
35790 PR gcov-profile/80081
35791 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
35792 * doc/gcc.texi: Include gcov-dump stuff.
35793 * doc/gcov-dump.texi: New file.
35795 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
35797 PR rtl-optimization/79150
35798 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
35799 conditional jump, if the jump is the last insn of the loop.
35801 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35802 Richard Biener <rguenth@suse.de>
35804 PR tree-optimization/79908
35805 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35806 been cast away, use force_gimple_operand to construct the side
35809 2017-03-21 Martin Liska <mliska@suse.cz>
35811 PR libfortran/79956
35812 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
35815 2017-03-21 Brad Spengler <spender@grsecurity.net>
35818 * plugin.c (htab_hash_plugin): New function.
35819 (add_new_plugin): Use it and adjust.
35820 (parse_plugin_arg_opt): Adjust.
35821 (init_one_plugin): Likewise.
35823 2017-03-21 Richard Biener <rguenther@suse.de>
35825 PR tree-optimization/80032
35826 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35827 if set force the cleanup to happen unconditionally.
35828 (gimplify_target_expr): Push inserted clobbers with force_uncond
35829 to avoid them being removed by control-dependent DCE.
35831 2017-03-21 Richard Biener <rguenther@suse.de>
35833 PR tree-optimization/80122
35834 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35835 va_arg_pack_len when the inlined call stmt requires pack
35837 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35839 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35842 * tsan.c (instrument_builtin_call): If the memory model argument
35843 is not a constant, assume it is valid.
35846 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
35849 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35851 PR rtl-optimization/79910
35852 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35853 if its dest is used by an insn before I2 (other than the combined
35854 insns themselves, which are properly handled already).
35856 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35859 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35861 * combine.c (record_used_regs): New static function.
35862 (try_combine): Handle situations where there is an additional
35863 instruction between I2 and I3 which needs to have a LOG_LINK
35867 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35869 * combine.c (try_combine): Delete redundant i1 test. Call
35870 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35872 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35875 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35876 alternatives 13/14.
35878 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35880 PR tree-optimization/80054
35881 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35882 the optimization if a PHI or any of its arguments is not dominated
35883 by the candidate's basis. Use gphi* rather than gimple* as
35885 (replace_profitable_candidates): Clean up a gimple* variable that
35886 should be a gphi* variable.
35888 2017-03-20 Martin Sebor <msebor@redhat.com>
35891 * doc/extend.texi (attribute constructor): Document present limitation.
35893 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35896 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35897 __POWER9_VECTOR__ #ifdef control, change template definition to
35898 use Power9-specific built-in function.
35899 (vec_any_eq): Likewise.
35900 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35901 to control outcomes from this test.
35902 (vector_ae_<mode>p): For VEC_F modes, likewise.
35904 2017-03-20 Ian Lance Taylor <iant@google.com>
35906 * config/i386/i386.c (ix86_function_regparm): Save an extra
35907 register for -fsplit-stack with DECL_STATIC_CHAIN.
35909 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35912 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35913 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35915 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35917 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35919 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35922 2017-03-20 Marek Polacek <polacek@redhat.com>
35925 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35927 2017-03-20 Richard Biener <rguenther@suse.de>
35929 PR tree-optimization/80113
35930 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35931 allocate extra SSA name for PHI def.
35932 (add_close_phis_to_outer_loops): Likewise.
35933 (add_close_phis_to_merge_points): Likewise.
35934 (copy_loop_close_phi_args): Likewise.
35935 (copy_cond_phi_nodes): Likewise.
35937 2017-03-20 Martin Liska <mliska@suse.cz>
35939 PR middle-end/79753
35940 * tree-chkp.c (chkp_build_returned_bound): Do not build
35941 returned bounds for a LHS that's not a BOUNDED_P type.
35943 2017-03-20 Martin Liska <mliska@suse.cz>
35947 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35948 COMPLEX_CST and VECTOR_CST.
35950 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35953 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35954 target operand. A new splitter adds the clobber statement in case
35955 the target operand is dead anyway.
35957 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35959 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35960 to age-old versions of binutils and glibc.
35962 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35964 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35966 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35968 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35970 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35972 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35973 requirement for binutils 2.13.
35975 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35977 * combine.c (try_combine): Delete redundant i1 test. Call
35978 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35980 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35982 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35983 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35985 <riscv64-*-elf>: Re-arrange section
35986 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35987 <riscv32-*-linux>: Likewise.
35988 <riscv64-*-elf>: Likewise
35989 <riscv64-*-linux>: Likewise.
35991 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35994 * aarch64.opt(verbose-cost-dump): Fix typo.
35996 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35999 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36000 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36002 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36004 * reload.c (find_reloads): When reloading a nonoffsettable address,
36005 use RELOAD_OTHER for it and its address reloads.
36007 PR rtl-optimization/79910
36008 * combine.c (record_used_regs): New static function.
36009 (try_combine): Handle situations where there is an additional
36010 instruction between I2 and I3 which needs to have a LOG_LINK
36013 2017-03-17 Jeff Law <law@redhat.com>
36015 PR tree-optimization/71437
36016 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36017 conditional in the hash table first.
36018 (vrp_dom_walker::before_dom_children): Extract condition from
36019 ASSERT_EXPR. Record condition, its inverion and any implied
36020 conditions as well.
36022 2017-03-17 Marek Polacek <polacek@redhat.com>
36023 Markus Trippelsdorf <markus@trippelsdorf.de>
36025 PR tree-optimization/80079
36026 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36029 2017-03-17 Richard Biener <rguenther@suse.de>
36031 PR middle-end/80075
36032 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36033 Properly verify the LHS before the RHS possibly claims to be
36035 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36038 2017-03-17 Martin Jambor <mjambor@suse.cz>
36040 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36041 (List of -O2 options): Likewise.
36042 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36045 2017-03-17 Tom de Vries <tom@codesourcery.com>
36047 * gcov-dump.c (print_usage): Print bug_report_url.
36049 2017-03-17 Richard Biener <rguenther@suse.de>
36051 PR middle-end/80050
36052 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36053 (parser::peek): Likewise.
36055 2017-03-17 Richard Biener <rguenther@suse.de>
36057 PR tree-optimization/80048
36058 * sese.c (free_sese_info): Properly release rename_map and
36059 copied_bb_map elements.
36061 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36063 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36064 Add linked-list forward and backlinks. Insert on
36065 construction, remove on destruction.
36066 (class pass_store_merging): Add m_stores_head field.
36067 (pass_store_merging::terminate_and_process_all_chains):
36068 Iterate over m_stores_head list.
36069 (pass_store_merging::terminate_all_aliasing_chains):
36071 (pass_store_merging::execute): Check for debug stmts first.
36072 Push new chains onto the m_stores_head stack.
36074 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36077 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36078 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36079 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36081 2017-03-16 Jeff Law <law@redhat.com>
36083 PR tree-optimization/71437
36084 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36085 member function. Implementation moved into after_dom_children
36086 member function and into the threader's thread_outgoing_edges
36088 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36089 some code into new thread_outgoing_edges.
36090 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36091 definition. Simplify marker handling (do it here). Assume we always
36092 have the available expression and the const/copies tables.
36093 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36095 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36096 * tree-vrp.c (equiv_stack): No longer file scoped.
36097 (vrp_dom_walker): New class.
36098 (vrp_dom_walker::before_dom_children): New member function.
36099 (vrp_dom_walker::after_dom_children): Likewise.
36100 (identify_jump_threads): Setup domwalker. Use it rather than
36101 walking edges in a random order by hand. Simplify setup/finalization.
36102 (finalize_jump_threads): Remove.
36103 (vrp_finalize): Do not call identify_jump_threads here.
36104 (execute_vrp): Do it here instead and call thread_through_all_blocks
36107 PR tree-optimization/71437
36108 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36110 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36112 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36113 (dom_opt_dom_walker::thread_across_edge): Remove
36114 handle_dominating_asserts argument. All callers changed.
36115 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36116 changes. Remove calls to lhs_of_dominating_assert. Other
36117 uses of handle_dominating_asserts turn into unconditional code
36118 (simplify_control_stmt_condition_1): Likewise.
36119 (simplify_control_stmt_condition): Likewise.
36120 (thread_through_normal_block, thread_across_edge): Likewise.
36121 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36122 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36123 object if it is not an SSA_NAME.
36124 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36125 before calling into the VRP specific simplifiers.
36126 (identify_jump_threads): Remove handle_dominating_asserts
36129 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36132 * tree-diagnostic.c (default_tree_printer): No longer static.
36133 * tree-diagnostic.h (default_tree_printer): New prototype.
36135 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36137 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36138 Change ins into fmov.
36140 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36142 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36143 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36144 Use h_con constraint for operand 1.
36145 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36146 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36148 2017-03-15 Jeff Law <law@redhat.com>
36150 PR tree-optimization/71437
36151 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36152 (record_temporary_equivalences): Use it.
36154 PR tree-optimization/71437
36155 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36156 tree-ssa-scopedtables.
36157 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36158 (record_conditions, record_cond, vuse_eq): Likewise.
36159 (record_edge_info): Adjust to API tweak of record_conditions.
36160 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36161 (record_temporary_equivalences, optimize_stmt): Likewise.
36162 (eliminate_redundant_computations): Likewise.
36163 (record_equivalences_from_stmt): Likewise.
36164 * tree-ssa-scopedtables.c: Include options.h and params.h.
36165 (vuse_eq): New function, moved from tree-ssa-dom.c
36166 (build_and_record_new_cond): Likewise.
36167 (record_conditions): Likewise. Accept vector of conditions rather
36168 than edge_equivalence structure for first argument.
36169 for the first argument.
36170 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36171 from tree-ssa-dom.c.
36172 (avail_exprs_stack::record_cond): Likewise.
36173 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36174 from tree-ssa-dom.c.
36175 (avail_exprs_stack): Add new member functions lookup_avail_expr
36177 (record_conditions): Declare.
36179 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36182 * lra-constraints.c (process_alt_operands): Increase reject for
36183 reloading an input/output operand.
36185 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36188 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36189 insns to convert from signed/unsigned char/short to IEEE 128-bit
36191 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36193 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36196 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36197 subreg of inner mode for values already in registers.
36199 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36201 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36202 iteration reg is used after the loop.
36204 2017-03-14 Martin Sebor <msebor@redhat.com>
36206 PR tree-optimization/79800
36207 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36208 precision in negative-positive range.
36209 (format_floating): Call non-const overload with adjusted precision.
36211 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36214 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36217 2017-03-14 Martin Sebor <msebor@redhat.com>
36219 PR middle-end/80020
36220 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36221 * builtins.def (aligned_alloc): Use it.
36224 * Makefile.in (GTFILES): Add calls.c.
36225 * calls.c: Include "gt-calls.h".
36227 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36229 PR rtl-optimization/79728
36230 * regs.h (struct target_regs): New field
36231 x_contains_allocatable_regs_of_mode.
36232 (contains_allocatable_regs_of_mode): New macro.
36233 * reginfo.c (init_reg_sets_1): Initialize it, and change
36234 contains_reg_of_mode so it includes global regs as well.
36235 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36236 rather than contains_regs_of_mode.
36238 2017-03-14 Martin Liska <mliska@suse.cz>
36240 * doc/invoke.texi: Document options that can't be combined with
36241 -fcheck-pointer-bounds.
36243 2017-03-14 Martin Liska <mliska@suse.cz>
36245 PR middle-end/79831
36246 * doc/invoke.texi (-Wchkp): Document the option.
36248 2017-03-14 Martin Liska <mliska@suse.cz>
36250 * Makefile.in: Install gcov-dump.
36252 2017-03-14 Martin Liska <mliska@suse.cz>
36254 * multiple_target.c (expand_target_clones): Bail out for
36255 an invalid attribute.
36257 2017-03-14 Richard Biener <rguenther@suse.de>
36259 * alias.c (struct alias_set_entry): Pack properly.
36260 * cfgloop.h (struct loop): Likewise.
36261 * cse.c (struct set): Likewise.
36262 * ipa-utils.c (struct searchc_env): Likewise.
36263 * loop-invariant.c (struct invariant): Likewise.
36264 * lra-remat.c (struct cand): Likewise.
36265 * recog.c (struct change_t): Likewise.
36266 * rtl.h (struct address_info): Likewise.
36267 * symbol-summary.h (function_summary): Likewise.
36268 * tree-loop-distribution.c (struct partition): Likewise.
36269 * tree-object-size.c (struct object_size_info): Likewise.
36270 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36271 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36272 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36273 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36274 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36275 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36276 (struct _stmt_vec_info): Likewise.
36278 2017-03-14 Martin Liska <mliska@suse.cz>
36281 * multiple_target.c (create_dispatcher_calls): Check that
36282 a target can create a function dispatcher.
36284 2017-03-14 Martin Liska <mliska@suse.cz>
36287 * multiple_target.c (expand_target_clones): Drop local.local
36288 flag for default implementation.
36290 2017-03-14 Richard Biener <rguenther@suse.de>
36292 PR tree-optimization/80030
36293 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36295 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36297 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36298 gcc_fallthrough() instead of __attribute__((fallthrough));
36300 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36302 * doc/gcc.texi: Remove "up" link to (DIR).
36303 * doc/gccint.texi: Ditto.
36305 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36307 * doc/install.texi (Specific) <avr>: Remove reference to
36310 2017-03-13 Jeff Law <law@redhat.com>
36312 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36313 attribute rather than comments.
36315 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36316 match_scratch operand is highest.
36318 2017-03-13 Martin Liska <mliska@suse.cz>
36320 PR middle-end/78339
36321 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36322 is a CHKP clone, use original declaration.
36324 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36326 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36327 (arc_conditional_register_usage): Use a different allocation order
36328 when optimizing for size.
36329 * common/config/arc/arc-common.c (arc_option_optimization_table):
36330 Section anchors default on when optimizing for size.
36332 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36334 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36336 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36338 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36339 * config/arc/arc.md (cpu_facility): Add cd variant.
36340 (*movqi_insn): Add code density variant.
36341 (*movhi_insn): Likewise.
36342 (*movqi_insn): Likewise.
36343 (*addsi3_mixed): Likewise.
36344 (subsi3_insn): Likewise.
36346 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36348 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36350 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36352 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36353 expressions with MINUS and UNARY ops.
36355 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36358 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36360 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36361 between vec_select and vector argument.
36362 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36363 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36364 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36365 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36366 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36367 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36369 2017-03-13 Richard Biener <rguenther@suse.de>
36372 * params.def (vect-max-peeling-for-alignment): Fix typo.
36374 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36376 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36377 issue that only occurred with binutils below 2.18.
36379 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36381 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36382 refer to binutils 2.11/2.12 minimum.
36384 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36386 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36387 ftp.kernel.org and simplify binutils requirement.
36389 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36391 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36393 (Optimize Options): Ditto. Also remove redundancy.
36395 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36397 PR translation/79848
36398 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36400 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36401 to G_ to avoid double translation.
36403 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36405 PR translation/79923
36406 * auto-profile.c (get_combined_location): Convert leading
36407 character of diagnostics to lower case and remove trailing period.
36408 (read_profile): Likewise for various diagnostics.
36409 * config/arm/arm.c (arm_option_override): Remove trailing period
36410 from various diagnostics.
36411 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36412 (msp430_expand_delay_cycles): Likewise.
36414 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36417 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36418 full command-line argument, rather than just "str".
36419 (aarch64_validate_march): Likewise.
36420 (aarch64_validate_mtune): Likewise.
36422 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36424 PR rtl-optimization/78911
36425 * lra-assigns.c (must_not_spill_p): New function.
36426 (spill_for): Use it.
36428 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36430 PR tree-optimization/79981
36431 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36432 ATOMIC_COMPARE_EXCHANGE ifn result.
36433 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36434 IFN_ATOMIC_COMPARE_EXCHANGE.
36436 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36439 * opts.c (parse_sanitizer_options): Add missing question mark to
36440 "did you mean" message.
36442 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36444 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36446 (VMULEUH_UNS): Likewise.
36447 (VMULOUB_UNS): Likewise.
36448 (VMULOUH_UNS): Likewise.
36449 * config/rs6000/rs6000.c (builtin_function_type): Remove
36450 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36452 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36455 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36456 x with result of extra_parsing_for_operand_code_0.
36457 (function_reader::extra_parsing_for_operand_code_0): Convert
36458 return type from void to rtx, returning x. When reading
36459 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36460 larger size containing struct block_symbol.
36462 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36464 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36465 -mfloat128-hardware without -m64.
36467 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36470 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36471 entries to the case statement that marks unsigned arguments to
36472 overloaded functions.
36474 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36476 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36477 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36479 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36482 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36483 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36485 2017-03-10 Martin Liska <mliska@suse.cz>
36489 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36490 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36493 2017-03-10 Olivier Hainque <hainque@adacore.com>
36495 * tree-switch-conversion (array_value_type): Start by resetting
36496 candidate type to it's main variant.
36498 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36500 PR rtl-optimization/79909
36501 * combine.c (try_combine): Use simplify_replace_rtx on individual
36502 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36503 of the whole CALL_INSN_FUNCTION_USAGE.
36505 PR tree-optimization/79972
36506 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36507 get_range_info on SSA_NAMEs. Formatting fixes.
36509 2017-03-10 Richard Biener <rguenther@suse.de>
36510 Jakub Jelinek <jakub@redhat.com>
36512 PR tree-optimization/77975
36513 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36514 edge to be constant.
36515 (get_val_for): For constant x return it. Formatting fix.
36516 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36517 would use the same bases as the current one.
36519 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36521 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36522 instead of vec_select for V1TImode.
36523 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36525 (VSX_LE_128): Add V1TI to this mode iterator.
36526 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36527 (*vsx_le_perm_store_<mode>): Likewise.
36528 (pre-reload splitter for VSX stores): Likewise.
36529 (post-reload splitter for VSX stores): Likewise.
36530 (*vsx_xxpermdi2_le_<mode>): Likewise.
36531 (*vsx_lxvd2x2_le_<mode>): Likewise.
36532 (*vsx_stxvd2x2_le_<mode>): Likewise.
36534 2017-03-09 Michael Eager <eager@eagercon.com>
36536 Correct failures with --enable-checking=yes,rtl.
36538 * config/microblaze/microblaze.c (microblaze_expand_shift):
36539 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36540 test for const0_rtx.
36541 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36542 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36544 2017-03-09 Richard Biener <rguenther@suse.de>
36546 PR tree-optimization/79977
36547 * graphite-scop-detection.c (scop_detection::merge_sese):
36548 Handle the case of extra exits to blocks dominating the entry.
36550 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36552 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36555 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36557 PR rtl-optimization/79949
36558 * lra-constraints.c (process_alt_operands): Check memory when
36559 trying to predict a cycle. Print about the overall increase.
36561 2017-03-09 Richard Biener <rguenther@suse.de>
36563 PR middle-end/79971
36564 * gimple-expr.c (useless_type_conversion_p): Preserve
36565 TYPE_SATURATING for fixed-point types.
36567 2017-03-09 Richard Biener <rguenther@suse.de>
36570 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36571 alignment of BLKmode params.
36573 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36576 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36577 (VALL_NO_V2Q): Likewise.
36579 * config/aarch64/aarch64-simd.md
36580 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36582 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36583 VALL_NO_V2Q mode iterator.
36584 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36586 2017-03-09 Martin Liska <mliska@suse.cz>
36588 PR tree-optimization/79631
36589 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36590 tree_int_cst_sign_bit just for INTEGER constants.
36592 2017-03-09 Martin Liska <mliska@suse.cz>
36596 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36599 2017-03-09 Marek Polacek <polacek@redhat.com>
36602 * tree.c (inchash::add_expr): Handle TREE_VEC.
36604 2017-03-09 Martin Liska <mliska@suse.cz>
36607 (chkp_narrow_size_and_offset): New function.
36608 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36609 (void chkp_parse_bit_field_ref): New function.
36610 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36611 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36613 2017-03-09 Martin Liska <mliska@suse.cz>
36616 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36617 (chkp_find_bounds_1): Remove gcc_unreachable.
36619 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36622 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36623 BUILT_IN_SYNC*, determine the access type from the size suffix and
36624 always build a MEM_REF with that type. Handle forgotten
36625 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36628 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36629 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36630 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36631 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36632 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36633 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36634 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36635 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36636 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36637 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36638 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36639 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36640 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36641 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36642 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36643 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36644 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36645 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36646 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36647 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36648 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36649 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36650 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36651 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36652 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36653 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36654 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36655 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36656 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36657 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36658 definitions outside of __OPTIMIZE__ guarded section.
36661 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36662 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36663 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36664 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36667 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36669 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36670 ("vfenez<mode>"): Add missing constraints.
36672 2017-03-08 Martin Sebor <msebor@redhat.com>
36675 * config/nds32/nds32.c (nds32_option_override):
36676 Fix misspelled diagnostic.
36678 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36681 * gimplify.c (gimplify_omp_for): Replace index var in outer
36682 taskloop statement with an artificial variable and add
36683 OMP_CLAUSE_PRIVATE clause for it.
36685 2017-03-08 Richard Biener <rguenther@suse.de>
36687 PR tree-optimization/79955
36688 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36689 for accesses that are completely outside of the variable.
36691 2017-03-08 Andrew Haley <aph@redhat.com>
36693 PR tree-optimization/79943
36694 * tree-ssa-loop-split.c (compute_new_first_bound): When
36695 calculating the new upper bound, (END-BEG) should be added, not
36698 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36700 * config/avr/avr.md (setmemhi): Make sure match_dup
36701 operand number comes before match_scratch.
36703 2017-03-08 Richard Biener <rguenther@suse.de>
36705 PR tree-optimization/79920
36706 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36707 with ncopies == 1 to ...
36708 (vect_transform_slp_perm_load): ... here. Properly compute
36709 all element loads by iterating VF times over the group. Do
36710 not handle ncopies (computed in a broken way) in
36711 vect_create_mask_and_perm.
36713 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36716 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36717 is a uniform vector, use uniform_vector_p return value instead of
36718 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36720 2017-03-07 Marek Polacek <polacek@redhat.com>
36722 PR middle-end/79809
36723 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36724 (alloca_call_type): Likewise.
36726 2017-03-07 Martin Liska <mliska@suse.cz>
36728 * gcov.c (process_args): Put comment to correct location.
36730 2017-03-07 Martin Liska <mliska@suse.cz>
36732 PR middle-end/68270
36733 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36734 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36735 (chkp_narrow_bounds_for_field): Likewise.
36736 (chkp_parse_array_and_component_ref): Pass one more argument to
36739 2017-03-07 Richard Biener <rguenther@suse.de>
36741 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36744 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36746 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36747 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36749 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36752 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36753 to end of description.
36754 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36756 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36758 PR rtl-optimization/79901
36759 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36761 (*avx512f_<code><mode>3<mask_name>): ... this.
36762 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36763 iterator instead of VI8_AVX2_AVX512BW.
36765 PR rtl-optimization/79901
36766 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36767 min/max expander, expand it using expand_vec_cond_expr.
36770 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36773 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36776 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36777 to void * for PCH reasons.
36778 * dwarf2out.c (output_loc_operands, output_die): Cast
36779 v.val_vec.array to unsigned char *.
36781 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36784 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36787 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36789 PR rtl-optimization/79571
36790 * lra-constraints.c (process_alt_operands): Calculate static
36791 reject and subtract it from overall when only addresses will be
36794 2017-03-06 Julia Koval <julia.koval@intel.com>
36797 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
36798 incoming stack boundary to 128 for 64-bit targets.
36800 2017-03-06 Richard Biener <rguenther@suse.de>
36802 PR tree-optimization/79894
36803 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
36804 to NULL after folding it.
36806 2017-03-06 Richard Biener <rguenther@suse.de>
36808 PR tree-optimization/79824
36809 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
36810 check disabling peeling for gaps.
36812 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
36814 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
36815 attributes): Document gettimeofday.
36817 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36819 * config/s390/s390.c (s390_option_override_internal): Set
36820 PARAM_MIN_VECT_LOOP_BOUND
36822 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36824 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36825 * config/s390/s390.md: Likewise.
36827 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36830 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36831 (<avx2_avx512>_perm<mode>): Rename to ...
36832 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36834 (<avx512>_perm<mode>_mask): Rename to ...
36835 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36837 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36838 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36839 instead of VI8F_256_512.
36840 (avx512f_perm<mode>): New define_expand.
36841 (avx512f_perm<mode>_mask): Likewise.
36842 (avx512f_perm<mode>_1<mask_name>): New define_insn.
36843 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
36845 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36847 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
36848 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
36850 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36852 2017-03-06 Martin Liska <mliska@suse.cz>
36855 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
36856 when having a SSA NAME w/o VAR_DECL assigned to it.
36858 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36860 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
36861 msa_dpsub_<su>_d): Fix MODE for vec_select.
36863 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36865 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
36867 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
36869 2017-03-06 Richard Biener <rguenther@suse.de>
36871 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
36872 * plugin.c (register_plugin_info): Likewise.
36873 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36875 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36877 * config/i386/sse.md (sse_storehps, sse_storelps,
36878 avx_<castmode><avxsizesuffix>_<castmode>,
36879 avx512f_<castmode><avxsizesuffix>_<castmode>,
36880 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36881 in condition that at least one operand is not a MEM.
36883 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36885 PR middle-end/79805
36886 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36887 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36889 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36890 gimple_call_nothrow_p flag based on whether original builtin can throw.
36891 If it can, emit following stmts on the fallthrough edge.
36892 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36893 don't create new bb if inserting just debug stmts on the edge, try to
36894 insert them on the fallthru bb or just reset debug stmts.
36896 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36899 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36900 restore recog_data (including the operand rtxes inside it) around
36901 the call to get_insn_template.
36903 2017-03-03 Martin Sebor <msebor@redhat.com>
36905 PR tree-optimization/79699
36906 * context.c (context::~context): Free MPFR caches to avoid
36907 a memory leak on program exit.
36909 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36911 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36912 Use wide_int::ulow () instead of .elt (0).
36914 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36916 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36917 (*pushxf): Limit oF constraint to 32bit targets and add oC
36918 constraint for 64bit targets.
36919 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36920 (*pushdf): Change rmF constraint to rmC.
36922 2017-03-03 Martin Liska <mliska@suse.cz>
36924 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36925 Remove unused variable.
36927 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36930 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36931 is a memory operand, increase num_memory.
36932 (ix86_expand_args_builtin): Likewise.
36934 2017-03-03 Jan Hubicka <jh@suse.cz>
36937 * ipa-devirt.c (maybe_record_node): Properly handle
36938 __cxa_pure_virtual visibility.
36940 2017-03-03 Martin Liska <mliska@suse.cz>
36942 PR tree-optimization/79803
36943 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36945 (pass_loop_prefetch::execute): Disabled optimization if an
36946 assumption about L1 cache size is not met.
36948 2017-03-03 Martin Liska <mliska@suse.cz>
36950 PR rtl-optimization/79574
36951 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36952 (hash_scan_set): Likewise.
36953 (dump_hash_table): Likewise.
36954 (hoist_code): Likewise.
36956 2017-03-03 Richard Biener <rguenther@suse.de>
36958 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36959 in place of uhigh ().
36960 (fixed_convert_from_real): Likewise.
36962 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36965 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36967 2017-03-03 Richard Biener <rguenther@suse.de>
36969 PR middle-end/79818
36970 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36971 TYPE_OVERFLOW_UNDEFINED check.
36973 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36975 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36977 (vector_ae_<mode>_p): Likewise.
36978 (vector_nez_<mode>_p): Likewise.
36979 (vector_ne_v2di_p): Likewise.
36980 (vector_ae_v2di_p): Likewise.
36981 (vector_ne_<mode>_p): Likewise.
36982 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36984 (vsx_tsqrt<mode>2_fe): Likewise.
36986 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36989 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36991 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36993 PR rtl-optimization/79780
36994 * cprop.c (one_cprop_pass): When second and further conditional trap
36995 in a single basic block is turned into an unconditional trap, turn it
36996 into a deleted note to avoid RTL verification failures.
36998 2017-03-02 Richard Biener <rguenther@suse.de>
37000 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37002 2017-03-02 Richard Biener <rguenther@suse.de>
37004 PR tree-optimization/79345
37006 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37007 param and abort the walk, returning -1 if it is hit.
37008 (walk_aliased_vdefs): Take a limit param and pass it on.
37009 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37010 defaulting to 0 and return a signed int.
37011 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37012 (check_defs): New helper.
37013 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37014 about uninitialized memory.
37015 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37016 bogus uninitialized warning.
37017 (fixed_convert_from_real): Likewise.
37019 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37021 PR tree-optimization/66768
37022 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37023 iv_use if base object can't be determined.
37025 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37027 PR tree-optimization/79345
37028 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37029 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37030 (get_pattern_stats): Initialize it.
37031 * genemit.c (gen_expand): Verify match_scratch numbers come after
37032 match_operand/match_dup numbers.
37033 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37034 match_scratch numbers.
37035 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37037 * config/s390/s390.md (trunctdsd2): Likewise.
37039 2017-03-02 Richard Biener <rguenther@suse.de>
37041 * wide-int.h (wide_int_storage::operator=): Implement in terms
37044 2017-03-02 Richard Biener <rguenther@suse.de>
37046 PR tree-optimization/79777
37047 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37048 the to insert expression to sth existing.
37050 2017-03-01 Martin Sebor <msebor@redhat.com>
37052 PR middle-end/79692
37053 * gimple-ssa-sprintf.c
37054 (directive::known_width_and_precision): New function.
37055 (format_integer): Use it.
37056 (get_mpfr_format_length): Consider the full range of precision
37057 when computing %g output with the # flag. Set the likely byte
37058 count to 3 rather than 1 when precision is indeterminate.
37059 (format_floating): Correct the lower bound of precision.
37061 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37063 * doc/invoke.texi: Document default code model for 64-bit Linux.
37065 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37068 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37069 udiv rather than div since input pattern is unsigned.
37071 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37073 * config/i386/i386.c (print_reg): Warn for values of
37074 unsupported size in integer register.
37076 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37079 * config/rs6000/predicates.md (current_file_function_operand): Do
37080 not allow self calls to be local if the function is replaceable.
37082 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37085 * config/rs6000/altivec.h (vec_ctz and others): Change the
37086 preprocessor macro that controls conditional compilation from
37087 _ARCH_PWR9 to __POWER9_VECTOR__.
37088 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37089 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37090 control (instead of _ARCH_PWR9 control) so that template
37091 definition uses power9-specific function.
37092 (vec_any_eq): Likewise.
37093 (vec_all_ne): Change macro definition to use a power9-specific
37094 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37095 _ARCH_PWR9 control).
37096 (vec_any_eq) Likewise.
37097 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37098 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37099 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37100 support for xvcmpnedp instruction.
37101 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37102 macro expansion so that Power9 implementation of vec_all_ne does
37103 not use the AltiVec predicate framework.
37104 (VCMPNEH_P): Likewise.
37105 (VCMPNEW_P): Likewise.
37106 (VCMPNED_P): Likewise.
37107 (VCMPNEFP_P): Likewise.
37108 (VCMPNEDP_P): Likewise.
37109 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37110 implementation of vec_any_eq to not use AltiVec predicate
37112 (VCMPAEH_P): Likewise.
37113 (VCMPAEW_P): Likewise.
37114 (VCMPAED_P): Likewise.
37115 (VCMPAEFP_P): Likewise.
37116 (VCMPAEDP_P): Likewise.
37117 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37118 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37119 not use the AltiVec predicate framework.
37120 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37121 of vec_any_eq to not use AltiVec predicate framework.
37122 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37123 support for predefined __POWER9_VECTOR__ macro to indicate that
37124 Power9 instruction selection is enabled.
37125 (altivec_overloaded_builtins): Remove extraneous
37126 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37127 function argument types RS6000_BTI_bool_V16QI and
37128 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37129 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37130 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37131 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37132 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37133 Power9 for implementations of vec_cmpne. Change the signature for
37134 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37135 (representing vec_all_ne) to remove the previously described first
37136 argument of type RS6000_BTI_INTSI, as this was an artifact of
37137 reliance on the AltiVec predicate framework, which is no longer
37138 used in the implementation of these functions. Add
37139 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37140 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37141 since, unlike the AltiVec predicate framework implementation, we
37142 do not share function descriptors between vec_alle and vec_anyeq.
37143 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37144 set of modes that receive special treatment even when
37145 TARGET_P9_VECTOR is true. The special treatment emits code that
37146 does not depend on Power9 instructions.
37147 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37148 define_expand to not rely on AltiVec predicate framework.
37149 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37151 (vector_ne_v2di_p): Change this define_expand to not rely on
37152 AltiVec predicate framework.
37153 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37155 (vector_ne_<mode>_p): Change this define_expand to not rely on
37156 AltiVec predicate framework.
37157 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37159 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37160 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37161 define_insn pattern.
37162 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37163 define_insn pattern because the xvcmpne<VSs>. instruction is not
37165 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37166 instruction is not supported.
37168 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37170 * config/nvptx/nvptx.c: Include intl.h.
37172 2017-03-01 Martin Jambor <mjambor@suse.cz>
37175 * ipa-prop.h (ipa_bits): Removed field known.
37176 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37177 to pointers. Adjusted their comments to warn about their sharing.
37178 (ipcp_transformation_summary): Change bits to a vector of pointers.
37179 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37180 (ipa_get_ipa_bits_for_value): Declare.
37181 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37182 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37183 (ipa_bits_hash_table): Likewise.
37184 (ipa_vr_ggc_hash_traits): Likewise.
37185 (ipa_vr_hash_table): Likewise.
37186 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37187 being pointers and vr_known being removed.
37188 (ipa_set_jf_unknown): Likewise.
37189 (ipa_get_ipa_bits_for_value): New function.
37190 (ipa_set_jfunc_bits): Likewise.
37191 (ipa_get_value_range): New overloaded functions.
37192 (ipa_set_jfunc_vr): Likewise.
37193 (ipa_compute_jump_functions_for_edge): Use the above functions to
37194 construct bits and vr parts of jump functions.
37195 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37196 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37198 (ipcp_grow_transformations_if_necessary): Also allocate
37199 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37201 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37202 them. Fix too long lines.
37203 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37204 vr_known being removed.
37205 (ipa_read_jump_function): Use new setter functions to construct bits
37206 and vr parts of jump functions or set them to NULL.
37207 (write_ipcp_transformation_info): Adjust for bits being pointers.
37208 (read_ipcp_transformation_info): Likewise.
37209 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37211 Include gt-ipa-prop.h.
37212 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37214 (ipcp_store_bits_results): Likewise.
37215 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37216 Do not write to existing jump functions but use a temporary instead.
37218 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37221 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37222 attempt to use its first operand as BIT_FIELD_REF base.
37224 2017-03-01 Richard Biener <rguenther@suse.de>
37226 PR middle-end/79721
37227 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37228 interpolating formula in wrapping arithmetic.
37229 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37231 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37233 PR tree-optimization/79734
37234 * tree-vect-generic.c (expand_vector_condition): Optimize
37235 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37236 Handle VEC_COND_EXPR where comparison has different inner width from
37237 type's inner width.
37239 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37241 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37242 markup, and similar issues. Remove @opindex entries for things
37243 that aren't options. Add missing -mmpy-option entries.
37245 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37247 PR tree-optimization/79737
37248 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37249 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37250 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37251 instead of byte_size. Formatting fix.
37252 (shift_bytes_in_array_right): Formatting fix.
37254 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37257 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37258 condition on optimize for the leaf function test.
37260 2017-02-28 Martin Liska <mliska@suse.cz>
37263 * read-rtl-function.c (function_reader::handle_unknown_directive):
37264 Bail out when one uses -flto.
37266 2017-02-28 Martin Liska <mliska@suse.cz>
37268 * common.opt: Replace space with tabular for options of <number>
37270 * config/i386/i386.opt: Show <number> value for
37271 -mlarge-data-threshold.
37272 * opts.c (print_filtered_help): Do not display number in hexadecimal
37275 2017-02-28 Martin Liska <mliska@suse.cz>
37277 * common.opt: Fix --help=option -Q for options which are of
37280 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37282 * config/i386/i386.c (print_reg): Error out for values
37283 of 8-bit size in invalid integer register.
37285 2017-02-28 Martin Sebor <msebor@redhat.com>
37287 PR tree-optimization/79691
37288 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37290 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37293 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37294 gcc_unreachable with output_operand_lossage.
37296 2017-02-28 Richard Biener <rguenther@suse.de>
37298 PR tree-optimization/79740
37299 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37301 (visit_nary_op): Insert the nary into the hashtable if we
37302 pattern-matched sth.
37303 * tree-ssa-pre.c (eliminate_insert): Robustify.
37305 2017-02-28 Richard Biener <rguenther@suse.de>
37307 PR middle-end/79731
37308 * fold-const.c (decode_field_reference): Reject out-of-bound
37311 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37313 * config/i386/i386.c: Include intl.h.
37314 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37315 instead of just cond ? "..." : "...".
37316 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37317 * coverage.c (read_counts_file): Likewise.
37318 * omp-offload.c: Include intl.h.
37319 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37320 of just cond ? "..." : "...".
37321 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37322 of just cond ? "..." : "...".
37324 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37327 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37329 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37330 'tune for' CPU name.
37331 * config/arm/arm-cpu-data.h: Regenerated.
37333 2017-02-28 Richard Biener <rguenther@suse.de>
37335 PR tree-optimization/79732
37336 * tree-inline.c (expand_call_inline): Do not shadow var.
37338 2017-02-28 Richard Biener <rguenther@suse.de>
37340 PR tree-optimization/79723
37341 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37342 address-space properly.
37344 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37346 * doc/optinfo.texi (Optimization groups): Fix option used for
37348 * doc/invoke.texi (-fopt-info): Document "omp".
37349 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37350 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37351 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37352 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37353 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37355 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37357 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37358 OPTGROUP_OMP with "omp".
37360 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37363 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37364 for arithmetic shift of unsigned V2DI.
37366 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37368 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37369 arc/linux.h headers.
37370 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37371 (LINK_SPEC): Likewise.
37372 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37373 (EXTRA_SPECS): Likewise.
37374 (STARTFILE_SPEC): Likewise.
37375 (ENDFILE_SPEC): Likewise.
37376 (LIB_SPEC): Likewise.
37377 (TARGET_SDATA_DEFAULT): Likewise.
37378 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37379 (MULTILIB_DEFAULTS): Likewise.
37380 (DWARF2_UNWIND_INFO): Likewise.
37381 * config/arc/big.h: New file.
37382 * config/arc/elf.h: Likewise.
37383 * config/arc/linux.h: Likewise.
37384 * config/arc/t-uClibc: Remove.
37386 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37388 PR tree-optimization/77536
37389 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37390 (tree_transform_and_unroll_loop): Use above function to compute the
37391 estimated niter of unrolled loop and use it when scaling profile.
37392 Also use count info rather than frequency if it's non-zero.
37393 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37394 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37395 (vect_transform_loop): Call above function.
37397 2017-02-27 Richard Biener <rguenther@suse.de>
37399 PR tree-optimization/45397
37400 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37401 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37402 (visit_nary_op): Add pattern matching for CSEing sign-changed
37403 or truncated operations with wider ones.
37405 2017-02-27 Richard Biener <rguenther@suse.de>
37407 PR tree-optimization/79690
37408 * tree-vect-stmts.c (vectorizable_store): Use vector type
37409 built from the DR with address-space.
37411 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37413 * doc/invoke.texi (Optimize Options): Refine the description
37414 of asan-use-after-return.
37416 2017-02-25 Alan Modra <amodra@gmail.com>
37418 PR rtl-optimization/79584
37419 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37420 base, not ad->base_term, the reg within base. Remove assertion
37421 that ad->base == ad->base_term. Replace gen_int_mode using
37422 bogus mode with const0_rtx.
37424 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37426 PR middle-end/79396
37427 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37428 FMA_EXPR like tcc_binary or tcc_unary.
37430 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37433 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37435 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37436 (output_loc_operands): Handle DW_OP_call_ref and
37437 DW_OP_GNU_variable_value.
37438 (struct variable_value_struct): New type.
37439 (struct variable_value_hasher): Likewise.
37440 (variable_value_hash): New variable.
37441 (string_types): Remove.
37442 (copy_loc_descr): New function.
37443 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37444 (prepend_loc_descr_to_each): New function.
37445 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37446 instead of add_loc_descr_to_each if the first argument is single
37447 location list and the second has multiple.
37448 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37449 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37450 when looking for variable value which doesn't have other location info.
37451 (loc_list_from_tree): Formatting fix.
37452 (gen_array_type_die): Simplify DW_AT_string_length handling.
37453 (adjust_string_types): Remove.
37454 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37455 string_types. Call resolve_variable_values.
37456 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37457 (resolve_addr_in_expr): Likewise. Add A argument.
37458 (copy_deref_exprloc): Remove deref argument. Adjust for the
37459 original expression being DW_OP_GNU_variable_value with optionally
37460 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37461 optionally after it.
37462 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37463 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37464 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37465 (variable_value_hasher::hash, variable_value_hasher::equal): New
37467 (resolve_variable_value_in_expr, resolve_variable_value,
37468 resolve_variable_values, note_variable_value_in_expr,
37469 note_variable_value): New functions.
37470 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37473 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37476 * opts.h (handle_generated_option): Add GENERATED_P argument.
37477 * opts-common.c (handle_option): Adjust function comment.
37478 (handle_generated_option): Add GENERATED_P argument, pass it to
37480 (control_warning_option): Pass false to handle_generated_option
37482 * opts.c (maybe_default_option): Pass true to handle_generated_option
37484 * optc-gen.awk: Likewise.
37486 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37488 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37489 a REG, look at the REG it is a SUBREG of.
37490 (splitter for cmpeqsi_t): Ditto.
37492 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37494 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37495 the special USEs with the pattern of the insn, not the insn itself.
37497 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37500 * doc/invoke.texi: Document -mload-store-pairs.
37502 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37503 Sandra Loosemore <sandra@codesourcery.com>
37505 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37506 argument isn't a CONST_INT.
37507 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37508 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37509 (nios2_validate_compare): Bypass alternate compare logic if *op2
37510 is not a CONST_INT.
37511 (ldstwm_operation_p): Return false if first_base is not a REG or
37512 if first_offset is not a CONST_INT.
37514 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37516 * config/cris/cris.md: Use correct operand in a define_peephole2.
37518 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37520 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37522 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37524 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37525 this_insn if it is an INSN or JUMP_INSN.
37526 (force_offsettable): Look at base, not at addr.
37527 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37528 on things that aren't necessarily CONST_INTs.
37530 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37532 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37533 -mfpmath=sse is the default also for x86-32 targets with SSE2
37534 instruction set when @option{-ffast-math} is enabled
37536 2017-02-24 Jeff Law <law@redhat.com>
37538 PR rtl-optimizatoin/79286
37539 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37542 2017-02-24 Richard Biener <rguenther@suse.de>
37544 PR tree-optimization/79389
37545 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37548 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37550 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37551 function comment to reflect reality.
37552 (loop_exits_before_overflow): Fix typo in function description.
37554 2017-02-24 Richard Biener <rguenther@suse.de>
37556 PR tree-optimization/79389
37557 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37558 properly that a threading opportunity exists. Detect conditional
37559 copy/constant propagation opportunities.
37561 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37563 * config/visium/visium.md (type): Add trap.
37564 (b): New mode attribute.
37565 (*btst): Rename into...
37566 (*btst<mode>): ...this and adjust.
37567 (*cbranchsi4_btst_insn): Rename into...
37568 (*cbranch<mode>4_btst_insn): ...this and adjust.
37569 (trap): New define_insn.
37571 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37573 PR tree-optimization/79389
37574 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37575 (noce_reversed_cond_code): New function.
37576 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37577 reversed_comparison_code. Formatting fix.
37578 (noce_try_store_flag): Test rev_cond != NULL in addition to
37579 reversed_comparison_code.
37580 (noce_try_store_flag_constants): Likewise.
37581 (noce_try_store_flag_mask): Likewise.
37582 (noce_try_addcc): Use rev_cond if non-NULL instead of
37583 reversed_comparison_code.
37584 (noce_try_cmove_arith): Likewise. Formatting fixes.
37585 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37586 (noce_find_if_block): Initialize rev_cond.
37587 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37588 instead of false as last argument never attempt to reverse it
37591 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37593 PR tree-optimization/79663
37594 * tree-predcom.c (combine_chains): Process refs in reverse order
37595 only for ZERO length chains, and add explaining comment.
37597 2017-02-23 Jeff Law <law@redhat.com>
37599 PR tree-optimization/79578
37600 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37601 in call to operand_equal_p.
37603 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37606 * config/i386/cpuid.h: Fix another undefined behavior.
37608 2017-02-23 Richard Biener <rguenther@suse.de>
37610 PR tree-optimization/79683
37611 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37612 vector types for data-refs.
37614 2017-02-23 Martin Liska <mliska@suse.cz>
37616 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37618 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37620 PR middle-end/79665
37621 * internal-fn.c (get_range_pos_neg): Moved to ...
37622 * tree.c (get_range_pos_neg): ... here. No longer static.
37623 * tree.h (get_range_pos_neg): New prototype.
37624 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37625 are known to be in between 0 and signed maximum inclusive, try to
37626 expand both unsigned and signed divmod and use the cheaper one from
37629 2017-02-22 Jeff Law <law@redhat.com>
37631 PR tree-optimization/79578
37632 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37633 to compare base operands.
37635 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37638 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37639 gpc_reg_operand instead of fpr_reg_operand.
37641 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37643 * config/mips/mips.c (mips_return_in_memory): Force FP
37644 vector types to be returned in memory for o32 ABI.
37646 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37648 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37649 instead of DW_TAG_member for static data member declarations and don't
37650 set no_linkage_name for static inline data members.
37651 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37654 2017-02-22 Martin Liska <mliska@suse.cz>
37656 * doc/invoke.texi: Replace inequality signs with square brackets
37659 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37662 * lra-constraints.c (simplify_operand_subreg): Handle
37663 WORD_REGISTER_OPERATIONS targets.
37665 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37668 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37669 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37670 elimination by swapping fld*.
37672 2017-02-22 Richard Biener <rguenther@suse.de>
37674 PR tree-optimization/79673
37675 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37676 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37677 irrelevant address-space qualifiers and avoiding a
37678 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37680 2017-02-22 Richard Biener <rguenther@suse.de>
37682 PR tree-optimization/79666
37683 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37684 to not symbolically negate if that may introduce undefined
37687 2017-02-22 Martin Liska <mliska@suse.cz>
37690 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37691 * data-streamer-out.c (streamer_write_gcov_count_stream):
37693 * value-prof.c (stream_out_histogram_value): Make assert more
37694 precise based on type of counter.
37696 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37699 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37700 Use nonimmediate_operand instead of memory_operand for operand 1.
37701 (float-extend standard_x87sse_constant_load splitter): Ditto.
37703 2017-02-21 Jeff Law <law@redhat.com>
37705 PR tree-optimization/79621
37706 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37707 blocks with edges to themselves.
37709 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37712 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37713 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37714 Use gimple_call_builtin_p.
37717 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37718 on temporarily removed DEBUG_INSNs.
37720 PR tree-optimization/79649
37721 * tree-loop-distribution.c (classify_partition): Give up on
37722 non-generic address space loads/stores.
37724 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37726 * doc/loop.texi (Loop manipulation): Remove nonexistent
37727 tree_ssa_loop_version from the documentation.
37728 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37730 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37733 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37734 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37735 * config/rs6000/rs6000.c: Include except.h.
37736 (rs6000_expand_split_stack_prologue): Call
37737 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37739 2017-02-21 Martin Jambor <mjambor@suse.cz>
37742 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37743 have been analyzed.
37745 2017-02-21 Martin Jambor <mjambor@suse.cz>
37747 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37748 for backward compatibility only.
37749 * doc/invoke.texi (Option Summary): Remove all references to
37750 -fipa-cp-alignment.
37752 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37756 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37758 * lra-constraints.c (curr_insn_transform): Handle
37759 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37761 2017-02-21 Martin Liska <mliska@suse.cz>
37763 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37765 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37767 PR translation/79638
37768 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37770 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37773 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37774 (arm_function_ok_for_sibcall): Return false for an indirect call by
37775 descriptor if all the argument registers are used.
37776 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37777 alignment of the function.
37779 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37781 PR tree-optimization/61441
37782 * simplify-rtx.c (simplify_const_unary_operation): For
37783 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37784 the sNaN unmodified.
37786 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37788 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37789 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
37790 instead of SYSTEM_HEADER_DIR.
37792 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
37793 Martin Liška <mliska@suse.cz>
37795 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
37796 Fix typos and grammar, use active voice, and clarify.
37798 2017-02-20 Marek Polacek <polacek@redhat.com>
37800 PR middle-end/79537
37801 * gimplify.c (gimplify_expr): Handle unused *&&L;.
37804 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
37806 2017-02-20 Jakub Jelinek <jakub@redhat.com>
37809 * config/i386/i386.c (ix86_expand_builtin): Handle
37810 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
37811 ix86_builtins_isa[fcode].isa as a requirement of those
37812 flags and any other flag in the bitmask.
37813 (ix86_init_mmx_sse_builtins): Use 0 instead of
37814 ~OPTION_MASK_ISA_64BIT as mask.
37815 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
37816 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
37817 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37818 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37820 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37823 * lra-constraints.c (split_reg): Check requested split mode
37824 is supported by the register.
37826 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37828 * lra-constraints.c (simplify_operand_subreg): Remove early
37831 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37834 * lra-constraints.c (curr_insn_transform): Tighten condition
37835 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37837 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37840 * lra-constraints.c (curr_insn_transform): Handle
37841 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37843 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
37846 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
37848 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
37850 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37853 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
37856 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37858 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
37861 2017-02-18 Jakub Jelinek <jakub@redhat.com>
37863 * final.c (last_columnnum, override_columnnum): New variables.
37864 (final_start_function): Set last_columnnum, pass it to begin_prologue
37865 hook and pass 0 to dwarf2out_begin_prologue.
37866 (final_scan_insn): Update override_columnnum. Pass last_columnnum
37867 to source_line debug hook.
37868 (notice_source_line): Compute last_columnnum and for debug_column_info
37869 return true on column changes.
37870 * debug.h (struct gcc_debug_hooks): Add column argument to
37871 source_line and begin_prologue hooks.
37872 (debug_nothing_int_charstar_int_bool): Remove prototype.
37873 (debug_nothing_int_int_charstar,
37874 debug_nothing_int_int_charstar_int_bool): New prototypes.
37875 (dwarf2out_begin_prologue): Add column argument.
37876 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37877 begin_prologue hooks.
37878 (debug_nothing_int_charstar_int_bool): Remove.
37879 (debug_nothing_int_int_charstar,
37880 debug_nothing_int_int_charstar_int_bool): New functions.
37881 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37882 through to dwarf2out_source_line.
37883 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37884 (dwarf2out_source_line): Add column argument, emit it if requested.
37885 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37887 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37888 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37889 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37890 through to dwarf2out_begin_prologue.
37891 (vmsdbgout_source_line): Add column argument, pass it through to
37892 dwarf2out_source_line.
37893 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37894 dbxout_source_line caller.
37895 (dbxout_source_line): Add column argument.
37897 * common.opt (gno-column-info, gcolumn-info): New options.
37898 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37899 (check_die): Also test for multiple DW_AT_decl_column attributes.
37900 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37901 DW_AT_decl_column if requested.
37902 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37904 (gen_variable_die): Likewise.
37905 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37906 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37909 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37910 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37911 (ix86_handle_option): Handle OPT_m3dnowa.
37912 * doc/invoke.texi (-m3dnowa): Document.
37913 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37914 -m3dnowa instead of -m3dnow -march=athlon.
37917 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37918 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37920 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37923 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37924 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37925 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37926 generator for vsx_xxpermdi_<mode>_be.
37927 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37928 force big-endian semantics.
37929 (vsx_xxpermdi_<mode>_be): New define_expand with same
37930 implementation as previous version of vsx_xxpermdi_<mode>.
37932 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37934 PR tree-optimization/79327
37935 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37936 variable, its initialization and use.
37938 2017-02-17 Julia Koval <julia.koval@intel.com>
37940 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37941 (OPTION_MASK_ISA_PKU_UNSET): New.
37942 (ix86_handle_option): Handle -mrdpid.
37943 * config/i386/cpuid.h (bit_RDPID): New.
37944 * config/i386/driver-i386.c (host_detect_local_cpu):
37945 Detect RDPID feature.
37946 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37947 * config/i386/i386-c.c (ix86_target_macros_internal):
37949 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37950 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37951 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37952 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37953 * config/i386/i386.md (define_insn "rdpid"): New.
37954 * config/i386/i386.opt Add -mrdpid.
37955 * config/i386/immintrin.h (_rdpid_u32): New.
37957 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37959 PR rtl-optimization/79541
37960 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37961 instead of transforming it into USE.
37963 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37965 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37966 If HONOR_SNANS (SFmode) force the input to a register.
37967 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37968 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37969 an frsp or similar insn.
37971 2017-02-17 Martin Liska <mliska@suse.cz>
37973 PR rtl-optimization/79577
37974 * params.def (selsched-max-sched-times): Increase minimum to 1.
37976 2017-02-17 Martin Liska <mliska@suse.cz>
37978 PR rtl-optimization/79574
37979 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37981 2017-02-17 Martin Liska <mliska@suse.cz>
37983 PR tree-optimization/79529
37984 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37985 ssa_defined_default_def_p to handle cases which are implicitly
37987 * tree-ssa.c (ssa_defined_default_def_p): New function.
37988 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37989 which are implicitly defined.
37990 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37992 2017-02-17 Richard Biener <rguenther@suse.de>
37994 PR middle-end/79576
37995 * params.def (max-ssa-name-query-depth): Limit to 10.
37997 2017-02-17 Richard Biener <rguenther@suse.de>
37999 PR tree-optimization/79552
38000 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38003 2017-02-17 Richard Biener <rguenther@suse.de>
38006 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38008 2017-02-17 Marek Polacek <polacek@redhat.com>
38010 PR middle-end/79536
38011 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38012 (fold_negate_expr): New wrapper.
38014 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38016 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38017 Correct terminology and de-emphasize pre-standard behavior.
38019 2017-02-16 Alan Modra <amodra@gmail.com>
38021 PR rtl-optimization/79286
38022 * ira.c (def_dominates_uses): New function.
38023 (update_equiv_regs): Don't create an equivalence for insns that
38024 may trap where the register def does not dominate the use.
38026 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38028 PR rtl-optimization/78127
38029 * lra.c (lra): Call lra_eliminate before finish the loop after
38032 2017-02-16 Richard Biener <rguenther@suse.de>
38034 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38036 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38037 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38038 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38039 (isl_val_int_from_wi): New function.
38040 (extract_affine_gmp): Rename to ...
38041 (extract_affine_wi): ... this, take a widest_int.
38042 (extract_affine_int): Just wrap extract_affine_wi.
38043 (add_param_constraints): Use isl_val_int_from_wi.
38044 (add_loop_constraints): Likewise, and extract_affine_wi.
38046 2017-02-15 Jeff Law <law@redhat.com>
38048 PR middle-end/79521
38049 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38050 ira_init_register_move_cost_if_necessary.
38052 2017-02-15 Martin Sebor <msebor@redhat.com>
38054 PR middle-end/32003
38055 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38056 removed in a prior commit.
38058 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38060 PR tree-optimization/79347
38061 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38062 counters during peeling.
38064 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38066 * Makefile.in (site.exp): Remove "set ISLVER".
38068 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38071 * real.c (real_from_integer): Call real_convert even for decimal.
38073 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38076 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38078 2017-02-14 Andrew Pinski <apinski@cavium.com>
38080 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38081 cores and change the partno/implementer to be correct.
38082 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38083 the 'B" as the implementer.
38084 * config/aarch64/aarch64-tune.md: Regenerate.
38086 2017-02-14 Carl Love <cel@us.ibm.com>
38088 * config/rs6000/rs6000.c: Add case statement entry to make the
38089 xvcvuxdsp built-in argument unsigned.
38090 * config/rs6000/vsx.md: Fix the source and return operand types so they
38091 match the instruction definitions from the ISA document. Fix typo
38092 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38095 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38098 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38099 member early_clobber_alts.
38100 * lra-lives.c (reg_early_clobber_p): New.
38101 (process_bb_lives): Use it.
38102 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38103 (debug_operand_data): Initialize early_clobber_alts.
38104 (setup_operand_alternative): Set up early_clobber_alts.
38105 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38106 alternatives to new_insn_reg.
38107 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38109 (lra_update_insn_regno_info): Pass the new arg.
38111 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38113 PR middle-end/79505
38114 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38115 (new_oacc_loop_raw): Don't clear already cleared fields.
38118 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38119 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38120 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38122 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38125 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38127 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38130 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38131 the extra instruction to the right place to store 128-bit constant
38134 2017-02-14 Martin Sebor <msebor@redhat.com>
38136 PR middle-end/79448
38137 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38138 warning for strings of unknown length.
38140 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38142 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38144 2017-02-14 Jeff Law <law@redhat.com>
38147 * ira-costs.c (scan_one_insn): Initialize register move costs
38148 for pseudos seen in USE/CLOBBER insns.
38150 PR tree-optimization/79095
38151 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38152 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38153 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38154 if the operands are known to be not equal, then the resulting range
38156 (intersect_ranges): If the new range is ~[0,0] and the old range is
38157 wide, then prefer ~[0,0].
38158 * tree-vrp.c (overflow_comparison_p_1): New function.
38159 (overflow_comparison_p): New function.
38160 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38161 if NAME is used in an overflow test.
38162 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38163 overflow check that can be expressed as an equality test, then adjust
38164 ops to be that equality test.
38166 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38168 * config/s390/s390-builtin-types.def: Remove flags argument.
38169 * config/s390/s390.c (s390_init_builtins): Likewise.
38171 2017-02-14 Martin Liska <mliska@suse.cz>
38173 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38174 vector. Fix trailing white spaces.
38176 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38178 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38181 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38183 PR rtl-optimization/68664
38184 * config/arm/arm.c (arm_sched_can_speculate_insn):
38185 New function. Declare prototype.
38186 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38188 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38190 PR rtl-optimization/68664
38191 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38193 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38195 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38197 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38198 max skip bytes for function, loop and jump.
38200 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38202 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38203 ABS_EXPR for gimple dump.
38205 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38208 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38210 PR tree-optimization/79408
38211 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38212 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38213 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38214 also if rhs1 is INTEGER_CST.
38216 2017-02-14 Richard Biener <rguenther@suse.de>
38218 PR middle-end/79432
38219 * tree-into-ssa.c (insert_phi_nodes): When the function can
38220 have abnormal edges rewrite SSA names with broken use-def
38221 dominance out of SSA and register them for PHI insertion.
38223 2017-02-13 Martin Sebor <msebor@redhat.com>
38225 PR middle-end/79496
38226 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38227 clearing info.nowrite flag when snprintf size argument is a range.
38229 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38231 * cprop.c (cprop_jump): Add missing space in string literal.
38232 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38233 (get_constraint_for_component_ref): Likewise.
38234 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38235 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38236 * lra-constraints.c (process_alt_operands): Likewise.
38237 * ipa-inline.c (inline_small_functions): Likewise.
38238 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38239 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38240 * trans-mem.c (diagnose_tm_1_op): Likewise.
38241 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38242 (grid_parallel_clauses_gridifiable): Likewise.
38244 * config/nvptx/mkoffload.c (process): Add space in between
38247 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38248 "MOD4_SSE_REGS" and "ALL_REGS".
38250 * spellcheck.c (test_data): Add , in between "foo" and "food".
38252 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38255 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38256 boundary crossing check and subsequent code generation agree.
38258 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38260 * config/aarch64/aarch64.c (has_memory_op): Delete.
38261 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38264 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38266 PR rtl-optimization/79388
38267 PR rtl-optimization/79450
38268 * combine.c (distribute_notes): When removing TEM_INSN for which
38269 corresponding dest has last value recorded, invalidate that last
38272 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38274 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38275 of explicit '@'. Add missing assembly comment marker on branch costs
38278 2017-02-13 Nathan Sidwell <nathan@acm.org>
38280 * gengtype-lex.l (<in_struct>): Add '/'.
38282 2017-02-13 Martin Liska <mliska@suse.cz>
38285 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38287 2017-02-13 Richard Biener <rguenther@suse.de>
38289 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38291 * configure: Re-generate.
38292 * config.in: Likewise.
38293 * graphite-dependences.c: Simplify as if
38294 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38295 * graphite-isl-ast-to-gimple.c: Likewise.
38296 * graphite-optimize-isl.c: Likewise.
38297 * graphite-poly.c: Likewise.
38298 * graphite-sese-to-poly.c: Likewise.
38299 * graphite.h: Likewise.
38300 * toplev.c: Include isl/version.h and use isl_version () for
38301 printing the ISL version.
38302 * doc/install.texi: Update ISL requirement.
38304 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38306 * doc/standards.texi (Standards): Update reference to
38309 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38311 * doc/extend.texi (Named Address Spaces): sourceware.org now
38313 * doc/install.texi (Binaries): Ditto.
38316 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38318 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38319 standard terminology "stringize"/"stringizing" throughout.
38320 * doc/cppinternals.texi: Likewise.
38322 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38324 * doc/extend.texi: Fix some spelling mistakes and typos.
38325 * doc/invoke.texi: Likewise.
38327 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38330 * params.def (inline-min-speedup) Change from 10 to 8.
38332 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38334 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38337 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38340 * ipa-inline-analysis.c (get_minimal_bb): New function.
38341 (record_modified): Use it.
38342 (remap_edge_change_prob): Handle also ancestor functions.
38344 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38346 * doc/contrib.texi (Contributors): Remove broken link into
38347 the Mauve CVS repository.
38349 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38351 PR middle-end/79454
38352 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38353 result computation whenever lhs doesn't have vector mode, not
38354 just when it has BLKmode.
38356 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38358 * doc/makefile.texi (profiledbootstrap): Refer to the
38359 installation instructions only in textual form.
38361 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38364 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38366 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38368 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38369 (Specific): Update mingw-w64 reference.
38371 (Specific): Remove broken link to Renesas RX processor.
38373 2017-02-10 Richard Biener <rguenther@suse.de>
38375 * toplev.c (process_options): Do not mention obsolete graphite
38376 options when printing sorry message about missing graphite support.
38377 Mention -floop-nest-optimize.
38379 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38381 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38382 (vtst_p16): Likewise.
38383 (vtstq_p8): Likewise.
38384 (vtstq_p16): Likewise.
38386 (vtstq_p64): Likewise.
38387 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38388 (vset_lane_p64): New.
38389 (vsetq_lane_p64): New.
38391 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38393 PR tree-optimization/79411
38394 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38395 stmt operands are SSA_NAMEs used in abnormal phis.
38396 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38399 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38402 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38405 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38407 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38409 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38411 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38415 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38416 not arbitrary TREE_CONSTANT.
38419 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38420 "omp declare target link" attribute unless is_global_var.
38421 * omp-offload.c (find_link_var_op): Likewise.
38423 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38424 Chung-Lin Tang <cltang@codesourcery.com>
38426 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38428 (gimplify_adjust_omp_clauses): Don't delete TILE.
38429 (gimplify_omp_for): Deal with TILE.
38430 * internal-fn.c (expand_GOACC_TILE): New function.
38431 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38433 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38434 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38436 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38437 avoid DIV for outermost collapse var.
38438 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38439 Remove out of date comments, fix whitespace.
38440 * omp-general.c (omp_extract_for_data): Deal with tiling.
38441 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38442 adjust OLF_DIM_BASE value.
38443 (struct omp_for_data): Add tiling field.
38444 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38445 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38446 for auto loops. Remove default auto determining, moved to
38447 oacc_loop_fixed_partitions.
38448 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38449 stmts, add e_mask field.
38450 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38451 (oacc_thread_numbers): Use oacc_dim_call.
38452 (oacc_xform_tile): New.
38453 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38454 (finish_oacc_loop): Adjust for ifns vector.
38455 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38456 add case for GOACC_TILE fns.
38457 (oacc_loop_xform_loop): Delete.
38458 (oacc_loop_process): Iterate over call list directly, and add
38459 handling for GOACC_TILE fns.
38460 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38462 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38463 vector partitioning to outer loops. Assign 2 partitions to loops
38464 when available. Add TILE handling.
38465 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38466 (execite_oacc_device_lower): Process GOACC_TILE fns,
38467 ignore unknown specs.
38468 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38469 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38470 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38472 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38474 * configure.ac (ACX_BUGURL): Update.
38475 * configure: Regenerate.
38477 2017-02-09 Richard Biener <rguenther@suse.de>
38479 PR tree-optimization/69823
38480 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38481 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38483 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38485 * config/arc/arc-c.def: Add __NPS400__ definition.
38486 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38487 (TARGET_NPS400): Define.
38489 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38491 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38493 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38494 pointer, arch_info.
38495 (arc_cpu_types): Fill the arch_info field with a pointer into the
38496 arc_arch_types table.
38497 (arc_selected_cpu): Declare.
38498 * config/arc/arc.c (arc_selected_cpu): Make global.
38499 (arc_selected_arch): Delete.
38500 (arc_base_cpu): Delete.
38501 (arc_override_options): Remove references to deleted variables,
38502 update access to arch information.
38503 (ARC_OPT): Update access to arch information.
38504 (ARC_OPTX): Likewise.
38505 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38506 (TARGET_ARC600): Update access to arch information.
38507 (TARGET_ARC601): Likewise.
38508 (TARGET_ARC700): Likewise.
38509 (TARGET_EM): Likewise.
38510 (TARGET_HS): Likewise.
38511 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38514 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38517 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38518 condition/operands for integer GE/LE/GEU/LEU operations.
38520 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38522 PR translation/79397
38523 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38526 2017-02-08 Martin Jambor <mjambor@suse.cz>
38529 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38530 whether allocation happened.
38531 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38532 nothing was allocated.
38534 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38536 PR tree-optimization/79408
38537 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38538 constant, but SSA_NAME with a known integer range, use the minimum
38539 of that range instead of op1 to determine if modulo can be replaced
38540 with its first operand.
38542 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38544 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38546 2017-02-08 Richard Biener <rguenther@suse.de>
38548 PR tree-optimization/71824
38549 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38550 Check all loops contained in the merged region.
38552 2017-02-07 Andrew Pinski <apinski@cavium.com>
38554 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38556 2017-02-07 Andrew Pinski <apinski@cavium.com>
38558 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38559 (thunderxt88): Likewise.
38560 (thunderxt81): Disable LSE and change v8.1 to v8.
38561 (thunderxt83): Likewise.
38563 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38564 Richard Biener <rguenther@suse.de>
38566 PR middle-end/79399
38567 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38568 type from int to size_t.
38569 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38571 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38573 PR rtl-optimization/79386
38574 * cprop.c (bypass_conditional_jumps): Initialize
38575 bypass_last_basic_block already before splitting bbs after
38576 unconditional traps...
38577 (bypass_conditional_jumps): ... rather than here.
38580 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38581 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38582 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38583 fix -masm=intel patterns.
38585 2017-02-07 Richard Biener <rguenther@suse.de>
38587 PR tree-optimization/79256
38588 PR middle-end/79278
38589 * builtins.c (get_object_alignment_2): Use min_align_of_type
38590 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38591 and ADJUST_FIELD_ALIGN.
38593 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38595 * doc/tm.texi: Regenerate.
38596 * stor-layout.c (layout_decl): Adjust.
38597 (update_alignment_for_field): Likewise.
38598 (place_field): Likewise.
38599 (min_align_of_type): Likewise.
38600 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38601 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38602 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38603 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38604 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38605 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38606 * config/i386/i386.c (x86_field_alignment): Likewise.
38607 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38608 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38609 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38610 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38611 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38612 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38616 2017-01-30 Richard Biener <rguenther@suse.de>
38618 PR tree-optimization/79256
38619 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38620 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38623 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38625 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38626 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38627 builtins to SImode and emit a zero-extend, if necessary.
38629 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38631 * docs/invoke.texi (RISC-V Options): Alphabetize.
38633 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38635 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38638 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38640 * config/riscv/riscv.c: New file.
38641 * common/config/riscv/riscv-common.c: Likewise.
38642 * config.gcc: Likewise.
38643 * config/riscv/constraints.md: Likewise.
38644 * config/riscv/elf.h: Likewise.
38645 * config/riscv/generic.md: Likewise.
38646 * config/riscv/linux.h: Likewise.
38647 * config/riscv/multilib-generator: Likewise.
38648 * config/riscv/peephole.md: Likewise.
38649 * config/riscv/pic.md: Likewise.
38650 * config/riscv/predicates.md: Likewise.
38651 * config/riscv/riscv-builtins.c: Likewise.
38652 * config/riscv/riscv-c.c: Likewise.
38653 * config/riscv/riscv-ftypes.def: Likewise.
38654 * config/riscv/riscv-modes.def: Likewise.
38655 * config/riscv/riscv-opts.h: Likewise.
38656 * config/riscv/riscv-protos.h: Likewise.
38657 * config/riscv/riscv.h: Likewise.
38658 * config/riscv/riscv.md: Likewise.
38659 * config/riscv/riscv.opt: Likewise.
38660 * config/riscv/sync.md: Likewise.
38661 * config/riscv/t-elf-multilib: Likewise.
38662 * config/riscv/t-linux: Likewise.
38663 * config/riscv/t-linux-multilib: Likewise.
38664 * config/riscv/t-riscv: Likewise.
38665 * configure.ac: Likewise.
38666 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38667 Waterman as RISC-V maintainers.
38668 * doc/install.texi: Add RISC-V entries.
38669 * doc/invoke.texi: Add RISC-V options section.
38670 * doc/md.texi: Add RISC-V constraints section.
38671 * configure: Regenerated.
38673 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38676 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38677 false values to be constant vectors with all 0 or all 1 bits set.
38678 (vcondu<mode><mode>): Likewise.
38679 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38681 (fpmask_comparison_operator): Update comment.
38682 (vecint_comparison_operator): New predicate.
38683 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38684 vector conditionals when the true and false values are constant
38685 vectors with all 0 bits or all 1 bits set.
38687 2017-02-06 Martin Sebor <msebor@redhat.com>
38689 PR tree-optimization/79376
38690 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38692 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38694 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38695 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38696 to simplify split condition.
38698 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38700 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38701 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38704 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38706 PR rtl-optimization/68664
38707 * target.def (can_speculate_insn): New hook.
38708 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38709 * doc/tm.texi: Regenerate.
38710 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38711 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38712 (rs6000_sched_can_speculate_insn): New function.
38714 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38716 PR tree-optimization/79284
38717 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38718 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38719 vectorizable_mask_load_store, vectorizable_operation,
38720 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38721 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38722 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38723 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38724 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38725 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38726 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38727 is_gimple_assign (stmt). Replace another such test with
38728 is_gimple_assign (stmt).
38730 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38733 * config/avr/avr.c (rtl-iter.h): Include it.
38734 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38735 (avr_legitimate_combined_insn): ...and implementation.
38737 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38739 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38740 * config/s390/s390.c (s390_const_operand_ok)
38741 (s390_canonicalize_comparison, s390_extract_part)
38742 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38743 (s390_contiguous_bitmask_p, s390_rtx_costs)
38744 (legitimize_pic_address): Likewise.
38745 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38746 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38747 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38748 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38749 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38751 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38753 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38754 REGNO($0) == REGNO($1).
38756 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38758 * config/s390/linux.h(SIZE_TYPE): Add comment.
38760 2017-02-06 Julian Brown <julian@codesourcery.com>
38761 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38762 Virendra Pathak <virendra.pathak@broadcom.com>
38764 * config/aarch64/aarch64-cores.def: Change the scheduler
38766 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38767 * config/aarch64/thunderx2t99.md: New file.
38769 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38771 * doc/standards.texi (Go Language): Update link to language
38774 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38776 * tree-eh.c (lower_resx): Sanitize profile.
38777 (cleanup_empty_eh_move_lp): Likewise.
38779 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38782 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38784 * cfgloopmanip.h (loop_version): Update prototype.
38785 * modulo-sched.c (sms_schedule): Update call of loop_version.
38786 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38787 * tree-parloops.c (gen_parallel_loop): Likewise.
38788 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38789 * tree-ssa-loop-split.c (split_loop): Likewise.
38790 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
38791 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
38793 2017-02-05 Martin Liska <mliska@suse.cz>
38796 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
38798 (print_operand_address): Initialize a struct to zero.
38800 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38802 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
38803 garbage collector only in textual form.
38805 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38807 * doc/extend.texi (x86 specific memory model extensions for
38808 transactional memory): Simplify a phrase.
38810 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
38813 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
38814 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
38815 (atomic_storedi_1): Likewise.
38817 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38819 PR tree-optimization/79338
38820 * tree-parloops.c (gather_scalar_reductions): Don't call
38821 vect_analyze_loop_form for loop->inner before destroying loop's
38824 2017-02-03 Martin Sebor <msebor@redhat.com>
38826 PR tree-optimization/79327
38827 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38828 when precision has resulted in leading zeros.
38829 (format_integer): Adjust the likely counter to assume an unknown
38830 argument that may be zero is non-zero.
38832 2017-02-03 Jason Merrill <jason@redhat.com>
38835 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38836 avoid copying non-taken branch.
38838 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38840 PR tree-optimization/79340
38841 * tree-vect-loop.c (vectorizable_reduction): Release
38842 vec_defs elements after safe_splicing them into other vectors.
38845 PR tree-optimization/79327
38846 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
38847 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
38849 (format_integer): Use wide_int_to_tree instead of build_int_cst
38850 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38851 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38852 of shortest and longest sequence.
38854 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
38856 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
38857 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
38859 2017-02-03 Walter Lee <walt@tilera.com>
38862 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
38863 after initial stackframe link reg save.
38864 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
38866 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38869 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
38870 wu for stxssp alternative.
38872 2017-02-03 Martin Sebor <msebor@redhat.com>
38874 PR tree-optimization/79352
38875 * gimple-fold.c (get_range_strlen): Add argument.
38876 (get_range_strlen): Change return type to bool.
38877 (get_maxval_strlen): Pass in a dummy argument.
38878 * gimple-fold.h (get_range_strlen): Change return type to bool.
38879 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38880 * tree.h (array_at_struct_end_p): Add argument.
38881 * tree.c (array_at_struct_end_p): Handle it.
38883 2017-02-03 Martin Liska <mliska@suse.cz>
38886 * multiple_target.c (create_dispatcher_calls): Redirect edge
38887 from a caller of a dispatcher.
38888 (expand_target_clones): Make the clones local.
38889 (ipa_target_clone): Do both target clones and resolvers.
38890 (ipa_dispatcher_calls): Remove the pass.
38891 (pass_dispatcher_calls::gate): Likewise.
38892 (make_pass_dispatcher_calls): Likewise.
38893 * passes.def (pass_target_clone): Put as very first IPA early
38896 2017-02-03 Martin Liska <mliska@suse.cz>
38898 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38899 in case of a function with ifunc attribute.
38901 2017-02-03 Martin Liska <mliska@suse.cz>
38903 * cgraph.c (cgraph_node::dump): Dump function version info.
38904 * symtab.c (symtab_node::dump_base): Add missing new line.
38906 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38908 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38909 (ifcombine_ifandif): Use it.
38911 2017-02-03 Martin Liska <mliska@suse.cz>
38913 * doc/invoke.texi: Document default value for
38914 use-after-scope-direct-emission-threshold.
38916 2017-02-03 Martin Liska <mliska@suse.cz>
38918 PR tree-optimization/79339
38919 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38920 (format_floating): Likewise.
38922 2017-02-03 Martin Liska <mliska@suse.cz>
38925 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38927 (ipa_node_params_t::remove): Likewise.
38928 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38929 initialization from removed ipa_node_params_t::insert.
38930 (ipa_node_params::~ipa_node_params): Move from removed
38931 ipa_node_params_t::release.
38932 * symbol-summary.h (symbol_summary::m_released): New member.
38933 Do not release a summary twice. Do not allow to call finalizer
38934 for types of a summary that live in GGC memory.
38936 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38938 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38941 2017-02-02 Martin Sebor <msebor@redhat.com>
38943 PR middle-end/79275
38944 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38945 (format_string): Tighten up the range of output for non-constant
38946 strings and correct the expected range for wide non-constant strings.
38948 2017-02-02 Martin Sebor <msebor@redhat.com>
38950 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38952 PR middle-end/32003
38953 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38955 (-fdump-tree-@var): Add to index and document how to come up
38956 with pass-specific option and dump file names.
38957 (-fdump-passes): Clarify where to look for output.
38959 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38961 PR middle-end/77445
38962 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38963 statistics of the analyzed path; allow threading for speed when
38964 any of BBs along the path are optimized for speed.
38966 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38968 PR middle-end/78468
38969 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38970 settings of the virtual registers.
38973 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38975 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38976 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38979 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38981 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38982 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38984 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38986 * config/s390/s390.md: Add missing comments with the expanded
38988 * config/s390/vector.md: Likewise.
38989 * config/s390/vx-builtins.md: Likewise.
38991 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38994 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38995 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38996 conditions on a single line.
38998 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39000 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39001 __S390_VX__ to __VX__.
39003 2017-02-01 Andrew Pinski <apinski@cavium.com>
39005 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39006 stmt_info to record_stmt_cost.
39007 (vect_get_known_peeling_cost): Pass stmt_info if known to
39009 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39010 cpu_vector_cost field into
39011 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39012 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39013 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39014 splitting of scalar_stmt_cost and vec_stmt_cost.
39015 (thunderx_vector_cost): Likewise.
39016 (cortexa57_vector_cost): LIkewise.
39017 (exynosm1_vector_cost): Likewise.
39018 (xgene1_vector_cost): Likewise.
39019 (thunderx2t99_vector_cost): Improve after the splitting of the two
39021 (aarch64_builtin_vectorization_cost): Update for the splitting of
39022 scalar_stmt_cost and vec_stmt_cost.
39024 2017-02-01 Torvald Riegel <triegel@redhat.com>
39025 Richard Henderson <rth@redhat.com>
39027 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39028 conditional on existance of a fast atomic load.
39029 * optabs-query.c (can_atomic_load_p): New function.
39030 * optabs-query.h (can_atomic_load_p): Declare it.
39031 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39032 no fast atomic load is available for the particular size of access.
39033 (expand_atomic_compare_and_swap): Likewise.
39034 (expand_atomic_load): Likewise.
39035 (expand_atomic_store): Likewise.
39036 (expand_atomic_fetch_op): Likewise.
39037 * testsuite/lib/target-supports.exp
39038 (check_effective_target_sync_int_128): Remove x86 because it provides
39039 no fast atomic load.
39040 (check_effective_target_sync_int_128_runtime): Likewise.
39042 2017-02-01 Richard Biener <rguenther@suse.de>
39044 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39045 (graphite_transform_loops): Provide opt-info for optimized nests.
39046 * tree-parloop.c (parallelize_loops): Provide opt-info for
39047 parallelized loops.
39049 2017-02-01 Richard Biener <rguenther@suse.de>
39051 PR middle-end/79315
39052 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39053 was not set before.
39055 2017-02-01 Richard Biener <rguenther@suse.de>
39057 PR tree-optimization/71824
39058 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39059 Verify the loops are valid in the merged SESE region.
39060 (scop_detection::can_represent_loop_1): Check analyzing the
39061 evolution of the number of iterations in the region succeeds.
39063 2017-01-31 Ian Lance Taylor <iant@golang.org>
39065 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39066 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39068 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39070 PR preprocessor/79210
39071 * input.c (get_substring_ranges_for_loc): Replace line_width
39072 assertion with error-handling.
39074 2017-01-31 Richard Biener <rguenther@suse.de>
39076 PR tree-optimization/77318
39077 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39078 (create_pw_aff_from_tree): Take loop parameter.
39079 (add_condition_to_pbb): Pass loop of the condition to
39080 create_pw_aff_from_tree.
39082 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39084 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39085 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39087 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39091 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39093 (convert_int_to_float128): Likewise.
39094 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39095 (convert_int_to_float128): Likewise.
39096 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39097 (UNSPEC_IEEE128_CONVERT): Likewise.
39098 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39099 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39100 Use local variables for IBM extended format.
39101 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39102 (fix_trunc<mode>si2_fprs): Likewise.
39103 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39104 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39105 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39106 to know that we can now have integers of all sizes in vector
39108 (fix<uns>_<mode>di2_hw): Likewise.
39109 (float<uns>_<mode>si2_hw): Likewise.
39110 (fix_<mode>si2_hw): Likewise.
39111 (fixuns_<mode>si2_hw): Likewise.
39112 (float<uns>_<mode>di2_hw): Likewise.
39113 (float_<mode>di2_hw): Likewise.
39114 (float_<mode>si2_hw): Likewise.
39115 (floatuns_<mode>di2_hw): Likewise.
39116 (floatuns_<mode>si2_hw): Likewise.
39117 (xscvqp<su>wz_<mode>): Delete, no longer used.
39118 (xscvqp<su>dz_<mode>): Likewise.
39119 (xscv<su>dqp_<mode>): Likewise.
39120 (ieee128_mfvsrd_64bit): Likewise.
39121 (ieee128_mfvsrd_32bit): Likewise.
39122 (ieee128_mfvsrwz): Likewise.
39123 (ieee128_mtvsrw): Likewise.
39124 (ieee128_mtvsrd_64bit): Likewise.
39125 (ieee128_mtvsrd_32bit): Likewise.
39127 2017-01-31 Martin Liska <mliska@suse.cz>
39130 * ipa-prop.c (ipa_free_all_node_params): Call release method
39131 instead of ~sumbol_summary to not to trigger double times
39134 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39136 PR tree-optimization/71691
39137 * bitmap.h (class auto_bitmap): New.
39138 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39139 is_maybe_undefined instead of ssa_undefined_value_p.
39141 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39143 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39144 __S390_ARCH_LEVEL__ to __ARCH__.
39146 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39148 PR tree-optimization/79267
39149 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39150 if should_remove_lhs_p is true.
39152 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39155 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39156 (add_alignment_attribute): New.
39157 (base_type_die): Add alignment attribute.
39158 (subrange_type_die): Likewise.
39159 (modified_type_die): Likewise.
39160 (gen_array_type_die): Likewise.
39161 (gen_descr_array_type_die: Likewise.
39162 (gen_enumeration_type_die): Likewise.
39163 (gen_subprogram_die): Likewise.
39164 (gen_variable_die): Likewise.
39165 (gen_field_die): Likewise.
39166 (gen_ptr_to_mbr_type_die): Likewise.
39167 (gen_struct_or_union_type_die): Likewise.
39168 (gen_subroutine_type_die): Likewise.
39169 (gen_typedef_die): Likewise.
39170 (base_type_cmp): Compare alignment attribute.
39172 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39175 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39176 (setb_unsigned) New pattern for setb with CCUNS.
39177 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39178 subfc./subfe sequence to avoid overflow problems. Generate a
39179 shorter sequence with cmpld/setb for power9.
39180 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39181 for generating subfc. instruction.
39182 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39183 now uses this instruction.
39185 2017-01-30 Ian Lance Taylor <iant@google.com>
39188 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39189 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39191 2017-01-30 Martin Sebor <msebor@redhat.com>
39193 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39194 Move constant to the right of a relational operator.
39195 (get_mpfr_format_length, format_character, format_string): Ditto.
39196 (should_warn_p, maybe_warn): Same.
39198 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39200 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39203 * asan.c (get_translation_unit_decl): Remove function.
39204 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39206 2017-01-30 Martin Liska <mliska@suse.cz>
39208 PR gcov-profile/79259
39209 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39210 -fprofile-generate.
39212 2017-01-30 Martin Liska <mliska@suse.cz>
39215 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39216 Initialize variables with NULL value.
39218 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39221 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39223 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39225 2017-01-30 Richard Biener <rguenther@suse.de>
39227 PR tree-optimization/79276
39228 * tree-vrp.c (process_assert_insertions): Properly adjust common
39229 when removing a duplicate.
39231 2017-01-30 Richard Biener <rguenther@suse.de>
39233 PR tree-optimization/79256
39234 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39235 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39237 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39239 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39242 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39243 ("*r<noxa>sbg_<mode>_sll_bitmask")
39244 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39245 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39246 Use contiguous_bitmask_nowrap_operand.
39248 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39251 * config/rs6000/altivec.h (vec_xl): Revise #define.
39252 (vec_xst): Likewise.
39254 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39256 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39258 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39260 PR rtl-optimization/79194
39261 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39262 traps before call to bypass_conditional_jumps.
39264 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39266 PR tree-optimization/71374
39267 * lra-constraints.c (check_conflict_input_operands): New.
39268 (match_reload): Use it.
39270 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39273 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39274 account to calculate conflict_set.
39276 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39278 PR rtl-optimization/78559
39279 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39280 other_insn in combine.
39282 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39284 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39285 uint16_type_node for BT_UINT16.
39287 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39289 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39290 "RTL Tests" to menu.
39291 (GIMPLE Tests): New node.
39292 (RTL Tests): New node.
39294 2017-01-27 Richard Biener <rguenther@suse.de>
39296 PR tree-optimization/79245
39297 * tree-loop-distribution.c (distribute_loop): Apply cost
39298 modeling also to detected patterns.
39300 2017-01-27 Richard Biener <rguenther@suse.de>
39302 PR tree-optimization/71433
39303 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39304 (compare_assert_loc): New function.
39305 (process_assert_insertions): Sort and optimize assert locations
39306 to remove duplicates and push down identical assertions on
39307 edges to their destination block.
39309 2017-01-27 Richard Biener <rguenther@suse.de>
39311 PR tree-optimization/79244
39312 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39313 out SSA names even if abnormal.
39315 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39317 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39318 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39319 instead of MPFR_RNDN.
39321 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39324 * arm.c (arm_option_override): Don't call build_target_option_node
39325 until after doing all option overrides.
39326 (arm_valid_target_attribute_tree): Likewise.
39328 2017-01-27 Martin Liska <mliska@suse.cz>
39330 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39331 for {cd}tors and C++ {cd}tors.
39333 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39335 * config/s390/s390.md ("*setmem_long_and")
39336 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39338 2017-01-26 Martin Sebor <msebor@redhat.com>
39340 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39343 2017-01-26 Martin Sebor <msebor@redhat.com>
39345 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39346 HAVE_DFmode before using XFmode or DFmode.
39347 (parse_directive): Avoid using the z length modifier to avoid
39348 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39350 PR middle-end/78703
39351 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39352 to accept adjustment as an array.
39353 (get_int_range): New function.
39354 (struct directive): Make width and prec arrays.
39355 (directive::set_width, directive::set_precision): Call get_int_range.
39356 (format_integer, format_floating): Handle width and precision ranges.
39357 (format_string, parse_directive): Same.
39359 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39362 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39363 comdat_type_p set, just clone them, but keep the children in the
39367 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39368 which have direct callers with -fvar-tracking-assignments enabled
39370 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39371 inside of type units.
39373 2017-01-26 Martin Sebor <msebor@redhat.com>
39375 PR middle-end/78703
39376 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39378 (struct format_result): Replace number_chars, number_chars_min,
39379 and number_chars_max with a single member of struct result_range.
39381 (format_result::operator+=): Adjust.
39382 (struct fmtresult): Remove bounded. Handle likely and unlikely
39384 (fmtresult::adjust_for_width_or_precision): New function.
39385 (fmtresult:type_max_digits): New function.
39386 (bytes_remaining): Handle likely and unlikely counters.
39387 (min_bytes_remaining): Remove.
39388 (format_percent): Simplify.
39389 (format_integer, format_floating): Set likely and unlikely counters.
39390 (get_string_length, format_character, format_string): Same.
39391 (format_plain, should_warn_p): New function.
39392 (maybe_warn): Call should_warn_p. Update diagnostic messages
39393 and handle those for all directives, including plain strings.
39394 (format_directive): Handle likely and unlikely counters.
39395 Remove unnecessary quoting from diagnostics. Add an informational
39397 (add_bytes): Remove.
39398 (pass_sprintf_length::compute_format_length): Simplify.
39399 (try_substitute_return_value): Handle likely and unlikely counters.
39401 2017-01-26 Carl Love <cel@us.ibm.com>
39403 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39404 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39406 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39409 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39410 endianess for subregs into account.
39411 * lra-constraints.c (lra_constraints): Do risky transformations
39412 always on the first iteration.
39413 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39414 last_call_used_reg_set.
39415 (process_bb_lives): Define and use last_call_used_reg_set.
39416 * lra.c (lra): Always continue after lra_constraints on the first
39419 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39421 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39423 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39425 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39427 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39428 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39429 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39430 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39431 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39432 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39433 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39434 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39435 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39437 2017-01-26 Marek Polacek <polacek@redhat.com>
39440 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39441 for the third operand.
39443 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39445 PR middle-end/79236
39446 * omp-low.c (struct omp_context): Add simt_stmt field.
39447 (scan_omp_for): Return omp_context *.
39448 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39449 context to the _simt_ SIMD stmt.
39450 (lower_omp_for): For combined SIMD with sibling _simt_
39451 SIMD, make sure to use the same decls in _looptemp_
39452 clauses as in the sibling.
39454 2017-01-26 David Sherwood <david.sherwood@arm.com>
39456 PR middle-end/79212
39457 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39460 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39463 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39464 emit fld b; fld a; if possible.
39466 * brig-builtins.def: Update copyright years.
39467 * config/arm/arm_acle_builtins.def: Update copyright years.
39469 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39472 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39473 constraint instead of o for the stxsd instruction.
39475 2017-01-25 Carl Love <cel@us.ibm.com>
39477 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39478 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39480 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39482 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39484 2017-01-25 Richard Biener <rguenther@suse.de>
39486 PR tree-optimization/69264
39487 * target.def (vector_alignment_reachable): Improve documentation.
39488 * doc/tm.texi: Regenerate.
39489 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39491 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39492 earlier changes with respect to TYPE_USER_ALIGN.
39493 (vector_alignment_reachable_p): Likewise. Improve dumping.
39495 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39498 * config/arm/arm.md (xordi3): Force constant operand into a register
39501 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39503 * doc/invoke.texi (-fstore-merging): Correct default optimization
39504 levels at which it is enabled.
39505 (-O): Move -fstore-merging from list to...
39508 2017-01-25 Richard Biener <rguenther@suse.de>
39511 * omp-expand.c: Include debug.h.
39512 (expand_omp_taskreg): Make sure to generate early debug before
39513 outlining anything from a function.
39514 (expand_omp_target): Likewise.
39515 (grid_expand_target_grid_body): Likewise.
39517 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39520 * asan.c (get_translation_unit_decl): New function.
39521 (asan_add_global): Extract modules file name from globals
39522 TRANSLATION_UNIT_DECL name.
39524 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39527 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39528 for long calls with APCS frame and VFP.
39530 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39532 * cfg.c (original_copy_tables_initialized_p): New function.
39533 * cfg.h (original_copy_tables_initialized_p): New decl.
39534 * cfgrtl.c (relink_block_chain): Guard the call to
39535 free_original_copy_tables with a call to
39536 original_copy_tables_initialized_p.
39537 * cgraph.h (symtab_node::native_rtl_p): New decl.
39538 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39539 (symtab_node::needed_p): Don't assert for early assembly output
39540 for __RTL functions.
39541 (cgraph_node::finalize_function): Set "force_output" for __RTL
39543 (cgraph_node::analyze): Bail out early for __RTL functions.
39544 (analyze_functions): Update assertion to support __RTL functions.
39545 (cgraph_node::expand): Bail out early for __RTL functions.
39546 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39548 * function.h (struct function): Update comment for field
39550 * gimple-expr.c: Include "tree-pass.h".
39551 (gimple_has_body_p): Return false for __RTL functions.
39552 * Makefile.in (OBJS): Add run-rtl-passes.o.
39553 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39555 (gcc::pass_manager::get_clean_slate): New accessor.
39556 * passes.c: Include "insn-addr.h".
39557 (should_skip_pass_p): Add logging. Update logic for running
39558 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39559 property-provider override so it is only done for gimple passes.
39561 (skip_pass): New function.
39562 (execute_one_pass): Call skip_pass when skipping passes.
39563 * read-md.c (md_reader::read_char): Support filtering
39564 the input to a subset of line numbers.
39565 (md_reader::md_reader): Initialize fields
39566 m_first_line and m_last_line.
39567 (md_reader::read_file_fragment): New function.
39568 * read-md.h (md_reader::read_file_fragment): New decl.
39569 (md_reader::m_first_line): New field.
39570 (md_reader::m_last_line): New field.
39571 * read-rtl-function.c (function_reader::create_function): Only
39572 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39573 curr_properties. Set DECL_INITIAL to a dummy block.
39574 (read_rtl_function_body_from_file_range): New function.
39575 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39577 * run-rtl-passes.c: New file.
39578 * run-rtl-passes.h: New file.
39580 2017-01-24 Jeff Law <law@redhat.com>
39582 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39585 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39587 PR tree-optimization/79159
39588 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39589 (record_nonwrapping_iv): Improve boundary using above function if no
39590 value range information.
39592 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39593 Martin Jambor <mjambor@suse.cz>
39595 * brig-builtins.def: New file.
39596 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39597 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39598 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39599 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39600 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39601 * builtin-types.def (BT_INT8): New.
39602 (BT_INT16): Likewise.
39603 (BT_UINT8): Likewise.
39604 (BT_UINT16): Likewise.
39605 (BT_FN_ULONG): Likewise.
39606 (BT_FN_UINT_INT): Likewise.
39607 (BT_FN_UINT_ULONG): Likewise.
39608 (BT_FN_UINT_LONG): Likewise.
39609 (BT_FN_UINT_PTR): Likewise.
39610 (BT_FN_ULONG_PTR): Likewise.
39611 (BT_FN_INT8_FLOAT): Likewise.
39612 (BT_FN_INT16_FLOAT): Likewise.
39613 (BT_FN_UINT32_FLOAT): Likewise.
39614 (BT_FN_UINT16_FLOAT): Likewise.
39615 (BT_FN_UINT8_FLOAT): Likewise.
39616 (BT_FN_UINT64_FLOAT): Likewise.
39617 (BT_FN_UINT16_UINT32): Likewise.
39618 (BT_FN_UINT32_UINT16): Likewise.
39619 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39620 (BT_FN_INT_PTR_INT): Likewise.
39621 (BT_FN_UINT_PTR_UINT): Likewise.
39622 (BT_FN_LONG_PTR_LONG): Likewise.
39623 (BT_FN_ULONG_PTR_ULONG): Likewise.
39624 (BT_FN_VOID_UINT64_UINT64): Likewise.
39625 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39626 (BT_FN_INT8_INT8_INT8): Likewise.
39627 (BT_FN_INT16_INT16_INT16): Likewise.
39628 (BT_FN_INT_INT_INT): Likewise.
39629 (BT_FN_UINT_FLOAT_UINT): Likewise.
39630 (BT_FN_FLOAT_UINT_UINT): Likewise.
39631 (BT_FN_ULONG_UINT_UINT): Likewise.
39632 (BT_FN_ULONG_UINT_PTR): Likewise.
39633 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39634 (BT_FN_UINT_UINT_UINT): Likewise.
39635 (BT_FN_VOID_UINT_PTR): Likewise.
39636 (BT_FN_UINT_UINT_PTR: Likewise.
39637 (BT_FN_UINT32_UINT64_PTR): Likewise.
39638 (BT_FN_INT_INT_UINT_UINT): Likewise.
39639 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39640 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39641 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39642 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39643 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39644 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39645 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39646 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39647 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39648 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39649 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39650 * doc/frontends.texi: List BRIG FE.
39651 * doc/install.texi (Testing): Add BRIG tesring requirements.
39652 * doc/invoke.texi (Overall Options): Mention BRIG.
39653 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39655 2017-01-24 Richard Biener <rguenther@suse.de>
39657 PR translation/79208
39658 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39660 2017-01-24 Martin Jambor <mjambor@suse.cz>
39663 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39664 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39665 and known_contexts.
39667 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39669 PR middle-end/79123
39670 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39671 casts from signed to unsigned really don't have a range.
39673 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39675 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39676 GMP_RNDx for compatiblity.
39678 2017-01-24 Martin Liska <mliska@suse.cz>
39681 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39682 that would prevent us to call alloca with -1 as argument.
39684 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39686 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39687 Avoid -Wformat-security warning.
39689 2017-01-23 Andrew Pinski <apinski@cavium.com>
39691 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39694 2017-01-23 Martin Sebor <msebor@redhat.com>
39696 PR middle-end/78703
39697 * gimple-ssa-sprintf.c (warn_level): New global.
39698 (format_integer): Use it here and throughout the rest of the file.
39699 Use the same switch to compute sign as base.
39700 (maybe_warn): New function.
39701 (format_directive): Factor out warnings into maybe_warn.
39702 Add debugging output. Use warn_level.
39703 (add_bytes): Use warn_level.
39704 (pass_sprintf_length::compute_format_length): Add debugging output.
39705 (try_substitute_return_value): Same.
39706 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39708 PR middle-end/78703
39709 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39710 (struct fmtresult, format_integer, format_floating): Adjust.
39711 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39712 (get_string_length, format_string,format_directive): Same.
39713 (pass_sprintf_length::compute_format_length): Same.
39714 (try_substitute_return_value): Simplify slightly.
39716 PR middle-end/78703
39717 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39718 (fmtresult::operator+=): Outlined.
39719 (struct fmtresult): Add ctors.
39720 (struct conversion_spec): Rename...
39721 (struct directive): ...to this. Add and remove data members.
39722 (directive::set_width, directive::set_precision): New functions.
39723 (format_percent): Use fmtresult ctor.
39724 (get_width_and_precision): Remove.
39725 (format_integer): Make naming changes. Avoid computing width and
39727 (format_floating): Same. Adjust indentation.
39728 (format_character, format_none): New functions.
39729 (format_string): Moved character handling to format_character.
39730 (format_directive): Remove arguments, change return type.
39731 (parse_directive): New function.
39732 (pass_sprintf_length::compute_format_length): Move directive
39733 parsing to parse_directive.
39735 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39737 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39738 (assign_assembler_name_if_needed): ... this.
39739 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39740 (assign_assembler_name_if_needed): ... this.
39741 (free_lang_data_in_cgraph): Adjust callers.
39742 * cgraphunit.c (cgraph_node::analyze): Likewise.
39743 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39746 2017-01-23 Richard Biener <rguenther@suse.de>
39748 PR tree-optimization/79088
39749 PR tree-optimization/79188
39750 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39751 resetting loop bounds after last path deletion. Reset loop
39752 bounds of the target loop, make code match the comments.
39753 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39754 Make sure loops need no fixups.
39756 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39758 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39759 exponent support with double type for first argument.
39760 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39761 type returned by __builtin_vec_extract_sig,
39762 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39763 functions from "vector int" to "vector unsigned int" or from
39764 "vector long long int" to "vector unsigned long long int".
39765 Changed type returned by __builtin_vec_extract_exp,
39766 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39767 functions from "vector int" to "vector unsigned int" or from
39768 "vector long long int" to "vector unsigned long long int".
39769 Changed return type of __builtin_vec_test_data_class,
39770 __builtin_vec_test_data_class_sp, and
39771 __builtin_vec_test_data_class_dp from "vector int" to
39772 "vector bool int" or from "vector long long int" to "vector bool
39773 long long int" and changed second argument type from "unsigned
39774 int" to "int". Added new overloaded function forms "vector float
39775 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39776 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39777 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39778 double, vector unsigned long long int)" and "vector double
39779 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39780 long int)". Changed return type of
39781 __builtin_scalar_test_data_class and
39782 __builtin_scalar_test_data_class_sp and
39783 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39784 int" and changed second argument from "unsigned int" to "int".
39785 Changed type returned by __builtin_scalar_test_neg,
39786 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39787 from "int" to "bool int". Added new overloaded function form
39788 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39789 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
39790 exponent double-precision with floating point first argument.
39791 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
39792 documentation of scalar_test_data_class, scalar_test_neg,
39793 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
39794 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
39795 vec_test_data_class built-in functions to reflect refinements in
39796 their type signatures.
39798 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
39800 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
39802 (aarch64_elf_asm_destructor): Likewise.
39804 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
39806 PR rtl-optimization/78634
39807 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
39808 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
39809 * ifcvt.c (noce_try_cmove): Add missing cost check.
39811 PR rtl-optimization/71724
39812 * combine.c (if_then_else_cond): Look for situations where it is
39813 beneficial to undo the work of one of the recursive calls.
39815 2017-01-23 Bin Cheng <bin.cheng@arm.com>
39817 PR tree-optimization/70754
39818 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39819 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39820 combined stmt before it if not NULL.
39821 (combine_chains): Process refs reversely and compute dominance point
39824 2017-01-23 Martin Liska <mliska@suse.cz>
39826 PR tree-optimization/79196
39827 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39828 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39830 (strlen_optimize_stmt): Call the renamed function.
39832 2017-01-23 Michael Matz <matz@suse.de>
39834 PR tree-optimization/78384
39835 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39837 2017-01-23 Richard Biener <rguenther@suse.de>
39839 PR tree-optimization/79186
39840 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39841 both incoming edges before moving an assert.
39843 2017-01-23 Martin Jambor <mjambor@suse.cz>
39845 * ipa-prop.c (load_from_param_1): Removed.
39846 (load_from_unmodified_param): Bits from load_from_param_1 put back
39848 (load_from_param): Removed.
39849 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
39850 with stmt. Reverted back to use of load_from_unmodified_param.
39852 2017-01-23 Martin Jambor <mjambor@suse.cz>
39855 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
39856 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
39857 field a pointer to garbage collected vector, mark lattices and
39858 ipcp_orig_node with GTY((skip)).
39859 (ipa_get_param_count): Adjust to descriptors being a pointer.
39860 (ipa_get_param): Likewise.
39861 (ipa_get_type): Likewise.
39862 (ipa_get_param_move_cost): Likewise.
39863 (ipa_set_param_used): Likewise.
39864 (ipa_get_controlled_uses): Likewise.
39865 (ipa_set_controlled_uses): Likewise.
39866 (ipa_is_param_used): Likewise.
39867 (ipa_node_params_t): Move into garbage collector. New methods insert
39869 (ipa_node_params_sum): Annotate wth GTY(()).
39870 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
39872 (ipa_load_from_parm_agg): Adjust declaration.
39873 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
39874 * ipa-profile.c (ipa_profile): Likewise.
39875 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39876 (ipa_populate_param_decls): Make descriptors parameter garbage
39878 (ipa_dump_param): Adjust to descriptors being a pointer.
39879 (ipa_alloc_node_params): Likewise.
39880 (ipa_initialize_node_params): Likewise.
39881 (load_from_param_1): Make descriptors parameter garbage collected.
39882 (load_from_unmodified_param): Likewise.
39883 (load_from_param): Likewise.
39884 (ipa_load_from_parm_agg): Likewise.
39885 (ipa_node_params::~ipa_node_params): Removed.
39886 (ipa_free_all_node_params): Remove call to delete operator.
39887 (ipa_node_params_t::insert): New.
39888 (ipa_node_params_t::remove): Likewise.
39889 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39890 copy known_csts and known_contexts vectors.
39891 (ipa_read_node_info): Adjust to descriptors being a pointer.
39892 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39894 (ipcp_transform_function): Make descriptors variable garbage
39897 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39899 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39900 * config/i386/avx512dqintrin.h: Ditto.
39901 * config/i386/avx512fintrin.h: Ditto.
39902 * config/i386/i386.c: Handle new builtins.
39903 * config/i386/i386-builtin.def: Add new builtins.
39904 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39905 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39907 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39908 Martin Liska <mliska@suse.cz>
39910 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39911 * asan.c (asan_expand_poison_ifn): Support stores and use
39912 appropriate ASAN report function.
39913 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39914 * internal-fn.def (ASAN_POISON_USE): Declare.
39915 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39916 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39917 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39918 ASAN_POISON calls w/o LHS.
39919 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39920 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39921 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39922 * gimplify.c (asan_poison_variables): Add attribute
39923 use_after_scope_memory to variables that really needs to live
39925 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39926 having the attribute.
39928 2017-01-23 Martin Liska <mliska@suse.cz>
39930 * asan.c (create_asan_shadow_var): New function.
39931 (asan_expand_poison_ifn): Likewise.
39932 * asan.h (asan_expand_poison_ifn): New declaration.
39933 * internal-fn.c (expand_ASAN_POISON): Likewise.
39934 * internal-fn.def (ASAN_POISON): New builtin.
39935 * sanopt.c (pass_sanopt::execute): Expand
39936 asan_expand_poison_ifn.
39937 * tree-inline.c (copy_decl_for_dup_finish): Make function
39939 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39940 * tree-ssa.c (is_asan_mark_p): New function.
39941 (execute_update_addresses_taken): Rewrite local variables
39942 (identified just by use-after-scope as addressable) into SSA.
39944 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39946 * doc/install.texi (Specific): opensource.apple.com uses https
39947 now. Remove trailing slash.
39949 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39951 * README.Portability: Remove note on an Irix compatibility issue.
39953 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39955 * gcov.c (INCLUDE_ALGORITHM): Define.
39956 (INCLUDE_VECTOR): Define.
39957 No longer include <vector> and <algorithm> directly.
39959 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39961 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39963 * doc/invoke.texi (Code Gen Options): Ditto.
39965 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39968 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39970 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39972 rtl-optimization/79125
39973 * cprop.c (local_cprop_pass): Handle cases where we make an
39974 unconditional trap.
39976 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39980 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39981 read from, for big endian.
39983 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39985 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39986 register pauth builtins for LP64 only.
39988 2017-01-20 Marek Polacek <polacek@redhat.com>
39991 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39994 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39996 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39998 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39999 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40000 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40002 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40005 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40006 in big-endian mode when they are not a single duplicated value.
40008 2017-01-20 Richard Biener <rguenther@suse.de>
40010 * BASE-VER: Bump to 7.0.1.
40012 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40014 * omp-low.c (omplow_simd_context): New struct. Use it...
40015 (lower_rec_simd_input_clauses): ...here and...
40016 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40017 references to idx, lane, max_vf, is_simt.
40019 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40021 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40024 2017-01-20 Martin Jambor <mjambor@suse.cz>
40026 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40027 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40029 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40030 (GTFILES): Rename hsa.c to hsa-common.c.
40031 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40032 * hsa-dump.c: Likewise.
40033 * hsa-gen.c: Likewise.
40034 * hsa-regalloc.c: Likewise.
40035 * ipa-hsa.c: Likewise.
40036 * omp-expand.c: Likewise.
40037 * omp-low.c: Likewise.
40038 * toplev.c: Likewise.
40040 2017-01-20 Marek Polacek <polacek@redhat.com>
40043 * doc/invoke.texi: Document -Wduplicated-branches.
40044 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40045 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40046 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40047 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40048 return 0 only when not OEP_LEXICOGRAPHIC.
40049 (fold_build_cleanup_point_expr): Use the expression
40050 location when building CLEANUP_POINT_EXPR.
40051 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40052 * tree.c (add_expr): Handle error_mark_node.
40054 2017-01-20 Martin Liska <mliska@suse.cz>
40057 * tree-profile.c (init_ic_make_global_vars): Do not call
40059 (gimple_init_gcov_profiler): Likewise.
40061 2017-01-20 Martin Liska <mliska@suse.cz>
40064 * cgraph.h (maybe_create_reference): Remove argument and
40066 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40068 * ipa-cp.c (create_specialized_node): Likewise.
40069 * symtab.c (symtab_node::maybe_create_reference): Handle
40070 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40072 2017-01-20 Martin Liska <mliska@suse.cz>
40074 * read-rtl-function.c (function_reader::create_function): Use
40075 build_decl instread of build_decl_stat.
40077 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40079 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40080 * config/i386/avx512dqintrin.h: Ditto.
40081 * config/i386/avx512fintrin.h: Ditto.
40082 * config/i386/i386-builtin-types.def: Add new types.
40083 * config/i386/i386.c: Handle new types.
40084 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40085 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40086 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40087 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40088 (__builtin_ia32_kshiftridi): New.
40089 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40091 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40095 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40096 define to rs6000_init_stack_protect_guard.
40097 (rs6000_init_stack_protect_guard): New function.
40099 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40100 Yunqiang Su <yunqiang.su@imgtec.com>
40102 * config.gcc (supported_defaults): Add madd4.
40103 (with_madd4): Add validation.
40104 (all_defaults): Add madd4.
40105 * config/mips/mips.opt (mmadd4): New option.
40106 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40108 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40110 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40111 (ISA_HAS_FUSED_MADD4): Likewise.
40112 * doc/invoke.texi (-mmadd4): Document the new option.
40113 * doc/install.texi (--with-madd4): Document the new option.
40115 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40117 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40118 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40119 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40120 (aarch64_init_pauth_hint_builtins): New.
40121 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40122 (aarch64_expand_builtin): Expand new builtins.
40124 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40126 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40127 * combine-stack-adj.c (no_unhandled_cfa): Handle
40128 REG_CFA_TOGGLE_RA_MANGLE.
40129 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40130 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40131 info for return address signing.
40132 (aarch64_expand_epilogue): Likewise.
40134 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40136 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40137 * config/aarch64/aarch64-protos.h
40138 (aarch64_return_address_signing_enabled): New declaration.
40139 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40141 (aarch64_expand_prologue): Sign return address before it's pushed onto
40143 (aarch64_expand_epilogue): Authenticate return address fetched from
40145 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40146 (aarch64_attributes): New function attributes for "sign-return-address".
40147 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40148 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40149 ("*do_return"): Generate combined instructions according to key index.
40150 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40151 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40153 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40154 * config/aarch64/aarch64.opt (msign-return-address=): New.
40155 * doc/extend.texi (AArch64 Function Attributes): Documents
40156 "sign-return-address=".
40157 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40159 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40161 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40162 overall option summary.
40164 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40166 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40167 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40168 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40169 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40171 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40173 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40174 -mpower9-minmax by default for -mcpu=power9.
40175 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40176 128-bit floating point.
40178 2017-01-20 Alan Modra <amodra@gmail.com>
40180 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40181 optimizing for size.
40183 2017-01-20 Alan Modra <amodra@gmail.com>
40186 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40187 for strcmp and strncmp from corresponding builtin decl.
40189 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40191 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40192 instead of i386/rtems-64.h.
40193 * config/i386/rtems-64.h: Remove.
40195 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40199 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40201 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40203 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40205 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40206 Change int to HOST_WIDE_INT.
40207 * config/aarch64/aarch64-protos.h
40208 (aarch64_simd_gen_const_vector_dup): Likewise.
40209 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40211 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40213 * langhooks-def.h (lhd_type_for_size): New decl.
40214 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40215 * langhooks.c (lhd_type_for_size): New function, taken from
40218 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40220 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40221 define_bypass for CR latency.
40222 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40223 (power9-alu2): Add define_bypass for CR latency.
40225 (power9-mul): Update insn latency.
40226 (power9-mul-compare): Update insn latency, bypass latency and remove
40229 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40231 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40233 * config/aarch64/aarch64.md
40234 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40235 aarch64_nopcrelative_literal_loads.
40236 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40238 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40240 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40241 TARGET_LOONGSON_3A.
40242 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40244 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40247 * config.gcc (supported_defaults): Add lxc1-sxc1.
40248 (with_lxc1_sxc1): Add validation.
40249 (all_defaults): Add lxc1-sxc1.
40250 * config/mips/mips.opt (mlxc1-sxc1): New option.
40251 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40253 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40254 __mips_no_lxc1_sxc1.
40255 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40256 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40257 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40259 2017-01-19 Richard Biener <rguenther@suse.de>
40261 PR tree-optimization/72488
40262 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40263 sure to restore SSA info.
40264 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40266 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40268 PR rtl-optimization/79121
40269 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40270 of the inner type when shifting an extended value.
40272 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40275 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40276 interposable aliases.
40278 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40281 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40282 Use the evmergelohi instruction.
40283 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40284 (mov_sitf_e500_subreg8_2_be): Likewise.
40285 (mov_sitf_e500_subreg12_2_le): Likewise.
40286 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40287 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40288 (mov_sitf_e500_subreg8_2_le): Likewise.
40289 (mov_sitf_e500_subreg12_2_be): Likewise.
40291 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40293 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40294 attribute from vecsimple to vecperm.
40295 (altivec_vbpermq2): Likewise.
40297 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40300 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40302 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40303 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40304 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40305 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40306 case where N arg is SIZE_MAX.
40307 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40308 (cmpstrsi): Add pattern.
40310 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40312 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40313 __builtin_vec_revb builtins.
40314 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40315 built-in functions to support generation of the ISA 3.0 XXBR<x>
40316 vector byte reverse instructions.
40317 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40318 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40319 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40320 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40321 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40322 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40323 (P9V_BUILTIN_VEC_REVB): Likewise.
40324 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40325 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40326 (p9_xxbrq_v16qi): Likewise.
40327 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40328 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40329 (p9_xxbrh_v8hi): Likewise.
40330 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40331 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40332 vec_revb built-in functions.
40334 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40336 PR rtl-optimization/78952
40337 * config/i386/i386.md (any_extract): New code iterator.
40338 (*insvqi_2): Use any_extract for source operand.
40339 (*insvqi_3): Use any_shiftrt for source operand.
40341 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40343 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40345 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40347 2017-01-18 Matthias Klose <doko@ubuntu.com>
40349 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40351 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40353 * config/rs6000/altivec.h (vec_bperm): Change #define.
40354 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40355 (altivec_vbpermq2): New define_insn.
40356 (altivec_vbpermd): Likewise.
40357 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40358 function interface.
40359 (VBPERMD): Likewise.
40360 (VBPERM): New polymorphic function interface.
40361 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40362 Add entries for P9V_BUILTIN_VEC_VBPERM.
40363 * doc/extend.texi: Add interfaces for vec_bperm.
40365 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40367 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40368 first letter of error messages.
40369 (s390_resolve_overloaded_builtin): Likewise.
40370 * config/s390/s390.c (s390_expand_builtin): Likewise.
40371 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40372 (s390_valid_target_attribute_inner_p): Likewise.
40373 * config/s390/s390.md ("tabort"): Likewise.
40375 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40377 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40378 (ISA_AVOID_DIV_HILO): New macro.
40379 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40380 (ISA_HAS_DDIV): Likewise.
40382 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40384 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40386 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40388 * doc/invoke.texi (fabi-version): Spelling fix.
40390 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40393 * doc/invoke.texi (fabi-version): Mention mangling fix for
40396 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40399 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40401 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40404 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40405 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40407 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40409 (stack_protect_test): Ditto.
40410 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40411 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40413 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40414 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40415 -mstack-protector-guard-offset=.
40416 (RS/6000 and PowerPC Options): Ditto.
40418 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40420 * config/i386/i386.h (MASK_CLASS_P): New define.
40421 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40422 there are no registers from different register sets also when
40423 mask registers are used. Update function comment.
40424 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40425 to (*k/*r) and (*k/*km) alternatives.
40427 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40429 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40430 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40431 (EH_RETURN_HANDLER_RTX): New define.
40432 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40433 Force frame pointer in EH return functions.
40434 (aarch64_expand_epilogue): Add barrier for eh_return.
40435 (aarch64_final_eh_return_addr): Remove.
40436 (aarch64_eh_return_handler_rtx): New function.
40437 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40439 (aarch64_eh_return_handler_rtx): New prototype.
40441 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40443 * config/rs6000/altivec.h (vec_rlmi): New #define.
40444 (vec_vrlnm): Likewise.
40445 (vec_rlnm): Likewise.
40446 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40447 (UNSPEC_VRLNM): Likewise.
40448 (VIlong): New mode iterator.
40449 (altivec_vrl<VI_char>mi): New define_insn.
40450 (altivec_vrl<VI_char>nm): Likewise.
40451 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40453 (VRLDNM): Likewise.
40454 (RLNM): New polymorphic function entry.
40455 (VRLWMI): New monomorphic function entry.
40456 (VRLDMI): Likewise.
40457 (RLMI): New polymorphic function entry.
40458 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40459 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40460 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40463 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40466 * dwarf2out.c (field_byte_offset): Restore the
40467 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40468 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40469 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40472 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40475 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40477 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40480 * dwarf2out.c (add_data_member_location_attribute): For constant
40481 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40482 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40483 DW_AT_byte_size attributes.
40485 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40487 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40488 after forcing to constant memory when the code model is medium.
40490 2017-01-17 Julia Koval <julia.koval@intel.com>
40493 * config/i386/avx512fintrin.h
40494 (_mm512_i32gather_ps): Change __addr type to void const*.
40495 (_mm512_mask_i32gather_ps): Ditto.
40496 (_mm512_i32gather_pd): Ditto.
40497 (_mm512_mask_i32gather_pd): Ditto.
40498 (_mm512_i64gather_ps): Ditto.
40499 (_mm512_mask_i64gather_ps): Ditto.
40500 (_mm512_i64gather_pd): Ditto.
40501 (_mm512_mask_i64gather_pd): Ditto.
40502 (_mm512_i32gather_epi32): Ditto.
40503 (_mm512_mask_i32gather_epi32): Ditto.
40504 (_mm512_i32gather_epi64): Ditto.
40505 (_mm512_mask_i32gather_epi64): Ditto.
40506 (_mm512_i64gather_epi32): Ditto.
40507 (_mm512_mask_i64gather_epi32): Ditto.
40508 (_mm512_i64gather_epi64): Ditto.
40509 (_mm512_mask_i64gather_epi64): Ditto.
40510 (_mm512_i32scatter_ps): Change __addr type to void*.
40511 (_mm512_mask_i32scatter_ps): Ditto.
40512 (_mm512_i32scatter_pd): Ditto.
40513 (_mm512_mask_i32scatter_pd): Ditto.
40514 (_mm512_i64scatter_ps): Ditto.
40515 (_mm512_mask_i64scatter_ps): Ditto.
40516 (_mm512_i64scatter_pd): Ditto.
40517 (_mm512_mask_i64scatter_pd): Ditto.
40518 (_mm512_i32scatter_epi32): Ditto.
40519 (_mm512_mask_i32scatter_epi32): Ditto.
40520 (_mm512_i32scatter_epi64): Ditto.
40521 (_mm512_mask_i32scatter_epi64): Ditto.
40522 (_mm512_i64scatter_epi32): Ditto.
40523 (_mm512_mask_i64scatter_epi32): Ditto.
40524 (_mm512_i64scatter_epi64): Ditto.
40525 (_mm512_mask_i64scatter_epi64): Ditto.
40526 * config/i386/avx512pfintrin.h
40527 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40528 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40529 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40530 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40531 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40532 (_mm512_prefetch_i32scatter_ps): Ditto.
40533 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40534 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40535 (_mm512_prefetch_i64scatter_pd): Ditto.
40536 (_mm512_prefetch_i64scatter_ps): Ditto.
40537 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40538 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40539 * config/i386/avx512vlintrin.h
40540 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40541 (_mm_mmask_i32gather_ps): Ditto.
40542 (_mm256_mmask_i32gather_pd): Ditto.
40543 (_mm_mmask_i32gather_pd): Ditto.
40544 (_mm256_mmask_i64gather_ps): Ditto.
40545 (_mm_mmask_i64gather_ps): Ditto.
40546 (_mm256_mmask_i64gather_pd): Ditto.
40547 (_mm_mmask_i64gather_pd): Ditto.
40548 (_mm256_mmask_i32gather_epi32): Ditto.
40549 (_mm_mmask_i32gather_epi32): Ditto.
40550 (_mm256_mmask_i32gather_epi64): Ditto.
40551 (_mm_mmask_i32gather_epi64): Ditto.
40552 (_mm256_mmask_i64gather_epi32): Ditto.
40553 (_mm_mmask_i64gather_epi32): Ditto.
40554 (_mm256_mmask_i64gather_epi64): Ditto.
40555 (_mm_mmask_i64gather_epi64): Ditto.
40556 (_mm256_i32scatter_ps): Change __addr type to void*.
40557 (_mm256_mask_i32scatter_ps): Ditto.
40558 (_mm_i32scatter_ps): Ditto.
40559 (_mm_mask_i32scatter_ps): Ditto.
40560 (_mm256_i32scatter_pd): Ditto.
40561 (_mm256_mask_i32scatter_pd): Ditto.
40562 (_mm_i32scatter_pd): Ditto.
40563 (_mm_mask_i32scatter_pd): Ditto.
40564 (_mm256_i64scatter_ps): Ditto.
40565 (_mm256_mask_i64scatter_ps): Ditto.
40566 (_mm_i64scatter_ps): Ditto.
40567 (_mm_mask_i64scatter_ps): Ditto.
40568 (_mm256_i64scatter_pd): Ditto.
40569 (_mm256_mask_i64scatter_pd): Ditto.
40570 (_mm_i64scatter_pd): Ditto.
40571 (_mm_mask_i64scatter_pd): Ditto.
40572 (_mm256_i32scatter_epi32): Ditto.
40573 (_mm256_mask_i32scatter_epi32): Ditto.
40574 (_mm_i32scatter_epi32): Ditto.
40575 (_mm_mask_i32scatter_epi32): Ditto.
40576 (_mm256_i32scatter_epi64): Ditto.
40577 (_mm256_mask_i32scatter_epi64): Ditto.
40578 (_mm_i32scatter_epi64): Ditto.
40579 (_mm_mask_i32scatter_epi64): Ditto.
40580 (_mm256_i64scatter_epi32): Ditto.
40581 (_mm256_mask_i64scatter_epi32): Ditto.
40582 (_mm_i64scatter_epi32): Ditto.
40583 (_mm_mask_i64scatter_epi32): Ditto.
40584 (_mm256_i64scatter_epi64): Ditto.
40585 (_mm256_mask_i64scatter_epi64): Ditto.
40586 (_mm_i64scatter_epi64): Ditto.
40587 (_mm_mask_i64scatter_epi64): Ditto.
40588 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40589 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40590 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40591 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40592 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40593 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40594 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40595 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40596 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40597 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40598 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40599 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40600 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40601 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40602 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40603 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40604 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40605 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40606 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40607 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40608 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40609 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40610 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40611 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40612 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40613 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40614 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40615 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40616 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40617 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40618 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40619 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40620 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40621 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40622 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40623 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40624 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40625 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40626 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40627 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40628 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40629 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40630 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40631 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40632 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40633 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40634 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40635 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40636 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40637 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40638 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40639 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40640 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40641 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40642 definitions accordingly.
40644 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40645 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40648 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40651 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40654 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40655 hard regs for inner regno.
40657 2017-01-17 Martin Liska <mliska@suse.cz>
40660 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40661 assumption and add comment.
40663 2017-01-17 Nathan Sidwell <nathan@acm.org>
40665 * ipa-visibility.c (localize_node): New function, broken out of ...
40666 (function_and_variable_visibility): ... here. Call it.
40668 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40670 PR middle-end/77445
40671 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40672 correctly set frequency of oudgoing edge.
40673 (duplicate_thread_path): Fix profile updating.
40675 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40678 * configure.ac: Add GCC_BASE_VER.
40679 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40680 version from BASE-VER file.
40681 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40682 (gcc.o): Depend on $(BASEVER).
40683 * common.opt (dumpfullversion): New option.
40684 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40685 * doc/invoke.texi: Document -dumpfullversion.
40686 * doc/install.texi: Document --with-gcc-major-version-only.
40687 * configure: Regenerated.
40689 2017-01-17 Richard Biener <rguenther@suse.de>
40691 PR tree-optimization/71433
40692 * tree-vrp.c (register_new_assert_for): Merge same asserts
40693 on all incoming edges.
40694 (process_assert_insertions_for): Handle insertions at the
40697 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40699 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40700 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40702 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40705 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40708 2017-01-17 Alan Modra <amodra@gmail.com>
40711 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40712 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40713 symbolic stack limit when pic.
40715 2017-01-16 Martin Sebor <msebor@redhat.com>
40717 PR tree-optimization/78608
40718 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40720 2017-01-16 Jeff Law <law@redhat.com>
40723 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40724 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40725 for several include directories that may be relative to sysroot.
40726 * config/i386/x-mingw32 (gplus_includedir): Define.
40727 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40728 (native_system_includedir): Likewise.
40729 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40730 override if TARGET_SYSTEM_ROOT is defined.
40731 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40733 PR tree-optimization/79090
40734 PR tree-optimization/33562
40735 PR tree-optimization/61912
40736 PR tree-optimization/77485
40737 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40738 and computed trims into the dump file.
40740 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40742 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40744 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40747 * gimplify.c (gimplify_init_constructor): If want_value and
40748 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40752 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40753 sequence. Formatting fixes.
40754 (doloop_optimize): Formatting fixes.
40757 * gcc.c (debug_level_greater_than_spec_func): New function.
40758 (static_spec_functions): Add debug-level-gt spec function.
40759 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40761 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40762 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40763 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40764 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40765 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40766 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40768 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40770 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40771 QImode fixups to general and mask registers only.
40773 2017-01-16 Carl Love <cel@us.ibm.com>
40775 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40776 for built-in functions
40777 vector signed char vec_nabs (vector signed char)
40778 vector signed short vec_nabs (vector signed short)
40779 vector signed int vec_nabs (vector signed int)
40780 vector signed long long vec_nabs (vector signed long long)
40781 vector float vec_nabs (vector float)
40782 vector double vec_nabs (vector double)
40783 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40785 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40786 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40787 * doc/extend.texi: Update the documentation file for the new built-in
40790 2017-01-16 Martin Sebor <msebor@redhat.com>
40792 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
40795 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40797 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
40798 UNSPEC_VSX__XXSPLTD to require special splat handling.
40800 2017-01-16 David Malcolm <dmalcolm@redhat.com>
40803 * system.h: Poison strndup.
40805 2017-01-16 Alan Modra <amodra@gmail.com>
40808 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
40811 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
40813 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
40815 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
40817 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40818 call recog here. Assert that INSN_CODE (insn) is non-negative.
40820 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40823 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40825 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40826 in the currently scheduled RTL fragment.
40828 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40830 PR rtl-optimization/78751
40831 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40834 2017-01-14 Jeff Law <law@redhat.com>
40836 PR tree-optimization/79090
40837 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40838 variable length stores.
40839 (compute_trims): Delete dead assignment to *trim_tail.
40840 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
40843 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
40845 PR rtl-optimization/78626
40846 PR rtl-optimization/78727
40847 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
40848 of a block, and split such blocks after everything else is finished.
40850 2017-01-14 Alan Modra <amodra@gmail.com>
40853 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40854 target legitimate_combined_insn.
40855 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
40856 (rs6000_legitimate_combined_insn): New function.
40857 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
40859 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
40860 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
40861 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
40863 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
40865 * doc/frontends.texi (G++ and GCC): Remove references to Java.
40867 2017-01-13 Jeff Law <law@redhat.com>
40869 PR tree-optimization/33562
40870 PR tree-optimization/61912
40871 PR tree-optimization/77485
40872 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
40874 (delete_dead_assignment): Likewise.
40875 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40876 statement to delete_dead_call and delete_dead_assignment.
40878 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40881 * substring-locations.c (format_warning_va): Strengthen case 1 so
40882 that both endpoints of the substring must be within the format
40883 range for just the substring to be printed.
40885 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40887 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40888 * config/i386/i386.c (ix86_target_string): Add missing options
40889 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40890 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40891 flags_other and ix86_target_other to flags2_other. Display unknown
40893 (ix86_valid_target_attribute_inner_p): Add missing options and
40894 reorder options by implied ISAs, as in ix86_target_string.
40896 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40898 * hash-table.h (hash_table::too_empty_p): New function.
40899 (hash_table::expand): Use it.
40900 (hash_table::traverse): Likewise.
40901 (hash_table::empty_slot): Use sizeof (value_type) instead of
40902 sizeof (PTR) to convert bytes to elements. Shrink the table
40903 if the current size is excessive for the current number of
40906 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40908 * ira-costs.c (record_reg_classes): Break from the inner loop
40909 early once alt_fail is known to be true. Update outer loop
40910 handling accordingly.
40912 2017-01-13 Jeff Law <law@redhat.com>
40914 * tree-ssa-dse.c (decrement_count): New function.
40915 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40916 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40917 when we know the partially dead statement is a mem* function.
40919 PR tree-optimization/61912
40920 PR tree-optimization/77485
40921 * tree-ssa-dse.c: Include expr.h.
40922 (maybe_trim_constructor_store): New function.
40923 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40925 PR tree-optimization/33562
40926 PR tree-optimization/61912
40927 PR tree-optimization/77485
40928 * doc/invoke.texi: Document new dse-max-object-size param.
40929 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40930 * tree-ssa-dse.c: Include params.h.
40931 (dse_store_status): New enum.
40932 (initialize_ao_ref_for_dse): New, partially extracted from
40934 (valid_ao_ref_for_dse, normalize_ref): New.
40935 (setup_live_bytes_from_ref, compute_trims): Likewise.
40936 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40937 (maybe_trim_partially_dead_store): Likewise.
40938 (maybe_trim_complex_store): Likewise.
40939 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40940 Track what bytes live from the original store. Return tri-state
40941 for dead, partially dead or live.
40942 (dse_dom_walker): Add constructor, destructor and new private members.
40943 (delete_dead_call, delete_dead_assignment): New extracted from
40945 (dse_optimize_stmt): Make a member of dse_dom_walker.
40946 Use initialize_ao_ref_for_dse.
40948 PR tree-optimization/33562
40949 PR tree-optimization/61912
40950 PR tree-optimization/77485
40951 * sbitmap.h (bitmap_count_bits): Prototype.
40952 (bitmap_clear_range, bitmap_set_range): Likewise.
40953 * sbitmap.c (bitmap_clear_range): New function.
40954 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40956 2017-01-13 Martin Liska <mliska@suse.cz>
40959 * function.c (set_cfun): Add new argument force.
40960 * function.h (set_cfun): Likewise.
40961 * ipa-inline-transform.c (inline_call): Use the function when
40962 strict alising from is dropped for function we inline to.
40964 2017-01-13 Richard Biener <rguenther@suse.de>
40966 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40967 for dumping GIMPLE INTEGER_CSTs.
40969 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40971 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40972 to 201112L since C++17.
40974 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40977 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40978 if -fsanitize=kernel-address is present.
40980 2017-01-13 Richard Biener <rguenther@suse.de>
40982 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40983 as _Literal ( type ) number in case usual suffixes do not
40984 preserve all information.
40986 2017-01-13 Richard Biener <rguenther@suse.de>
40988 PR tree-optimization/77283
40989 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40990 and ssa-iterators.h.
40991 (is_feasible_trace): Implement a cost model based on joiner
40994 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40997 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40998 char or short to __float128/_Float128 directly.
41000 2017-01-12 Martin Sebor <msebor@redhat.com>
41002 to -Wformat-overflow.
41003 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41004 (min_bytes_remaining): Same.
41005 (get_string_length): Same.
41006 (format_string): Same.
41007 (format_directive): Same.
41009 (pass_sprintf_length::handle_gimple_call): Same.
41011 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41013 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41014 info.nowrite calls with no lhs that can't throw. Return bool
41015 whether gsi_remove has been called or not.
41016 (pass_sprintf_length::handle_gimple_call): Return bool whether
41017 try_substitute_return_value called gsi_remove. Formatting fix.
41018 (pass_sprintf_length::execute): Don't use gsi_remove if
41019 handle_gimple_call returned true.
41022 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41023 be removed due to side-effects, don't remove following barrier nor
41024 turn the successor edge into fallthru edge.
41026 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41029 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41030 element-reversing loads and stores as not swappable.
41032 2017-01-12 Nathan Sidwell <nathan@acm.org>
41033 Nicolai Stange <nicstange@gmail.com>
41035 * combine.c (try_combine): Don't ignore result of overlap checking
41036 loop. Combine overlap & asm check into single loop.
41038 2017-01-12 Richard Biener <rguenther@suse.de>
41040 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41041 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41043 2017-01-12 Richard Biener <rguenther@suse.de>
41045 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41046 and TS_TARGET_OPTION directly derive from TS_BASE.
41047 * tree-core.h (tree_optimization_option): Derive from tree_base.
41048 (tree_target_option): Likewise.
41050 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41052 * config/i386/i386.c (memory_address_length): Increase len
41053 only when rip_relative_addr_p returns false.
41055 2017-01-11 Julia Koval <julia.koval@intel.com>
41057 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41058 (OPTION_MASK_ISA_SGX_SET): New.
41059 (ix86_handle_option): Handle OPT_msgx.
41060 * config.gcc: Added sgxintrin.h.
41061 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41062 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41063 * config/i386/i386.c (ix86_target_string): Add -msgx.
41065 (ix86_option_override_internal): Handle new options.
41066 (ix86_valid_target_attribute_inner_p): Add sgx.
41067 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41068 * config/i386/i386.opt: Add msgx.
41069 * config/i386/sgxintrin.h: New file.
41070 * config/i386/x86intrin.h: Add sgxintrin.h.
41072 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41075 * fold-const.c (maybe_nonzero_address): Return 1 for function
41077 (tree_single_nonzero_warnv_p): Don't handle function local objects
41081 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41084 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41087 * opts.c: Include "spellcheck.h"
41088 (struct string_fragment): New struct.
41089 (struct edit_distance_traits<const string_fragment &>): New
41091 (get_closest_sanitizer_option): New function.
41092 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41094 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41096 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41098 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41099 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41100 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41101 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41102 for initial die_offset if dwarf_split_debug_info.
41103 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41104 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41106 (output_skeleton_debug_sections): Formatting fix. Use
41107 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41108 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41110 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41112 * config/arm/cortex-a53.md: Add bypasses for
41113 cortex_a53_r2f_cvt.
41114 (cortex_a53_r2f): Only use for transfers.
41115 (cortex_a53_f2r): Likewise.
41116 (cortex_a53_r2f_cvt): Add reservation for conversions.
41117 (cortex_a53_f2r_cvt): Likewise.
41119 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41121 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41122 to all inlined functions, change static to extern.
41124 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41127 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41129 (arm_assemble_integer): Likewise.
41131 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41133 * config.gcc: Use new awk script to check CPU, FPU and architecture
41134 parameters for --with-... options.
41135 * config/arm/parsecpu.awk: New file
41136 * config/arm/arm-cpus.in: New file.
41137 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41139 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41141 * config/arm/t-arm: Update dependency rules.
41142 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41143 of processing .def files.
41144 * config/arm/genopt.sh: Deleted.
41145 * config/arm/gentune.sh: Deleted.
41146 * config/arm/arm-cores.def: Deleted.
41147 * config/arm/arm-arches.def: Deleted.
41148 * config/arm/arm-fpus.def: Deleted.
41149 * config/arm/arm-tune.md: Regenerated.
41150 * config/arm/arm-tables.opt: Regenerated.
41151 * config/arm/arm-cpu.h: New generated file.
41152 * config/arm/arm-cpu-data.h: New generated file.
41153 * config/arm/arm-cpu-cdata.h: New generated file.
41155 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41158 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41160 (input_varpool_node): Unpack dynamically_initialized bit.
41162 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41164 PR rtl-optimization/79032
41165 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41166 the alignment of the adjusted memory reference against that of MODE,
41167 instead of the alignment of the original memory reference.
41169 2017-01-11 Martin Jambor <mjambor@suse.cz>
41171 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41173 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41174 decorated functions.
41176 2017-01-11 Richard Biener <rguenther@suse.de>
41178 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41179 set range/nonnull info for PHI results. Do not set it on
41180 stmts marked for removal.
41182 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41184 * expr.c (store_field): In the bitfield case, fetch the return value
41185 from the registers before applying a single big-endian adjustment.
41186 Always do a final load for a BLKmode value not larger than a word.
41188 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41191 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41192 that we correctly handle column numbers greater than
41193 LINE_MAP_MAX_COLUMN_NUMBER.
41195 2017-01-10 Martin Sebor <msebor@redhat.com>
41197 PR middle-end/78245
41198 * gimple-ssa-sprintf.c (get_destination_size): Call
41199 {init,fini}object_sizes.
41200 * tree-object-size.c (addr_object_size): Adjust.
41201 (pass_through_call): Adjust.
41202 (pass_object_sizes::execute): Adjust.
41203 * tree-object-size.h (fini_object_sizes): Declare.
41205 2017-01-10 Martin Sebor <msebor@redhat.com>
41207 PR tree-optimization/78775
41208 * builtins.c (get_size_range): Move...
41209 * calls.c: ...to here.
41210 (alloc_max_size): Accept zero argument.
41211 (operand_signed_p): Remove.
41212 (maybe_warn_alloc_args_overflow): Call get_size_range.
41213 * calls.h (get_size_range): Declare.
41215 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41217 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41218 from TI's devices.csv file as of September 2016.
41219 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41221 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41223 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41224 * doc/invoke.texi: Likewise.
41225 * doc/md.texi: Likewise.
41226 * doc/objc.texi: Likewise.
41228 2017-01-10 Joshua Conner <joshconner@google.com>
41230 * config/arm/fuchsia-elf.h: New file.
41231 * config/fuchsia.h: New file.
41232 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41233 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41235 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41237 2016-01-10 Richard Biener <rguenther@suse.de>
41239 PR tree-optimization/79034
41240 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41241 Propagate out degenerate PHIs in the joiner.
41243 2017-01-10 Martin Liska <mliska@suse.cz>
41245 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41246 (sort_congruence_classes_by_decl_uid): Likewise.
41247 (sort_congruence_class_groups_by_decl_uid): Likewise.
41248 (sem_item_optimizer::merge_classes): Sort class, groups in these
41249 classes and members in the groups by DECL_UID of declarations.
41250 This would make merge operations stable.
41252 2017-01-10 Martin Liska <mliska@suse.cz>
41254 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41255 usage of m_classes_vec.
41256 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41257 (sem_item_optimizer::get_group_by_hash): Likewise.
41258 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41259 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41260 (sem_item_optimizer::verify_classes): Likewise.
41261 (sem_item_optimizer::process_cong_reduction): Likewise.
41262 (sem_item_optimizer::dump_cong_classes): Likewise.
41263 (sem_item_optimizer::merge_classes): Likewise.
41264 * ipa-icf.h (congruence_class_hash): Rename from
41265 congruence_class_group_hash. Remove declaration of m_classes_vec.
41267 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41269 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41270 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41271 * config.gcc: Add avx512vpopcntdqintrin.h.
41272 * config/i386/avx512vpopcntdqintrin.h: New.
41273 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41274 * config/i386/i386-builtin-types.def: Add new types.
41275 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41276 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41277 __builtin_ia32_vpopcountq_v8di_mask): New.
41278 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41279 __AVX512VPOPCNTDQ__.
41280 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41281 (PTA_AVX512VPOPCNTDQ): Define.
41282 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41283 TARGET_AVX512VPOPCNTDQ_P): Define.
41284 * config/i386/i386.opt: Add mavx512vpopcntdq.
41285 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41286 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41288 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41290 PR middle-end/77484
41291 * predict.def (PRED_CALL): Set to 67.
41293 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41295 * expr.c (store_field): In the bitfield case, if the value comes from
41296 a function call and is of an aggregate type returned in registers, do
41297 not modify the field mode; extract the value in all cases if the mode
41298 is BLKmode and the size is not larger than a word.
41300 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41303 * config/i386/cpuid.h: Fix undefined behavior.
41305 2017-01-04 Jeff Law <law@redhat.com>
41307 PR tree-optimization/79007
41308 PR tree-optimization/67955
41309 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41310 conservative for pt.null when flag_non_call_exceptions is on.
41312 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41314 PR translation/79019
41315 PR translation/79020
41316 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41317 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41318 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41320 * config/avr/avr.opt (maccumulate-args): Likewise.
41321 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41322 * common.opt (freport-bug): Likewise.
41323 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41324 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41325 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41326 translatable string.
41327 * config/i386/i386.c (function_value_32): Likewise.
41328 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41329 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41331 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41332 * common/config/msp430/msp430-common.c (msp430_handle_option):
41334 * symtab.c (symtab_node::verify_base): Likewise.
41335 * opts.c (set_debug_level): Likewise.
41336 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41337 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41338 missing whitespace to translatable strings.
41339 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41340 * config/sh/superh.h: Likewise.
41341 * config/i386/xopintrin.h: Likewise.
41342 * config/i386/znver1.md: Likewise.
41343 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41344 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41345 * double-int.h (struct double_int): Likewise.
41346 * double-int.c (div_and_round_double): Likewise.
41347 * wide-int.cc: Likewise.
41348 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41349 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41350 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41351 (crossjumps_occurred): ... this.
41352 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41355 PR tree-optimization/78899
41356 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41357 returning bool return struct loop *, NULL for failure and the new
41359 (versionable_outer_loop_p): Don't version outer loop if it has
41360 dont_vectorized bit set.
41361 (tree_if_conversion): When versioning outer loop, ensure
41362 tree_if_conversion is performed also on the inner loop of the
41363 non-vectorizable outer loop copy.
41364 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41365 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41366 prevent vectorization of it.
41367 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41368 the outer loop vectorization of the non-scalar version is attempted
41369 before vectorization of the inner loop in scalar version. If
41370 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41371 vectorization of its inner loop.
41372 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41373 has 2 inner loops, rename also on edges from bb whose single pred
41374 is outer_loop->header. Fix typo in function comment.
41376 2017-01-09 Martin Sebor <msebor@redhat.com>
41379 * asan.c (asan_emit_stack_protection): Increase local buffer size
41380 to avoid snprintf truncation warning.
41382 2017-01-09 Andrew Pinski <apinski@cavium.com>
41384 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41385 to reference thunderx2t99 for the tuning structure
41386 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41388 (thunderx2t99_extra_costs): This.
41389 * config/aarch64/aarch64-tune.md: Regenerate.
41390 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41391 (vulcan_addrcost_table): This.
41392 (vulcan_regmove_cost): Rename to ...
41393 (thunderx2t99_regmove_cost): This.
41394 (vulcan_vector_cost): Rename to ...
41395 (thunderx2t99_vector_cost): this.
41396 (vulcan_branch_cost): Rename to ...
41397 (thunderx2t99_branch_cost): This.
41398 (vulcan_tunings): Rename to ...
41399 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41400 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41402 2017-01-09 Martin Jambor <mjambor@suse.cz>
41406 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41407 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41408 (propagate_vr_accross_jump_function): Use the above function for all
41409 value range computations for pass-through jump functions and type
41410 converasion from explicit value range values.
41411 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41412 parameters from TYPE_ARG_TYPES.
41413 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41414 (ipa_write_node_info): Stream type of the actual argument.
41415 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41417 2017-01-09 Martin Liska <mliska@suse.cz>
41420 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41421 (lookup_compiler): Do not show error message with have_E.
41423 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41425 PR tree-optimization/78938
41426 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41427 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41428 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41429 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41432 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41434 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41437 2017-01-09 Richard Biener <rguenther@suse.de>
41439 PR tree-optimization/78997
41440 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41441 name condition properly.
41443 2017-01-09 Richard Biener <rguenther@suse.de>
41446 * dwarf2out.c (is_cxx): New overload with context.
41447 (is_naming_typedef_decl): Use it.
41449 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41451 * invoke.texi (Option Summary): Correct spacing in option lists
41452 and add line breaks to fix over-long lines.
41454 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41456 PR middle-end/17660
41458 * extend.texi (Common Variable Attributes): Add xref to GCC
41459 Internals manual to explain mode attribute keywords.
41461 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41464 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41465 and Preprocessor Options.
41466 (Options for Linking): Document -pthread here....
41467 (RS/6000 and PowerPC Options): ...not here.
41468 (Solaris 2 Options): ...or here.
41469 * doc/cppopts.texi: Document -pthread.
41471 2017-01-08 Martin Sebor <msebor@redhat.com>
41473 PR middle-end/77708
41474 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41475 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41476 New member functions.
41477 (format_directive): Used them.
41479 (pass_sprintf_length::handle_gimple_call): Same.
41480 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41481 to avoid truncation for any argument.
41482 (extract_affine_mul): Same.
41483 * tree.c (get_file_function_name): Same.
41485 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41487 PR middle-end/77484
41488 * predict.def (PRED_INDIR_CALL): Set to 86.
41490 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41492 PR preprocessor/54124
41493 * doc/cppopts.texi: Reformat -d subtable to list the full name
41494 of the options. Add cross-reference to the docs for the general
41495 compiler -d options.
41496 * doc/invoke.texi (Developer Options): Add cross-reference to the
41497 preprocessor-specific -d option documentation.
41499 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41501 PR preprocessor/13498
41502 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41503 redudant material, and reflect new command-line options.
41504 (System Headers): Likewise.
41506 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41508 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41509 -isystem, and -idirafter. Copy-edit.
41510 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41511 default for -ftrack-macro-expansion. Delete obsolete and
41512 badly-formatted implementation details about -fdebug-cpp output.
41513 * doc/cppwarnopts.texi: Copy-edit.
41515 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41518 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41519 that the transition from a max line width >= 1<<10 to narrower
41520 lines works correctly.
41522 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41524 * doc/options.texi (PerFunction): New.
41525 * opt-functions.awk (switch_flags): Map both Optimization and
41526 PerFunction to CL_OPTIMIZATION.
41527 * opth-gen.awk: Test for PerFunction flag along with
41529 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41530 it only when the latter is present. Skip those that don't in
41531 the hash function generator.
41532 * common.opt (fvar-tracking): Mark as PerFunction instead of
41534 (fvar-tracking-assignments): Likewise.
41535 (fvar-tracking-assignments-toggle): Likewise.
41536 (fvar-tracking-uninit): Likewise.
41538 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41540 PR translation/79018
41541 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41544 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41547 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41548 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41549 TARGET_LONG_JUMP_TABLE_OFFSETS.
41550 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41551 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41552 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41553 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41554 * config/m68k/m68k.md (tablejump expander): Likewise.
41555 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41556 TARGET_LONG_JUMP_TABLE_OFFSETS.
41557 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41558 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41560 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41561 David Holsgrove <david.holsgrove@xilinx.com>
41563 * common/config/microblaze/microblaze-common.c
41564 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41565 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41567 * config/microblaze/microblaze.c (microblaze_must_save_register)
41568 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41570 (microblaze_eh_return): New function.
41571 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41572 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41573 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41574 * config/microblaze/microblaze.md (eh_return): New pattern.
41576 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41578 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41579 GCC_DIAGNOSTIC_STRINGIFY): Define.
41581 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41583 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41585 * config/arm/arm.md (<mcrr>): New.
41587 * config/arm/arm.c (arm_arch5te): New.
41588 (arm_option_override): Set arm_arch5te.
41589 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41591 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41592 (arm_mcrr_qualifiers): ... this. New.
41593 (MRRC_QUALIFIERS): Define to...
41594 (arm_mrrc_qualifiers): ... this. New.
41595 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41597 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41598 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41599 (MRRCI, mrrc, MRRC): New.
41600 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41601 VUNSPEC_MRRC2): New.
41603 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41605 * config/arm/arm.md (<mcr>): New.
41607 * config/arm/arm.c (arm_coproc_builtin_available): Add
41608 support for mcr, mrc, mcr2 and mrc2.
41609 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41610 (arm_mcr_qualifiers): ... this. New.
41611 (MRC_QUALIFIERS): Define to ...
41612 (arm_mrc_qualifiers): ... this. New.
41613 (MCR_QUALIFIERS): Define to ...
41614 (arm_mcr_qualifiers): ... this. New.
41615 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41617 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41618 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41619 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41620 VUNSPEC_MRC2): New.
41622 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41624 * config/arm/arm.md (*ldc): New.
41628 * config/arm/arm.c (arm_coproc_builtin_available): Add
41629 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41630 (arm_coproc_ldc_stc_legitimate_address): New.
41631 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41632 'qualifier_const_pointer'.
41633 (LDC_QUALIFIERS): Define to...
41634 (arm_ldc_qualifiers): ... this. New.
41635 (STC_QUALIFIERS): Define to...
41636 (arm_stc_qualifiers): ... this. New.
41637 * config/arm/arm-protos.h
41638 (arm_coproc_ldc_stc_legitimate_address): New.
41639 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41640 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41641 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41642 stc2, stcl, stc2l): New.
41643 * config/arm/constraints.md (Uz): New.
41644 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41645 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41646 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41647 VUNSPEC_STC2L): New.
41649 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41651 * config/arm/arm.md (<cdp>): New.
41652 * config/arm/arm.c (neon_const_bounds): Rename this ...
41653 (arm_const_bounds): ... this.
41654 (arm_coproc_builtin_available): New.
41655 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41656 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41657 (CDP_QUALIFIERS): Define to...
41658 (arm_cdp_qualifiers): ... this. New.
41660 (arm_expand_builtin_args): Add case for 6 arguments.
41661 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41662 (arm_const_bounds): ... this.
41663 (arm_coproc_builtin_available): New.
41664 * config/arm/arm_acle.h (__arm_cdp): New.
41666 * config/arm/arm_acle_builtins.def (cdp): New.
41668 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41669 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41670 'arm_const_bounds'.
41671 * config/arm/types.md (coproc): New.
41672 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41673 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41674 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41675 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41677 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41679 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41680 (UBINOP_QUALIFIERS): New.
41682 (acle_builtin_data): New. Change comment.
41683 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41684 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41685 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41686 arm_acle_builtins.def.
41687 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41688 (arm_init_acle_builtins): New.
41689 (CRC32_BUILTIN): Remove.
41690 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41691 crc32cb, crc32ch and crc32cw.
41692 (arm_init_crc32_builtins): Remove.
41693 (arm_init_builtins): Use arm_init_acle_builtins rather
41694 than arm_init_crc32_builtins.
41695 (arm_expand_acle_builtin): New.
41696 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41697 * config/arm/arm_acle_builtins.def: New.
41699 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41701 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41702 (arm_builtin_datum): ... this.
41703 (arm_init_neon_builtin): Rename to ...
41704 (arm_init_builtin): ... this. Add a new parameters PREFIX
41705 and USE_SIG_IN_NAME.
41706 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41707 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41708 'arm_builtin_datum'.
41709 (arm_init_vfp_builtins): Likewise.
41710 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41711 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41712 (arm_expand_neon_args): Rename to ...
41713 (arm_expand_builtin_args): ... this. Rename builtin_arg
41714 enum values and differentiate between ARG_BUILTIN_MEMORY
41715 and ARG_BUILTIN_NEON_MEMORY.
41716 (arm_expand_neon_builtin_1): Rename to ...
41717 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41718 values, arm_expand_builtin_args and add bool parameter NEON.
41719 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41720 (arm_expand_vfp_builtin): Likewise.
41721 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41723 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41725 PR middle-end/77484
41726 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41727 * predict.c (tree_estimate_probability_bb): Reverse direction of
41728 polymorphic call predictor.
41730 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41732 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41733 (determine_pass_name_match): ...this new function and...
41734 (should_skip_pass_p): ...this new function.
41736 2017-01-06 Nathan Sidwell <nathan@acm.org>
41738 * ipa-visibility.c (function_and_variable_visibility): Reformat
41739 comments and long lines. Remove extrneous if.
41740 * symtab.c (symtab_node::make_decl_local): Fix code format.
41741 (symtab_node::set_section_for_node): Fix comment typo.
41743 2017-01-06 Martin Liska <mliska@suse.cz>
41746 * lra-constraints.c: Rename invariant to lra_invariant.
41747 * predict.c (set_even_probabilities): Initialize e to NULL.
41749 2017-01-05 Martin Sebor <msebor@redhat.com>
41751 PR tree-optimization/78910
41752 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41753 (format_integer): Correct off-by-one error in the handling
41754 of precision with negative numbers in signed conversions..
41756 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41758 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41760 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41762 PR tree-optimization/71016
41763 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41764 factor_out_conditional_conversion. Formatting fix.
41765 (factor_out_conditional_conversion): Add cond_stmt argument.
41766 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41767 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41770 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41772 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41773 read-rtl-function.o, and selftest-rtl.o.
41774 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41775 (selftest::aarch64_test_loading_full_dump): New function.
41776 (selftest::aarch64_run_selftests): New function.
41777 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41778 selftest::aarch64_run_selftests.
41779 * config/i386/i386.c
41780 (selftest::ix86_test_loading_dump_fragment_1): New function.
41781 (selftest::ix86_test_loading_call_insn): New function.
41782 (selftest::ix86_test_loading_full_dump): New function.
41783 (selftest::ix86_test_loading_unspec): New function.
41784 (selftest::ix86_run_selftests): Call the new functions.
41785 * emit-rtl.c (maybe_set_max_label_num): New function.
41786 * emit-rtl.h (maybe_set_max_label_num): New decl.
41787 * function.c (instantiate_decls): Guard call to
41788 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41789 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
41791 * gensupport.c (gen_reader::gen_reader): Pass "false"
41792 for new "compact" param of rtx_reader.
41793 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
41794 rather than an empty string for NULL strings.
41795 * read-md.c: Potentially include config.h rather than bconfig.h.
41796 Wrap include of errors.h with #ifdef GENERATOR_FILE.
41797 (have_error): New global, copied from errors.c.
41798 (md_reader::read_name): Rename to...
41799 (md_reader::read_name_1): ...this, adding "out_loc" param,
41800 and converting "missing name or number" to returning false, rather
41802 (md_reader::read_name): Reimplement in terms of read_name_1.
41803 (md_reader::read_name_or_nil): New function.
41804 (md_reader::read_string): Handle "(nil)" by returning NULL.
41805 (md_reader::md_reader): Add new param "compact".
41806 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
41807 (md_reader::read_file): New method.
41808 * read-md.h (md_reader::md_reader): Add new param "compact".
41809 (md_reader::read_file): New method.
41810 (md_reader::is_compact): New accessor.
41811 (md_reader::read_name): Convert return type from void to file_location.
41812 (md_reader::read_name_or_nil): New decl.
41813 (md_reader::read_name_1): New decl.
41814 (md_reader::m_compact): New field.
41815 (noop_reader::noop_reader): Pass "false" for new "compact" param
41817 (rtx_reader::rtx_reader): Add new "compact" param.
41818 (rtx_reader::read_rtx_operand): Make virtual and convert return
41819 type from void to rtx.
41820 (rtx_reader::read_until): New decl.
41821 (rtx_reader::handle_any_trailing_information): New virtual function.
41822 (rtx_reader::postprocess): New virtual function.
41823 (rtx_reader::finalize_string): New virtual function.
41824 (rtx_reader::m_in_call_function_usage): New field.
41825 (rtx_reader::m_reuse_rtx_by_id): New field.
41826 * read-rtl-function.c: New file.
41827 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41828 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41829 (selftest::verify_three_block_rtl_cfg): New decl.
41830 * read-rtl-function.h: New file.
41831 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41832 For host, include function.h, memmodel.h, and emit-rtl.h.
41833 (one_time_initialization): New function.
41834 (struct compact_insn_name): New struct.
41835 (compact_insn_names): New array.
41836 (find_code): Handle insn codes in compact dumps.
41837 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41838 (bind_subst_iter_and_attr): Likewise.
41839 (add_condition_to_string): Likewise.
41840 (add_condition_to_rtx): Likewise.
41841 (apply_attribute_uses): Likewise.
41842 (add_current_iterators): Likewise.
41843 (apply_iterators): Likewise.
41844 (initialize_iterators): Guard usage of apply_subst_iterator with
41845 #ifdef GENERATOR_FILE.
41846 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
41847 (md_reader::read_mapping): Likewise.
41848 (add_define_attr_for_define_subst): Likewise.
41849 (add_define_subst_attr): Likewise.
41850 (read_subst_mapping): Likewise.
41851 (check_code_iterator): Likewise.
41852 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41854 (one_time_initialization): New function.
41855 (rtx_reader::read_until): New method.
41856 (read_flags): New function.
41857 (parse_reg_note_name): New function.
41858 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
41859 Handle reuse_rtx ids.
41860 Wrap iterator lookup within #ifdef GENERATOR_FILE.
41861 Add parsing support for RTL dumps, mirroring the special-cases in
41862 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
41863 values, and calling handle_any_trailing_information.
41864 (rtx_reader::read_rtx_operand): Convert return type from void
41865 to rtx, returning return_rtx. Handle case 'e'. Call
41866 finalize_string on XSTR and XTMPL fields.
41867 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
41868 "(nil)" values were omitted. Call the postprocess vfunc on the
41870 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
41871 class ctor. Initialize m_in_call_function_usage. Call
41872 one_time_initialization.
41873 * rtl-tests.c (selftest::test_uncond_jump): Call
41874 set_new_first_and_last_insn.
41875 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41876 * selftest-rtl.c: New file.
41877 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41878 (selftest::get_insn_by_uid): New decl.
41879 * selftest-run-tests.c (selftest::run_tests): Call
41880 read_rtl_function_c_tests.
41881 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41882 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41885 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41887 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41888 operands in a special way. Assert that pos+len <= mode precision.
41890 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41892 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41893 3 argument Alias with unlimited for the negative form.
41894 (fno-vect-cost-model): Removed.
41896 2017-01-05 Martin Liska <mliska@suse.cz>
41898 * hsa-gen.c (gen_hsa_divmod): New function.
41899 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41901 2017-01-05 Martin Liska <mliska@suse.cz>
41904 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41907 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41909 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41910 small constant length operands.
41912 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41914 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41915 between loop iterations.
41917 2017-01-05 Martin Liska <mliska@suse.cz>
41920 * gimplify.c (gimplify_decl_expr): Compare to
41921 asan_poisoned_variables instread of checking flags.
41922 (gimplify_target_expr): Likewise.
41923 (gimplify_expr): Likewise.
41924 (gimplify_function_tree): Conditionally initialize
41925 asan_poisoned_variables.
41927 2017-01-04 Jeff Law <law@redhat.com>
41929 PR tree-optimizatin/78812
41930 * rtl.h (contains_mem_rtx_p): Prototype.
41931 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41932 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41933 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41934 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41935 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41937 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41939 * input.c (assert_char_at_range): Default-initialize actual_range.
41941 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41943 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41944 to match the caller.
41946 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41948 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41949 insns after final jump in test to emit dummy move.
41951 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41953 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41954 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41956 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41958 * multiple_target.c (create_dispatcher_calls): Init e_next.
41959 * tree-ssa-loop-split.c (split_loop): Init border.
41960 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41963 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41968 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41969 (altivec_register_operand): Do not return true if the operand
41970 contains a SUBREG mixing SImode and SFmode.
41971 (vsx_register_operand): Likewise.
41972 (vsx_reg_sfsubreg_ok): New predicate.
41973 (vfloat_operand): Do not return true if the operand contains a
41974 SUBREG mixing SImode and SFmode.
41975 (vint_operand): Likewise.
41976 (vlogical_operand): Likewise.
41977 (gpc_reg_operand): Likewise.
41978 (int_reg_operand): Likewise.
41979 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41980 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41981 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41983 (rs6000_emit_move_si_sf_subreg): New helper function.
41984 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41985 fixup SUBREGs involving SImode and SFmode.
41986 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41987 numbers for the new peephole2 optimization.
41988 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41989 the GLIBC math library that do AND/IOR/XOR operations on single
41990 precision floating point.
41991 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41992 target macros to say whether we need to avoid SUBREGs mixing
41994 (TARGET_ALLOW_SF_SUBREG): Likewise.
41995 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41996 (UNSPEC_SI_FROM_SF): Likewise.
41997 (iorxor): Change spacing.
41998 (and_ior_xor): New iterator for AND, IOR, and XOR.
41999 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42000 (movdi_from_sf_zero_ext): Likewise.
42001 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42002 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42003 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42004 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42005 (fms<mode>4): Likewise.
42006 (fnma<mode>4): Likewise.
42007 (fnms<mode>4): Likewise.
42008 (nfma<mode>4): Likewise.
42009 (nfms<mode>4): Likewise.
42011 2017-01-04 Marek Polacek <polacek@redhat.com>
42014 * doc/invoke.texi: Document -Wpointer-compare.
42016 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42018 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42021 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42022 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42025 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42028 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42029 documentation of the powerpc_popcntb_ok attribute.
42030 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42031 code to issue warning messages if a requested CPU configuration is
42032 not supported by the binary (assembler and loader) toolchain.
42033 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42034 made to define a built-in function that has been disabled.
42035 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42036 made to define a built-in function that has been disabled.
42037 (altivec_init_builtins): Add comment explaining why definition
42038 of the DST built-in functions is not preceded by an assertion
42039 check. Add assertions to prevent ICE if attempts are made to
42040 define an altivec predicate or an abs* built-in function that has
42042 (htm_init_builtins): Add comment explaining why definition of the
42043 htm built-in functions is not preceded by an assertion check.
42045 2017-01-04 Jeff Law <law@redhat.com>
42047 PR tree-optimizatin/67955
42048 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42049 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42050 the points-to solution does not include pt_null. Use DECL_PT_UID
42053 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42055 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42056 Use gen_int_mode instead of gen_lopwart for const_int operands.
42058 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42060 PR tree-optimization/71563
42061 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42062 out of range value - has low bits known to be zero.
42064 2017-01-04 Alan Modra <amodra@gmail.com>
42066 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42067 * configure: Regenerate.
42068 * config.in: Regenerate.
42070 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42073 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42074 a substring of the message, but strcmp with the whole message. Ifdef
42075 ENABLE_NLS, translate the message first using dgettext.
42077 2017-01-03 Jeff Law <law@redhat.com>
42079 PR tree-optimizatin/78856
42080 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42081 (mark_threaded_blocks): Remove code to truncate thread paths that
42082 cross multiple loop headers. Instead invalidate the cached loop
42083 iteration information and handle case of a thread path walking
42084 into an irreducible region.
42086 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42089 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42090 assertions. Add support for doing the signbit if the IEEE 128-bit
42091 floating point value is in a GPR.
42092 * config/rs6000/rs6000.md (Fsignbit): Delete.
42093 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42094 Update the length attribute if the value is in a GPR.
42095 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42096 the sign or zero extension instruction, since the value is always 0/1.
42097 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42100 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42101 extracting SImode to a GPR register so that we can generate a
42102 store, limit the vector to be in a traditional Altivec register
42103 for the vextuwrx instruction.
42105 2017-01-03 Ian Lance Taylor <iant@google.com>
42107 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42109 2017-01-03 Martin Sebor <msebor@redhat.com>
42111 PR tree-optimization/78696
42112 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42113 precision. Use MPFR for %f for greater fidelity. Correct handling
42115 (pass_sprintf_length::compute_format_length): Set width and precision
42116 specified by asrerisk to void_node for vararg functions.
42117 (try_substitute_return_value): Adjust dump output.
42119 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42121 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42123 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42125 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42126 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42127 -mlra/-mno-lra was passed to the compiler.
42129 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42131 PR rtl-optimization/65618
42132 * emit-rtl.c (try_split): Move initialization of "before" and
42133 "after" to just before the call to emit_insn_after_setloc.
42135 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42137 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42139 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42141 * dwarf2out.c (gen_enumeration_type_die): When
42142 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42144 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42146 PR tree-optimization/78965
42147 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42148 Change first argument from const call_info & to call_info &. For %n
42149 set info.nowrite to false.
42151 PR middle-end/78901
42152 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42153 possibly throwing calls.
42155 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42156 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42157 and fns handling, rather than in a separate case SSA_NAME.
42159 2017-01-02 Jeff Law <law@redhat.com>
42161 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42162 fixes for first_period and second_period variables.
42164 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42167 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42168 (*insvqi_1): New insn pattern.
42169 (*insvqi_1_mem_rex64): Ditto.
42170 (*insvqi_2): Ditto.
42171 (*insvqi_3): Rename from *insvqi.
42173 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42175 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42177 * doc/cfg.texi (Edges): Remove reference to Java.
42178 (Maintaining the CFG): Ditto.
42180 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42182 PR middle-end/77674
42183 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42184 transparent aliases.
42186 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42188 PR middle-end/77484
42189 * predict.def (PRED_CALL): Update hitrate.
42190 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42191 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42192 into direct/indirect/polymorphic variants.
42194 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42196 Update copyright years.
42198 * gcc.c (process_command): Update copyright notice dates.
42199 * gcov-dump.c (print_version): Ditto.
42200 * gcov.c (print_version): Ditto.
42201 * gcov-tool.c (print_version): Ditto.
42202 * gengtype.c (create_file): Ditto.
42203 * doc/cpp.texi: Bump @copying's copyright year.
42204 * doc/cppinternals.texi: Ditto.
42205 * doc/gcc.texi: Ditto.
42206 * doc/gccint.texi: Ditto.
42207 * doc/gcov.texi: Ditto.
42208 * doc/install.texi: Ditto.
42209 * doc/invoke.texi: Ditto.
42211 Copyright (C) 2017 Free Software Foundation, Inc.
42213 Copying and distribution of this file, with or without modification,
42214 are permitted in any medium without royalty provided the copyright
42215 notice and this notice are preserved.