1 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
3 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
4 before testing each element individually.
5 * tree-vect-generic.c (lower_vec_perm): Likewise.
7 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
9 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
10 * selftest-run-tests.c (selftest::run_tests): Call it.
11 * vector-builder.h (vector_builder::operator ==): New function.
12 (vector_builder::operator !=): Likewise.
13 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
14 (vec_perm_indices::all_from_input_p): New function.
15 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
16 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
17 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
18 instead of reading the VECTOR_CST directly. Detect whether both
19 vector inputs are the same before constructing the vec_perm_indices,
20 and update the number of inputs argument accordingly. Use the
21 utility functions added above. Only construct sel2 if we need to.
23 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
26 the broadcast of the low byte.
27 (expand_mult_highpart): Use an explicit encoding for the permutes.
28 * optabs-query.c (can_mult_highpart_p): Likewise.
29 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
30 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
31 (vectorizable_bswap): Likewise.
32 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
33 explicit encoding for the power-of-2 permutes.
34 (vect_permute_store_chain): Likewise.
35 (vect_grouped_load_supported): Likewise.
36 (vect_permute_load_chain): Likewise.
38 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
40 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
41 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
42 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
43 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
44 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
45 (vect_gen_perm_mask_any): Likewise.
47 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
49 * int-vector-builder.h: New file.
50 * vec-perm-indices.h: Include int-vector-builder.h.
51 (vec_perm_indices): Redefine as an int_vector_builder.
52 (auto_vec_perm_indices): Delete.
53 (vec_perm_builder): Redefine as a stand-alone class.
54 (vec_perm_indices::vec_perm_indices): New function.
55 (vec_perm_indices::clamp): Likewise.
56 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
57 (vec_perm_indices::new_vector): New function.
58 (vec_perm_indices::new_expanded_vector): Update for new
59 vec_perm_indices class.
60 (vec_perm_indices::rotate_inputs): New function.
61 (vec_perm_indices::all_in_range_p): Operate directly on the
62 encoded form, without computing elided elements.
63 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
64 encoding. Update for new vec_perm_indices class.
65 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
66 the given vec_perm_builder.
67 (expand_vec_perm_var): Update vec_perm_builder constructor.
68 (expand_mult_highpart): Use vec_perm_builder instead of
69 auto_vec_perm_indices.
70 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
71 vec_perm_indices instead of auto_vec_perm_indices. Use a single
72 or double series encoding as appropriate.
73 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
74 vec_perm_indices instead of auto_vec_perm_indices.
75 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
76 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
77 (vect_permute_store_chain): Likewise.
78 (vect_grouped_load_supported): Likewise.
79 (vect_permute_load_chain): Likewise.
80 (vect_shift_permute_load_chain): Likewise.
81 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
82 (vect_transform_slp_perm_load): Likewise.
83 (vect_schedule_slp_instance): Likewise.
84 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
85 (vectorizable_mask_load_store): Likewise.
86 (vectorizable_bswap): Likewise.
87 (vectorizable_store): Likewise.
88 (vectorizable_load): Likewise.
89 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
90 vec_perm_indices instead of auto_vec_perm_indices. Use
91 tree_to_vec_perm_builder to read the vector from a tree.
92 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
93 vec_perm_builder instead of a vec_perm_indices.
94 (have_whole_vector_shift): Use vec_perm_builder and
95 vec_perm_indices instead of auto_vec_perm_indices. Leave the
96 truncation to calc_vec_perm_mask_for_shift.
97 (vect_create_epilog_for_reduction): Likewise.
98 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
99 from auto_vec_perm_indices to vec_perm_indices.
100 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
101 instead of changing individual elements.
102 (aarch64_vectorize_vec_perm_const): Use new_vector to install
103 the vector in d.perm.
104 * config/arm/arm.c (expand_vec_perm_d::perm): Change
105 from auto_vec_perm_indices to vec_perm_indices.
106 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
107 instead of changing individual elements.
108 (arm_vectorize_vec_perm_const): Use new_vector to install
109 the vector in d.perm.
110 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
111 Update vec_perm_builder constructor.
112 (rs6000_expand_interleave): Likewise.
113 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
114 (rs6000_expand_interleave): Likewise.
116 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
118 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
119 to qimode could truncate the indices.
120 * optabs.c (expand_vec_perm_var): Likewise.
122 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
124 * Makefile.in (OBJS): Add vec-perm-indices.o.
125 * vec-perm-indices.h: New file.
126 * vec-perm-indices.c: Likewise.
127 * target.h (vec_perm_indices): Replace with a forward class
129 (auto_vec_perm_indices): Move to vec-perm-indices.h.
130 * optabs.h: Include vec-perm-indices.h.
131 (expand_vec_perm): Delete.
132 (selector_fits_mode_p, expand_vec_perm_var): Declare.
133 (expand_vec_perm_const): Declare.
134 * target.def (vec_perm_const_ok): Replace with...
135 (vec_perm_const): ...this new hook.
136 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
137 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
138 * doc/tm.texi: Regenerate.
139 * optabs.def (vec_perm_const): Delete.
140 * doc/md.texi (vec_perm_const): Likewise.
141 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
142 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
143 expand_vec_perm for constant permutation vectors. Assert that
144 the mode of variable permutation vectors is the integer equivalent
145 of the mode that is being permuted.
146 * optabs-query.h (selector_fits_mode_p): Declare.
147 * optabs-query.c: Include vec-perm-indices.h.
148 (selector_fits_mode_p): New function.
149 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
150 is defined, instead of checking whether the vec_perm_const_optab
151 exists. Use targetm.vectorize.vec_perm_const instead of
152 targetm.vectorize.vec_perm_const_ok. Check whether the indices
153 fit in the vector mode before using a variable permute.
154 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
155 vec_perm_indices instead of an rtx.
156 (expand_vec_perm): Replace with...
157 (expand_vec_perm_const): ...this new function. Take the selector
158 as a vec_perm_indices rather than an rtx. Also take the mode of
159 the selector. Update call to shift_amt_for_vec_perm_mask.
160 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
161 Use vec_perm_indices::new_expanded_vector to expand the original
162 selector into bytes. Check whether the indices fit in the vector
163 mode before using a variable permute.
164 (expand_vec_perm_var): Make global.
165 (expand_mult_highpart): Use expand_vec_perm_const.
166 * fold-const.c: Includes vec-perm-indices.h.
167 * tree-ssa-forwprop.c: Likewise.
168 * tree-vect-data-refs.c: Likewise.
169 * tree-vect-generic.c: Likewise.
170 * tree-vect-loop.c: Likewise.
171 * tree-vect-slp.c: Likewise.
172 * tree-vect-stmts.c: Likewise.
173 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
175 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
176 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
177 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
178 (aarch64_vectorize_vec_perm_const): ...this new function.
179 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
180 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
181 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
182 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
183 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
184 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
185 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
187 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
188 check for NEON modes.
189 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
190 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
191 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
192 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
194 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
195 the old VEC_PERM_CONST conditions.
196 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
197 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
198 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
199 (ia64_vectorize_vec_perm_const_ok): Merge into...
200 (ia64_vectorize_vec_perm_const): ...this new function.
201 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
202 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
203 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
204 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
205 * config/mips/mips.c (mips_expand_vec_perm_const)
206 (mips_vectorize_vec_perm_const_ok): Merge into...
207 (mips_vectorize_vec_perm_const): ...this new function.
208 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
209 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
210 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
211 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
212 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
213 (rs6000_expand_vec_perm_const): Delete.
214 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
216 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
217 (altivec_expand_vec_perm_const_le): Take each operand individually.
218 Operate on constant selectors rather than rtxes.
219 (altivec_expand_vec_perm_const): Likewise. Update call to
220 altivec_expand_vec_perm_const_le.
221 (rs6000_expand_vec_perm_const): Delete.
222 (rs6000_vectorize_vec_perm_const_ok): Delete.
223 (rs6000_vectorize_vec_perm_const): New function.
224 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
225 an element count and rtx array.
226 (rs6000_expand_extract_even): Update call accordingly.
227 (rs6000_expand_interleave): Likewise.
228 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
229 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
230 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
231 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
232 (rs6000_expand_vec_perm_const): Delete.
233 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
234 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
235 (altivec_expand_vec_perm_const_le): Take each operand individually.
236 Operate on constant selectors rather than rtxes.
237 (altivec_expand_vec_perm_const): Likewise. Update call to
238 altivec_expand_vec_perm_const_le.
239 (rs6000_expand_vec_perm_const): Delete.
240 (rs6000_vectorize_vec_perm_const_ok): Delete.
241 (rs6000_vectorize_vec_perm_const): New function. Remove stray
242 reference to the SPE evmerge intructions.
243 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
244 an element count and rtx array.
245 (rs6000_expand_extract_even): Update call accordingly.
246 (rs6000_expand_interleave): Likewise.
247 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
248 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
250 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
252 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
254 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
255 vector mode and that that mode matches the mode of the data
257 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
258 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
259 directly using expand_vec_perm_1 when forcing selectors into
261 (expand_vec_perm_var): New function, split out from expand_vec_perm.
263 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
265 * optabs-query.h (can_vec_perm_p): Delete.
266 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
267 * optabs-query.c (can_vec_perm_p): Split into...
268 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
269 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
270 particular selector is valid.
271 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
272 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
273 (vect_grouped_load_supported): Likewise.
274 (vect_shift_permute_load_chain): Likewise.
275 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
276 (vect_transform_slp_perm_load): Likewise.
277 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
278 (vectorizable_bswap): Likewise.
279 (vect_gen_perm_mask_checked): Likewise.
280 * fold-const.c (fold_ternary_loc): Likewise. Don't take
281 implementations of variable permutation vectors into account
282 when deciding which selector to use.
283 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
284 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
285 with a false third argument.
286 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
287 to test whether the constant selector is valid and can_vec_perm_var_p
288 to test whether a variable selector is valid.
290 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
292 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
293 * optabs-query.c (can_vec_perm_p): Likewise.
294 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
295 instead of vec_perm_indices.
296 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
297 (vect_gen_perm_mask_checked): Likewise,
298 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
299 (vect_gen_perm_mask_checked): Likewise,
301 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
303 * optabs-query.h (qimode_for_vec_perm): Declare.
304 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
305 (qimode_for_vec_perm): ...this new function.
306 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
308 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
310 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
311 does not have a conditional at the top.
313 2018-01-02 Richard Biener <rguenther@suse.de>
315 * ipa-inline.c (big_speedup_p): Fix expression.
317 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
320 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
323 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
327 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
328 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
329 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
330 cond_taken_branch_cost 3->4.
332 2018-01-01 Jakub Jelinek <jakub@redhat.com>
334 PR tree-optimization/83581
335 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
336 TODO_cleanup_cfg if any changes have been made.
339 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
340 convert_modes if target mode has the right side, but different mode
344 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
345 last argument when extracting from CONCAT. If either from_real or
346 from_imag is NULL, use expansion through memory. If result is not
347 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
348 the parts directly to inner mode, if even that fails, use expansion
352 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
353 check for bswap in mode rather than HImode and use that in expand_unop
356 2017-12-31 Jakub Jelinek <jakub@redhat.com>
359 * config/i386/i386.c (ix86_attribute_table): Remove excess
360 initializer for "shared" attribute.
362 2017-12-30 Tom de Vries <tom@codesourcery.com>
365 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
367 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
368 offload_funcs entries that no longer have a corresponding cgraph_node.
369 Mark the remaining ones as DECL_PRESERVE_P.
370 (output_lto): Call prune_offload_funcs.
372 2017-12-30 Jakub Jelinek <jakub@redhat.com>
374 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
375 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
376 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
377 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
378 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
379 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
380 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
381 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
382 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
383 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
384 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
385 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
386 vaesenc_<mode>, vpclmulqdq_<mode>,
387 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
389 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
391 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
392 _Float<N> and _Float<N>X variants for rounding built-in
394 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
395 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
396 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
397 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
398 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
399 * builtins.c (mathfn_built_in_2): Likewise.
400 * internal-fn.def (CEIL): Likewise.
402 (NEARBYINT): Likewise.
406 * convert.c (convert_to_integer_1): Likewise.
407 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
408 (integer_valued_real_call_p): Likewise.
409 * fold-const-call.c (fold_const_call_ss): Likewise.
410 * gencfn-macros.c (print_case_cfn): Change CFN and operator
411 printers to take a const char * suffix instead of a bool.
412 (print_define_operator_list): Likewise.
413 (fltall_suffixes): New list of suffixes, that include the
414 traditional suffixes as well as all of the _Float<N> and
416 (main): For _Float<N> and _Float<N>X functions, emit both
417 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
418 has the _Float<N> and _Float<N>X case names or operators. The
419 <name>_ALL variant has both the traditional and the
420 _Float<N>/_Float<N>X case names or operators.
421 * match.pd (COPYSIGN optimizations): Provide optimizations for
422 _Float<N> and _Float<N>X types where possible.
423 (MIN/MAX optimizations): Likewise.
424 (sqrt optimizations): Likewise.
425 (rounding optimizations): Likewise.
427 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
429 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
430 gen_int_mode rather than GEN_INT.
431 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
433 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
435 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
436 instead of CONSTANT_P.
437 (gen_vec_series): Likewise.
438 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
440 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
442 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
445 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
447 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
448 (aarch64_ld1x2<VDC:mode>): Likewise.
449 (aarch64_simd_ld1<mode>_x2): Likewise.
450 (aarch64_simd_ld1<mode>_x2): Likewise.
451 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
452 (vld1_s8_x2): Likewise.
453 (vld1_u16_x2): Likewise.
454 (vld1_s16_x2): Likewise.
455 (vld1_u32_x2): Likewise.
456 (vld1_s32_x2): Likewise.
457 (vld1_u64_x2): Likewise.
458 (vld1_s64_x2): Likewise.
459 (vld1_f16_x2): Likewise.
460 (vld1_f32_x2): Likewise.
461 (vld1_f64_x2): Likewise.
462 (vld1_p8_x2): Likewise.
463 (vld1_p16_x2): Likewise.
464 (vld1_p64_x2): Likewise.
465 (vld1q_u8_x2): Likewise.
466 (vld1q_s8_x2): Likewise.
467 (vld1q_u16_x2): Likewise.
468 (vld1q_s16_x2): Likewise.
469 (vld1q_u32_x2): Likewise.
470 (vld1q_s32_x2): Likewise.
471 (vld1q_u64_x2): Likewise.
472 (vld1q_s64_x2): Likewise.
473 (vld1q_f16_x2): Likewise.
474 (vld1q_f32_x2): Likewise.
475 (vld1q_f64_x2): Likewise.
476 (vld1q_p8_x2): Likewise.
477 (vld1q_p16_x2): Likewise.
478 (vld1q_p64_x2): Likewise.
480 2017-12-27 Martin Liska <mliska@suse.cz>
482 PR tree-optimization/83552
483 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
484 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
486 2017-12-27 Tom de Vries <tom@codesourcery.com>
488 * config/nvptx/nvptx.c (nvptx_option_override): Disable
489 -gstatement-frontiers.
491 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
493 PR rtl-optimization/83513
494 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
495 before priority comparison.
497 2017-12-25 Jakub Jelinek <jakub@redhat.com>
500 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
501 ix86_isa_flags2 to ix86_isa_flags.
502 * config/i386/i386-c.c (ix86_target_macros_internal): Test
503 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
504 isa_flags rather than isa_flags2.
505 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
506 and -mavx512bitalg from isa2_opts to isa_opts.
507 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
508 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
509 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
511 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
512 OPTION_MASK_ISA_AVX512F ored with another option similarly to
513 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
514 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
515 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
516 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
517 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
518 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
519 builtins from bdesc_args2 to bdesc_args section.
520 (__builtin_ia32_compressstoreuqi512_mask,
521 __builtin_ia32_compressstoreuhi512_mask,
522 __builtin_ia32_compressstoreuqi256_mask,
523 __builtin_ia32_expandloadqi512_mask,
524 __builtin_ia32_expandloadqi512_maskz,
525 __builtin_ia32_expandloadhi512_mask,
526 __builtin_ia32_expandloadhi512_maskz,
527 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
528 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
529 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
530 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
531 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
532 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
533 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
534 __builtin_ia32_vpshldv_v32hi_maskz,
535 __builtin_ia32_vpopcountb_v64qi_mask,
536 __builtin_ia32_vpopcountw_v32hi_mask,
537 __builtin_ia32_vpshufbitqmb512_mask,
538 __builtin_ia32_vpshufbitqmb256_mask): Add
539 " | OPTION_MASK_ISA_AVX512BW".
540 (__builtin_ia32_expandloadqi256_mask,
541 __builtin_ia32_expandloadqi256_maskz,
542 __builtin_ia32_vpopcountb_v32qi_mask): Add
543 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
544 (__builtin_ia32_expandloadhi256_mask,
545 __builtin_ia32_expandloadhi256_maskz,
546 __builtin_ia32_expandloadqi128_mask,
547 __builtin_ia32_expandloadqi128_maskz,
548 __builtin_ia32_expandloadhi128_mask,
549 __builtin_ia32_expandloadhi128_maskz,
550 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
551 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
552 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
553 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
554 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
555 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
556 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
557 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
558 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
559 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
560 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
561 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
562 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
563 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
564 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
565 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
566 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
567 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
568 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
569 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
570 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
571 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
572 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
573 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
574 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
575 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
576 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
577 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
578 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
579 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
580 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
581 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
582 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
583 __builtin_ia32_vpopcountw_v8hi_mask): Add
584 " | OPTION_MASK_ISA_AVX512VL".
585 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
586 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
587 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
588 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
589 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
590 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
591 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
592 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
593 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
594 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
595 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
596 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
597 _mm_mask_bitshuffle_epi64_mask): Likewise.
598 * common/config/i386/i386-common.c
599 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
600 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
601 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
602 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
603 OPTION_MASK_ISA_AVX512BITALG_UNSET.
604 (OPTION_MASK_ISA2_AVX512F_UNSET,
605 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
606 (ix86_handle_option): For -mno-general-regs-only, clear from
607 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
608 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
609 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
610 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
611 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
612 for moving from ix86_isa_flags2 to ix86_isa_flags.
614 2017-12-23 Jakub Jelinek <jakub@redhat.com>
617 * fold-const.c (struct contains_label_data): New type.
618 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
619 inside of a SWITCH_BODY seen during the walk.
620 (contains_label_p): Use walk_tree instead of
621 walk_tree_without_duplicates, prepare data for contains_label_1 and
624 2017-12-22 Jakub Jelinek <jakub@redhat.com>
627 * tree-iterator.c (alloc_stmt_list): Start with cleared
628 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
632 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
633 enable avx512f explicitly in #pragma GCC target.
634 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
635 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
636 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
637 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
638 __builtin_ia32_vpdpbusds_v8si_mask,
639 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
640 __builtin_ia32_vpdpbusds_v4si_mask,
641 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
642 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
643 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
644 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
645 __builtin_ia32_vpdpwssds_v8si_mask,
646 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
647 __builtin_ia32_vpdpwssds_v4si_mask,
648 __builtin_ia32_vpdpwssds_v4si_maskz): Use
649 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
650 just OPTION_MASK_ISA_AVX512VNNI.
652 2017-12-22 Martin Jambor <mjambor@suse.cz>
655 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
658 2017-12-22 Julia Koval <julia.koval@intel.com>
659 Sebastian Peryt <sebastian.peryt@intel.com>
661 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
662 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
663 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
664 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
665 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
666 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
667 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
668 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
669 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
670 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
671 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
672 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
673 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
674 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
675 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
676 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
677 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
678 _mm256_maskz_popcnt_epi64): New intrinsics.
679 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
680 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
681 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
682 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
683 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
684 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
685 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
686 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
687 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
688 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
689 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
690 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
691 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
692 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
693 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
694 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
695 __builtin_ia32_vpshufbitqmb256_mask,
696 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
697 * config/i386/i386-c.c (__AVX512BITALG__): New.
698 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
699 (ix86_valid_target_attribute_inner_p): Ditto.
700 (ix86_expand_args_builtin): Handle new types.
701 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
702 * config/i386/i386.opt: Add -mavx512bitalg.
703 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
704 avx512bitalgintrin.h.
705 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
706 (vpopcount<mode><mask_name>): Add more types.
707 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
708 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
710 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
712 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
713 Or in OPTION_MASK_ISA_AVX512F_SET.
714 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
715 OPTION_MASK_ISA_AVX512VNNI_UNSET.
716 (ix86_handle_option): Adjust for
717 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
718 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
720 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
721 isa_flag instead of isa_flag2.
722 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
723 isa_opts2 to isa_opts.
724 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
727 2017-12-22 Mike Stump <mikestump@comcast.net>
728 Eric Botcazou <ebotcazou@adacore.com>
730 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
732 2017-12-21 Jakub Jelinek <jakub@redhat.com>
735 * config/i386/i386.c (ix86_function_arg_boundary): Return
736 PARM_BOUNDARY for TYPE_EMPTY_P types.
739 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
740 if navail is >= dir.len.
742 2017-12-21 Steve Ellcey <sellcey@cavium.com>
744 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
747 2017-12-21 Jakub Jelinek <jakub@redhat.com>
749 PR rtl-optimization/80747
750 PR rtl-optimization/83512
751 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
752 succ edge from ENTRY, copy partition from e->dest to the newly
754 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
755 ENTRY, use BB_PARTITION of its successor block as current_partition.
756 Don't copy partition when splitting succ edge from ENTRY.
758 PR tree-optimization/83523
759 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
760 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
761 (convert_mult_to_fma): Likewise.
763 PR tree-optimization/83521
764 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
765 gimple_build_assign without code on result of
766 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
769 2017-12-21 Andrew Pinski <apinski@cavium.com>
770 Steve Ellcey <sellcey@cavium.com>
772 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
773 multi-arch for ilp32.
775 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
778 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
779 constraints to operand 2.
780 (*ashl<mode>3_mask_1): Ditto.
781 (*<shift_insn><mode>3_mask): Ditto.
782 (*<shift_insn><mode>3_mask_1): Ditto.
783 (*<rotate_insn><mode>3_mask): Ditto.
784 (*<rotate_insn><mode>3_mask_1): Ditto.
786 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
788 * reorg.c (make_return_insns): Reemit each insn with its own location.
790 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
792 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
793 generation for cases where splatting a value is not useful.
794 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
795 across a vec_duplicate and a paradoxical subreg forming a vector
796 mode to a vec_concat.
798 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
800 * combine.c (simplify_set): Do not transform subregs to zero_extends
801 if the destination is not a scalar int mode.
803 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
806 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
807 shared zero if the input has overflowed.
809 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
811 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
812 support for Cortex-A55 and Cortex-A75.
814 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
816 * common/config/arm/arm-common.c (compare_opt_names): Add function
817 comment. Use strcmp instead of manual loop.
819 2017-12-21 Martin Liska <mliska@suse.cz>
821 PR gcov-profile/83509
822 * gcov-dump.c (dump_gcov_file): Do not read info about
823 support_unexecuted_blocks for gcda files.
825 2017-12-21 Jakub Jelinek <jakub@redhat.com>
827 PR rtl-optimization/82973
828 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
829 (valid_for_const_vector_p): ... this.
830 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
831 (valid_for_const_vector_p): ... this. Adjust function comment.
832 (gen_vec_duplicate): Adjust caller.
833 * optabs.c (expand_vector_broadcast): Likewise.
834 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
835 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
837 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
838 instead of GET_CODE == CONST_FIXED.
839 (simplify_subreg): Use CONST_FIXED_P macro instead of
840 GET_CODE == CONST_FIXED.
843 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
844 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
845 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
846 (ix86_option_override_internal): Adjust for
847 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
848 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
849 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
850 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
851 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
852 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
853 def_builtin2 for CET builtins.
854 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
855 up formatting in IX86_BUILTIN_RDPID code.
856 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
857 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
858 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
859 ix86_isa_flags2 to ix86_isa_flags.
860 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
862 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
863 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
864 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
866 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
867 Or in OPTION_MASK_ISA_AVX512F_SET.
868 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
869 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
870 (ix86_handle_option): Adjust for
871 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
872 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
874 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
875 Alan Hayward <alan.hayward@arm.com>
876 David Sherwood <david.sherwood@arm.com>
878 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
879 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
880 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
881 Track polynomial offsets.
883 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
884 Alan Hayward <alan.hayward@arm.com>
885 David Sherwood <david.sherwood@arm.com>
887 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
888 Treat drb->init as a poly_int. Fail if its misalignment wrt
889 vector_alignment isn't known.
891 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
892 Alan Hayward <alan.hayward@arm.com>
893 David Sherwood <david.sherwood@arm.com>
895 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
897 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
898 (vect_loop_versioning): Take the loop versioning threshold as a
900 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
901 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
902 versioning_threshold.
903 (vect_analyze_loop_2): Compute the loop versioning threshold
904 whenever loop versioning is needed, and store it in the new
905 field rather than combining it with the cost model threshold.
906 (vect_transform_loop): Update call to vect_loop_versioning.
907 Try to combine the loop versioning and cost thresholds here.
909 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
910 Alan Hayward <alan.hayward@arm.com>
911 David Sherwood <david.sherwood@arm.com>
913 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
914 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
915 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
916 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
917 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
918 (group_compare_offset): Update accordingly.
919 (split_small_address_groups_p): Likewise.
920 (record_use): Take addr_offset as a poly_uint64 rather than
921 an unsigned HOST_WIDE_INT.
922 (strip_offset): Return the offset as a poly_uint64 rather than
923 an unsigned HOST_WIDE_INT.
924 (record_group_use, split_address_groups): Track polynomial offsets.
925 (add_iv_candidate_for_use): Likewise.
926 (addr_offset_valid_p): Take the offset as a poly_int64 rather
927 than a HOST_WIDE_INT.
928 (strip_offset_1): Return the offset as a poly_int64 rather than
931 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
932 Alan Hayward <alan.hayward@arm.com>
933 David Sherwood <david.sherwood@arm.com>
935 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
936 rather than a HOST_WIDE_INT.
937 * tree.c (get_binfo_at_offset): Likewise.
939 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
940 Alan Hayward <alan.hayward@arm.com>
941 David Sherwood <david.sherwood@arm.com>
943 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
944 rather than a HOST_WIDE_INT.
945 * tree-sra.c (build_ref_for_offset): Likewise.
947 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
948 Alan Hayward <alan.hayward@arm.com>
949 David Sherwood <david.sherwood@arm.com>
951 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
953 * tree.c (mem_ref_offset): Likewise.
954 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
955 * builtins.c (get_object_alignment_2): Likewise.
956 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
957 * gimple-fold.c (get_base_constructor): Likewise.
958 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
959 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
961 * ipa-polymorphic-call.c
962 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
963 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
964 (get_ancestor_addr_info): Likewise.
965 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
966 * match.pd: Likewise.
967 * tree-data-ref.c (dr_analyze_innermost): Likewise.
968 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
969 * tree-eh.c (tree_could_trap_p): Likewise.
970 * tree-object-size.c (addr_object_size): Likewise.
971 * tree-ssa-address.c (copy_ref_info): Likewise.
972 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
973 (indirect_refs_may_alias_p): Likewise.
974 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
975 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
976 (non_rewritable_mem_ref_base): Likewise.
977 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
978 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
979 * varasm.c (decode_addr_const): Likewise.
981 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
982 Alan Hayward <alan.hayward@arm.com>
983 David Sherwood <david.sherwood@arm.com>
985 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
986 offsets for MEM_REFs.
988 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
989 Alan Hayward <alan.hayward@arm.com>
990 David Sherwood <david.sherwood@arm.com>
992 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
993 as a poly_uint64 rather than an unsigned int.
994 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
996 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
997 Alan Hayward <alan.hayward@arm.com>
998 David Sherwood <david.sherwood@arm.com>
1000 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1002 (decode_addr_const): Update accordingly.
1004 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1005 Alan Hayward <alan.hayward@arm.com>
1006 David Sherwood <david.sherwood@arm.com>
1008 * tree.h (bit_field_size, bit_field_offset): New functions.
1009 * hsa-gen.c (gen_hsa_addr): Use them.
1010 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1011 (simplify_vector_constructor): Likewise.
1012 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1013 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1014 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1015 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
1018 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1019 Alan Hayward <alan.hayward@arm.com>
1020 David Sherwood <david.sherwood@arm.com>
1022 * expr.h (emit_group_load, emit_group_load_into_temps)
1023 (emit_group_store): Take the size as a poly_int64 rather than an int.
1024 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1025 (emit_group_load_into_temp, emit_group_store): Likewise.
1027 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1028 Alan Hayward <alan.hayward@arm.com>
1029 David Sherwood <david.sherwood@arm.com>
1031 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1032 an unsigned int to a poly_uint64.
1033 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1034 sizes as poly_uint64s rather than unsigned ints.
1035 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1038 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1039 Alan Hayward <alan.hayward@arm.com>
1040 David Sherwood <david.sherwood@arm.com>
1042 * lra-constraints.c (emit_inc): Change inc_amount from an int
1045 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1046 Alan Hayward <alan.hayward@arm.com>
1047 David Sherwood <david.sherwood@arm.com>
1049 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1051 (add_stack_var, stack_var_cmp, partition_stack_vars)
1052 (dump_stack_var_partition): Update accordingly.
1053 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1054 than a HOST_WIDE_INT.
1055 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1056 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1057 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1058 rather than a HOST_WIDE_INT.
1060 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1061 Alan Hayward <alan.hayward@arm.com>
1062 David Sherwood <david.sherwood@arm.com>
1064 * target.def (return_pops_args): Treat both the input and output
1065 sizes as poly_int64s rather than HOST_WIDE_INTS.
1066 * targhooks.h (default_return_pops_args): Update accordingly.
1067 * targhooks.c (default_return_pops_args): Likewise.
1068 * doc/tm.texi: Regenerate.
1069 * emit-rtl.h (incoming_args): Change pops_args, size and
1070 outgoing_args_size from int to poly_int64_pod.
1071 * function.h (expr_status): Change x_pending_stack_adjust and
1072 x_stack_pointer_delta from int to poly_int64.
1073 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1074 (ARGS_SIZE_RTX): Update accordingly.
1075 * calls.c (highest_outgoing_arg_in_use): Change from int to
1077 (stack_usage_watermark, stored_args_watermark): New variables.
1078 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1079 (emit_call_1): Change the stack_size and rounded_stack_size
1080 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1082 (save_fixed_argument_area): Check stack_usage_watermark.
1083 (initialize_argument_information): Change old_pending_adj from
1084 a HOST_WIDE_INT * to a poly_int64_pod *.
1085 (compute_argument_block_size): Return the size as a poly_int64
1087 (finalize_must_preallocate): Track polynomial argument sizes.
1088 (compute_argument_addresses): Likewise.
1089 (internal_arg_pointer_based_exp): Track polynomial offsets.
1090 (mem_overlaps_already_clobbered_arg_p): Rename to...
1091 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1092 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1093 Check stored_args_used_watermark.
1094 (load_register_parameters): Update accordingly.
1095 (check_sibcall_argument_overlap_1): Likewise.
1096 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1097 args size as a poly_int64 rather than an int. Return a bool
1098 indicating whether the optimization was possible and return
1099 the new adjustment by reference.
1100 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1101 Update stored_args_watermark.
1102 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1103 (expand_call): Likewise. Maintain stack_usage_watermark and
1104 stored_args_watermark. Update calls to
1105 combine_pending_stack_adjustment_and_call.
1106 (emit_library_call_value_1): Handle polynomial argument sizes.
1107 Call stack_region_maybe_used_p and mark_stack_region_used.
1108 Maintain stack_usage_watermark.
1109 (store_one_arg): Likewise. Update call to
1110 mem_overlaps_already_clobbered_arg_p.
1111 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1113 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1114 * config/microblaze/microblaze.c (microblaze_function_prologue):
1116 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1117 TARGET_RETURN_POPS_ARGS interface.
1118 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1120 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1121 * config/i386/i386.c (ix86_return_pops_args): Update for new
1122 TARGET_RETURN_POPS_ARGS interface.
1123 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1124 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1125 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1126 TARGET_RETURN_POPS_ARGS interface.
1127 * config/vax/vax.c (vax_return_pops_args): Likewise.
1128 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1129 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1130 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1131 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1132 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1133 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1134 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1135 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1136 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1137 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1138 and x_stack_pointer_delta from int to poly_int64.
1139 * dojump.c (do_pending_stack_adjust): Update accordingly.
1140 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1141 stack_pointer_deltas.
1142 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1143 (pad_to_arg_alignment): Track polynomial offsets.
1144 (assign_parm_find_stack_rtl): Likewise.
1145 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1146 * toplev.c (output_stack_usage): Update reference to
1147 current_function_pushed_stack_size.
1149 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1150 Alan Hayward <alan.hayward@arm.com>
1151 David Sherwood <david.sherwood@arm.com>
1153 * function.c (in_arg_offset, var_offset, dynamic_offset)
1154 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1155 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1156 rather than a HOST_WIDE_INT.
1157 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1158 (instantiate_virtual_regs_in_insn): Likewise.
1160 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1161 Alan Hayward <alan.hayward@arm.com>
1162 David Sherwood <david.sherwood@arm.com>
1164 * rtl.h (get_args_size, add_args_size_note): New functions.
1165 (find_args_size_adjust): Return a poly_int64 rather than a
1167 (fixup_args_size_notes): Likewise. Make the same change to the
1168 end_args_size parameter.
1169 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1170 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1171 * calls.c (emit_call_1): Likewise.
1172 * explow.c (adjust_stack_1): Likewise.
1173 * cfgcleanup.c (old_insns_match_p): Update use of
1174 find_args_size_adjust.
1175 * combine.c (distribute_notes): Track polynomial arg sizes.
1176 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1177 end_true_args_size, beg_delay_args_size and end_delay_args_size
1178 from HOST_WIDE_INT to poly_int64.
1179 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1180 than a HOST_WIDE_INT.
1181 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1182 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1183 polynomial arg sizes.
1184 * emit-rtl.c (try_split): Use get_args_size.
1185 * recog.c (peep2_attempt): Likewise.
1186 * reload1.c (reload_as_needed): Likewise.
1187 * expr.c (find_args_size_adjust): Return the adjustment as a
1188 poly_int64 rather than a HOST_WIDE_INT.
1189 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1190 to a poly_int64 and change the return type in the same way.
1191 (emit_single_push_insn): Track polynomial arg sizes.
1193 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1194 Alan Hayward <alan.hayward@arm.com>
1195 David Sherwood <david.sherwood@arm.com>
1197 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1198 from HOST_WIDE_INT to poly_int64.
1199 * expr.c (push_block, emit_push_insn): Likewise.
1201 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1202 Alan Hayward <alan.hayward@arm.com>
1203 David Sherwood <david.sherwood@arm.com>
1205 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1207 (get_frame_size): Return the size as a poly_int64 rather than a
1209 (frame_offset_overflow): Take the offset as a poly_int64 rather
1210 than a HOST_WIDE_INT.
1211 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1212 (assign_stack_temp): Likewise for the size.
1213 * function.c (get_frame_size): Return a poly_int64 rather than
1215 (frame_offset_overflow): Take the offset as a poly_int64 rather
1216 than a HOST_WIDE_INT.
1217 (try_fit_stack_local): Take the start, length and size as poly_int64s
1218 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1219 rather than a HOST_WIDE_INT.
1220 (add_frame_space): Take the start and end as poly_int64s rather than
1222 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1223 (assign_stack_temp): Likewise for the size.
1224 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1226 (find_temp_slot_from_address): Handle polynomial offsets.
1227 (combine_temp_slots): Likewise.
1228 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1230 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1231 poly_int64 rather than a HOST_WIDE_INT.
1232 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1233 than a HOST_WIDE_INT.
1234 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1235 polynomial frame offsets.
1236 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1237 as a poly_int64 rather than an int.
1238 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1239 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1240 * config/v850/v850.c (compute_frame_size): Likewise.
1241 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1242 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1243 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1244 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1245 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1246 rather than a HOST_WIDE_INT.
1247 * explow.c (get_dynamic_stack_base): Likewise.
1248 * final.c (final_start_function): Use the constant lower bound
1249 of the frame size for -Wframe-larger-than.
1250 * ira.c (do_reload): Adjust for new get_frame_size return type.
1251 * lra.c (lra): Likewise.
1252 * reload1.c (reload): Likewise.
1253 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1254 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1255 * rtlanal.c (get_initial_register_offset): Return the offset as
1256 a poly_int64 rather than a HOST_WIDE_INT.
1258 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1259 Alan Hayward <alan.hayward@arm.com>
1260 David Sherwood <david.sherwood@arm.com>
1262 * reload1.c (elim_table): Change initial_offset, offset and
1263 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1264 (offsets_at): Change the target array's element type from
1265 HOST_WIDE_INT to poly_int64_pod.
1266 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1267 (elimination_costs_in_insn, update_eliminable_offsets)
1268 (verify_initial_elim_offsets, set_offsets_for_label)
1269 (init_eliminable_invariants): Update after above changes.
1271 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1272 Alan Hayward <alan.hayward@arm.com>
1273 David Sherwood <david.sherwood@arm.com>
1275 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1276 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1277 (decomposition): Change start and end from HOST_WIDE_INT
1279 (decompose, immune_p): Update accordingly.
1280 (find_inc_amount): Return a poly_int64 rather than an int.
1281 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1284 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1285 Alan Hayward <alan.hayward@arm.com>
1286 David Sherwood <david.sherwood@arm.com>
1288 * tree.h (get_inner_reference): Return the bitsize and bitpos
1289 as poly_int64_pods rather than HOST_WIDE_INT.
1290 * fold-const.h (ptr_difference_const): Return the pointer difference
1291 as a poly_int64_pod rather than a HOST_WIDE_INT.
1292 * expr.c (get_inner_reference): Return the bitsize and bitpos
1293 as poly_int64_pods rather than HOST_WIDE_INT.
1294 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1296 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1297 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1298 (optimize_bit_field_compare): Update call to get_inner_reference.
1299 (decode_field_reference): Likewise.
1300 (fold_unary_loc): Track polynomial offsets and sizes.
1301 (split_address_to_core_and_offset): Return the bitpos as a
1302 poly_int64_pod rather than a HOST_WIDE_INT.
1303 (ptr_difference_const): Likewise for the pointer difference.
1304 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1305 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1306 * dbxout.c (dbxout_expand_expr): Likewise.
1307 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1308 (loc_list_from_tree_1, fortran_common): Likewise.
1309 * gimple-laddress.c (pass_laddress::execute): Likewise.
1310 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1311 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1312 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1313 * tree-affine.c (tree_to_aff_combination): Likewise.
1314 (get_inner_reference_aff): Likewise.
1315 * tree-data-ref.c (split_constant_offset_1): Likewise.
1316 (dr_analyze_innermost): Likewise.
1317 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1318 * tree-sra.c (ipa_sra_check_caller): Likewise.
1319 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1320 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1321 (instrument_bool_enum_load, instrument_object_size): Likewise.
1322 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1323 to get_inner_reference.
1324 * hsa-gen.c (gen_hsa_addr): Likewise.
1325 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1326 * tsan.c (instrument_expr): Likewise.
1327 * match.pd: Update call to ptr_difference_const.
1329 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1330 Alan Hayward <alan.hayward@arm.com>
1331 David Sherwood <david.sherwood@arm.com>
1333 * fold-const.c (fold_comparison): Track sizes and offsets as
1334 poly_int64s rather than HOST_WIDE_INTs when folding address
1337 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1338 Alan Hayward <alan.hayward@arm.com>
1339 David Sherwood <david.sherwood@arm.com>
1341 * expr.h (get_bit_range): Return the bitstart and bitend as
1342 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1343 as a poly_int64 rather than a HOST_WIDE_INT.
1344 * expr.c (get_bit_range): Likewise.
1345 (expand_assignment): Update call accordingly.
1346 * fold-const.c (optimize_bit_field_compare): Likewise.
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 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1353 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1354 as an optional final parameter.
1355 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1356 (aarch64_legitimate_address_p): Likewise.
1357 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1358 instead of an rtx code.
1359 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1360 (aarch64_legitimate_address_hook_p): Likewise.
1361 (aarch64_print_ldpstp_address): Likewise.
1362 (aarch64_print_operand_address): Likewise.
1363 (aarch64_address_cost): Likewise.
1364 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1365 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1366 (aarch64_mem_pair_lanes_operand): Likewise.
1368 2017-12-20 Richard Biener <rguenther@suse.de>
1370 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1371 update_stmt_if_modified.
1373 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1375 PR tree-optimization/83491
1376 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1377 before walking uses. Improve coding style and comments.
1379 2017-12-20 Tom de Vries <tom@codesourcery.com>
1381 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1383 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1386 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1388 (pass_ipa_free_fn_summary): Add small_p private data member,
1389 initialize to false in the ctor.
1390 (pass_ipa_free_fn_summary::clone,
1391 pass_ipa_free_fn_summary::set_pass_param,
1392 pass_ipa_free_fn_summary::gate): New methods.
1393 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1394 | TODO_dump_symtab if small_p.
1395 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1396 entry and add another instance of the pass with false parm after
1398 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1399 ipa_free_fn_summary here.
1401 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1403 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1405 2017-12-20 Martin Sebor <msebor@redhat.com>
1408 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1409 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1410 (expand_builtin_strncmp): Same.
1412 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1415 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1416 allowed debug stmts before labels.
1417 (expand_gimple_basic_block): Likewise.
1418 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1419 * gimple-iterator.h (gsi_after_labels): Likewise.
1420 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1421 rename reused variable, and simplify using gsi_move_before.
1422 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1423 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1424 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1425 (gimple_verify_flow_info, gimple_block_label): Likewise.
1426 (make_blocks): Move debug markers after adjacent labels.
1427 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1428 allowed debug insns outside blocks.
1429 * df-scan.c (df_insn_delete): Likewise.
1430 * lra-constraints.c (update_ebb_live_info): Likewise.
1431 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1432 (vt_initialize, delete_vta_debug_insns): Likewise.
1433 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1435 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1436 Alan Hayward <alan.hayward@arm.com>
1437 David Sherwood <david.sherwood@arm.com>
1439 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1440 (round_up_to_byte_boundary): Likewise.
1441 * expr.h (get_bit_range): Add temporary shim.
1442 * gimple-ssa-store-merging.c (store_operand_info): Change the
1443 bitsize, bitpos, bitregion_start and bitregion_end fields from
1444 unsigned HOST_WIDE_INT to poly_uint64.
1445 (merged_store_group): Likewise load_align_base.
1446 (compatible_load_p, compatible_load_p): Update accordingly.
1447 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1448 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1449 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1450 bitregion_start and bitregion_end as poly_uint64s rather than
1451 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1452 (handled_load): Take the bitsize, bitpos,
1453 bitregion_start and bitregion_end as poly_uint64s rather than
1454 unsigned HOST_WIDE_INTs.
1455 (pass_store_merging::process_store): Update call to
1456 mem_valid_for_store_merging.
1458 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1459 Alan Hayward <alan.hayward@arm.com>
1460 David Sherwood <david.sherwood@arm.com>
1462 * builtins.c (get_object_alignment_2): Track polynomial offsets
1463 and sizes. Update the alignment handling.
1465 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1466 Alan Hayward <alan.hayward@arm.com>
1467 David Sherwood <david.sherwood@arm.com>
1469 * tree.h (get_inner_reference): Add a version that returns the
1470 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1471 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1472 the case in which bitpos is not associated with the first byte.
1474 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1475 Alan Hayward <alan.hayward@arm.com>
1476 David Sherwood <david.sherwood@arm.com>
1478 * tree-affine.h (get_inner_reference_aff): Return the size as a
1480 * tree-affine.c (get_inner_reference_aff): Likewise.
1481 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1482 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1484 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1485 Alan Hayward <alan.hayward@arm.com>
1486 David Sherwood <david.sherwood@arm.com>
1488 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1489 HOST_WIDE_INT rather than a poly_int64.
1491 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1492 Alan Hayward <alan.hayward@arm.com>
1493 David Sherwood <david.sherwood@arm.com>
1495 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1496 HOST_WIDE_INT to poly_int64_pod.
1497 (perform_symbolic_merge): Update accordingly.
1498 (bswap_replace): Likewise.
1500 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1501 Alan Hayward <alan.hayward@arm.com>
1502 David Sherwood <david.sherwood@arm.com>
1504 * tree-affine.h (aff_tree::offset): Change from widest_int
1506 (wide_int_ext_for_comb): Delete.
1507 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1508 constants as poly_widest_int rather than widest_int.
1509 (aff_combination_constant_multiple_p): Return the multiplier
1510 as a poly_widest_int.
1511 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1513 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1514 version static and add an overload for poly_widest_int.
1515 (aff_combination_const, aff_combination_add_cst)
1516 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1517 the constants as poly_widest_int rather than widest_int.
1518 (tree_to_aff_combination): Generalize INTEGER_CST case to
1520 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1521 (aff_combination_add_product, aff_combination_mult): Handle
1523 (aff_combination_constant_multiple_p): Return the multiplier
1524 as a poly_widest_int.
1525 * tree-predcom.c (determine_offset): Return the offset as a
1527 (split_data_refs_to_components, suitable_component_p): Update
1529 (valid_initializer_p): Update call to
1530 aff_combination_constant_multiple_p.
1531 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1532 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1533 as a poly_int64 rather than a HOST_WIDE_INT.
1534 (get_address_cost): Handle polynomial offsets.
1535 (iv_elimination_compare_lt): Likewise.
1536 (rewrite_use_nonlinear_expr): Likewise.
1538 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1539 Alan Hayward <alan.hayward@arm.com>
1540 David Sherwood <david.sherwood@arm.com>
1542 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1543 as a poly_int64_pod rather than a HOST_WIDE_INT.
1544 (get_addr_base_and_unit_offset): Likewise.
1545 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1546 (get_addr_base_and_unit_offset): Likewise.
1547 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1548 to poly_int64 in example.
1549 * fold-const.c (fold_binary_loc): Update call to
1550 get_addr_base_and_unit_offset.
1551 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1552 (maybe_canonicalize_mem_ref_addr): Likewise.
1553 (gimple_fold_stmt_to_constant_1): Likewise.
1554 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1556 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1557 * match.pd: Likewise.
1558 * omp-low.c (lower_omp_target): Likewise.
1559 * tree-sra.c (build_ref_for_offset): Likewise.
1560 (build_debug_ref_for_model): Likewise.
1561 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1562 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1563 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1564 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1565 (constant_pointer_difference): Likewise.
1566 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1567 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1568 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1569 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1570 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1571 (set_ssa_val_to): Likewise.
1572 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1573 (maybe_diag_stxncpy_trunc): Likewise.
1574 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1575 * tree.c (build_simple_mem_ref_loc): Likewise.
1576 (array_at_struct_end_p): Likewise.
1578 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1579 Alan Hayward <alan.hayward@arm.com>
1580 David Sherwood <david.sherwood@arm.com>
1582 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1583 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1584 (get_ref_base_and_extent_hwi): Declare.
1585 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1586 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1587 (get_ref_base_and_extent_hwi): New function.
1588 * cfgexpand.c (expand_debug_expr): Update call to
1589 get_ref_base_and_extent.
1590 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1591 * gimple-fold.c (get_base_constructor): Return the offset as a
1592 poly_int64_pod rather than a HOST_WIDE_INT.
1593 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1594 * ipa-polymorphic-call.c
1595 (ipa_polymorphic_call_context::set_by_invariant)
1596 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1597 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1598 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1599 rather than get_ref_base_and_extent.
1600 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1601 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1602 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1604 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1605 call to get_ref_base_and_extent.
1606 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1607 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1608 (stmt_kills_ref_p): Likewise.
1609 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1610 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1612 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1613 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1614 when calling native_encode_expr.
1615 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1616 call to get_ref_base_and_extent.
1617 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1618 get_ref_base_and_extent.
1619 * var-tracking.c (track_expr_p): Likewise.
1621 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1622 Alan Hayward <alan.hayward@arm.com>
1623 David Sherwood <david.sherwood@arm.com>
1625 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1626 HOST_WIDE_INT to poly_int64_pod.
1627 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1628 polynomail parameter offsets.
1630 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1631 Alan Hayward <alan.hayward@arm.com>
1632 David Sherwood <david.sherwood@arm.com>
1634 * gengtype.c (main): Handle poly_int64_pod.
1635 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1636 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1637 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1638 from HOST_WIDE_INT to poly_int64_pod.
1639 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1640 (copy_cfa): New function.
1641 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1642 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1643 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1644 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1645 (initial_return_save): Treat offsets as poly_ints.
1646 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1647 offset is nonconstant.
1648 (reg_save): Take the offset as a poly_int64. Fall back to
1649 DW_CFA_expression for nonconstant offsets.
1650 (queue_reg_save): Take the offset as a poly_int64.
1651 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1653 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1654 Alan Hayward <alan.hayward@arm.com>
1655 David Sherwood <david.sherwood@arm.com>
1657 * rtl.h (operand_subword, operand_subword_force): Take the offset
1658 as a poly_uint64 an unsigned int.
1659 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1661 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1662 Alan Hayward <alan.hayward@arm.com>
1663 David Sherwood <david.sherwood@arm.com>
1665 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1666 'p' format code. Use INT_LIST rather than SUBREG as the example of
1667 a code with an XINT and an XEXP. Remove the implication that
1668 accessing an rtx field using XINT is expected to work.
1669 * rtl.def (SUBREG): Change format from "ei" to "ep".
1670 * rtl.h (rtunion::rt_subreg): New field.
1671 (XCSUBREG): New macro.
1672 (SUBREG_BYTE): Use it.
1673 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1674 Update constructor accordingly.
1675 (subreg_shape::operator ==): Update accordingly.
1676 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1677 than an unsigned int.
1678 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1679 a poly_uint64 rather than an unsigned int.
1680 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1681 than an unsigned int.
1682 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1683 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1684 an unsigned int. Take the sizes as poly_uint64s.
1685 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1686 an unsigned int. Take the shift as a poly_uint64 rather than
1688 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1689 as a poly_uint64 rather than an unsigned int.
1690 (simplify_subreg_regno): Likewise.
1691 (byte_lowpart_offset): Return the memory offset as a poly_int64
1693 (subreg_memory_offset): Likewise. Take the subreg offset as a
1694 poly_uint64 rather than an unsigned int.
1695 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1696 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1697 poly_uint64 rather than an unsigned int.
1698 * rtl.c (rtx_format): Describe 'p' in comment.
1699 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1700 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1701 offset as a poly_uint64 rather than an unsigned int.
1702 (byte_lowpart_offset): Return the memory offset as a poly_int64
1704 (subreg_memory_offset): Likewise. Take the subreg offset as a
1705 poly_uint64 rather than an unsigned int.
1706 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1707 mode sizes as poly_uint64s rather than unsigned ints. Return a
1708 poly_uint64 rather than an unsigned int.
1709 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1710 (copy_insn_1): Handle 'p'.
1711 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1712 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
1713 an unsigned int. Return the shift in the same way.
1714 (subreg_lsb): Return the shift as a poly_uint64 rather than an
1716 (subreg_size_offset_from_lsb): Take the sizes and shift as
1717 poly_uint64s rather than unsigned ints. Return the offset as
1719 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
1720 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
1722 * rtlhash.c (add_rtx): Handle 'p'.
1723 * genemit.c (gen_exp): Likewise.
1724 * gengenrtl.c (type_from_format, gendef): Likewise.
1725 * gensupport.c (subst_pattern_match, get_alternatives_number)
1726 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
1727 (subst_dup): Likewise.
1728 * gengtype.c (adjust_field_rtx_def): Likewise.
1729 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
1730 (match_pattern_2): Likewise.
1731 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
1732 (rtx_test::subreg_field): New function.
1733 (operator ==, safe_to_hoist_p, transition_parameter_type)
1734 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1735 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1736 * genpeep.c (match_rtx): Likewise.
1737 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1738 (rtx_writer::print_rtx_operand): Handle 'p'.
1739 (print_value): Handle SUBREG.
1740 * read-rtl.c (apply_int_iterator): Likewise.
1741 (rtx_reader::read_rtx_operand): Handle 'p'.
1742 * alias.c (rtx_equal_for_memref_p): Likewise.
1743 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1744 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1746 * calls.c (expand_call): Likewise.
1747 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1748 (make_extraction, gen_lowpart_for_combine): Likewise.
1749 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1751 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1752 rather than an unsigned int. Treat subreg offsets as poly_ints.
1753 (exp_equiv_p): Handle 'p'.
1754 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1755 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1756 * dse.c (find_shift_sequence): Likewise.
1757 * dwarf2out.c (rtl_for_decl_location): Likewise.
1758 * expmed.c (extract_low_bits): Likewise.
1759 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1760 (expand_expr_real_2): Likewise.
1761 * final.c (alter_subreg): Likewise.
1762 (leaf_renumber_regs_insn): Handle 'p'.
1763 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1764 Treat subreg offsets as poly_ints.
1765 * fwprop.c (forward_propagate_and_simplify): Likewise.
1766 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
1767 * ira.c (get_subreg_tracking_sizes): Likewise.
1768 * ira-conflicts.c (go_through_subreg): Likewise.
1769 * ira-lives.c (process_single_reg_class_operands): Likewise.
1770 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1771 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1772 as a poly_uint64 rather than an unsigned int.
1773 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1774 subreg offsets as poly_ints.
1775 * lra-constraints.c (operands_match_p): Handle 'p'.
1776 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1777 * lra-spills.c (assign_mem_slot): Likewise.
1778 * postreload.c (move2add_valid_value_p): Likewise.
1779 * recog.c (general_operand, indirect_operand): Likewise.
1780 * regcprop.c (copy_value, maybe_mode_change): Likewise.
1781 (copyprop_hardreg_forward_1): Likewise.
1782 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
1783 (record_subregs_of_mode): Likewise.
1784 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
1785 * reload.c (operands_match_p): Handle 'p'.
1786 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
1787 * reload1.c (alter_reg, choose_reload_regs): Likewise.
1788 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1789 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1790 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1791 (simplify_const_poly_int_tests<N>::run): Likewise.
1792 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1793 a poly_uint64 rather than an unsigned int.
1794 * valtrack.c (debug_lowpart_subreg): Likewise.
1795 * var-tracking.c (var_lowpart): Likewise.
1796 (loc_cmp): Handle 'p'.
1798 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1799 Alan Hayward <alan.hayward@arm.com>
1800 David Sherwood <david.sherwood@arm.com>
1802 * ira.c (get_subreg_tracking_sizes): New function.
1803 (init_live_subregs): Take an integer size rather than a register.
1804 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1805 to init_live_subregs.
1807 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1808 Alan Hayward <alan.hayward@arm.com>
1809 David Sherwood <david.sherwood@arm.com>
1811 * expr.c (store_constructor_field): Change bitsize from a
1812 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1813 HOST_WIDE_INT to a poly_int64.
1814 (store_constructor): Change size from a HOST_WIDE_INT to
1816 (store_field): Likewise bitsize and bitpos.
1818 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1819 Alan Hayward <alan.hayward@arm.com>
1820 David Sherwood <david.sherwood@arm.com>
1822 * expmed.h (store_bit_field): Change bitregion_start and
1823 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1824 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1825 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1826 (store_fixed_bit_field, store_split_bit_field): Likewise.
1827 * expr.c (store_constructor_field, store_field): Likewise.
1828 (optimize_bitfield_assignment_op): Likewise. Make the same change
1829 to bitsize and bitpos.
1830 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1831 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1832 same change in the constructor arguments.
1833 (get_best_mode): Change bitregion_start and bitregion_end from
1834 unsigned HOST_WIDE_INT to poly_uint64.
1835 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1836 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1838 (bit_field_mode_iterator::next_mode): Update for new types
1839 of m_bitregion_start and m_bitregion_end.
1840 (get_best_mode): Change bitregion_start and bitregion_end from
1841 unsigned HOST_WIDE_INT to poly_uint64.
1843 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1844 Alan Hayward <alan.hayward@arm.com>
1845 David Sherwood <david.sherwood@arm.com>
1847 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1848 accepts poly_uint64 offsets.
1849 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1850 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1851 * expmed.c (lowpart_bit_field_p): Likewise.
1852 (extract_bit_field_as_subreg): New function, split out from...
1853 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1854 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1855 extractions, check that BITSIZE matches the size of the extracted
1856 value and that BITNUM is an exact multiple of that size.
1857 If all else fails, try forcing the value into memory if
1858 BITNUM is variable, and adjusting the address so that the
1859 offset is constant. Split the part that can only handle constant
1860 bitsize and bitnum out into...
1861 (extract_integral_bit_field): ...this new function.
1862 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1863 rather than unsigned HOST_WIDE_INTs.
1865 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1866 Alan Hayward <alan.hayward@arm.com>
1867 David Sherwood <david.sherwood@arm.com>
1869 * expmed.h (store_bit_field): Take bitsize and bitnum as
1870 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1871 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1872 that returns the byte size.
1873 (store_bit_field_1): Take bitsize and bitnum as
1874 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1875 to simple_mem_bitfield_p. Split the part that can only handle
1876 constant bitsize and bitnum out into...
1877 (store_integral_bit_field): ...this new function.
1878 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1879 than unsigned HOST_WIDE_INTs.
1880 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1882 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1883 Alan Hayward <alan.hayward@arm.com>
1884 David Sherwood <david.sherwood@arm.com>
1886 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1887 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1889 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1890 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1891 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1892 offset as a poly_int64 rather than an int.
1893 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1894 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1895 * lra-constraints.c (equiv_address_substitution): Track offsets
1897 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1898 (curr_insn_transform): Handle the new form of sp_offset.
1899 * lra-eliminations.c (lra_elim_table): Change previous_offset
1900 and offset from HOST_WIDE_INT to poly_int64.
1901 (print_elim_table, update_reg_eliminate): Update accordingly.
1902 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1903 (get_elimination): Update accordingly.
1904 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1905 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1906 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1907 poly_int64 offsets generally.
1908 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1909 (mark_not_eliminable, init_elimination): Update accordingly.
1910 (remove_reg_equal_offset_note): Return a bool and pass the new
1911 offset back by pointer as a poly_int64.
1912 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1913 rather than a HOST_WIDE_INT.
1914 (do_remat): Track offsets poly_int64s.
1915 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1917 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1918 Alan Hayward <alan.hayward@arm.com>
1919 David Sherwood <david.sherwood@arm.com>
1921 * rtl.h (mem_attrs): Add a default constructor. Change size and
1922 offset from HOST_WIDE_INT to poly_int64.
1923 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1924 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1925 (widen_memory_access): Take the sizes and offsets as poly_int64s
1926 rather than HOST_WIDE_INTs.
1927 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1928 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1930 (adjust_offset_for_component_ref): Change the offset from a
1931 HOST_WIDE_INT to a poly_int64.
1932 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1933 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1934 * dce.c (find_call_stack_args): Likewise.
1935 * dse.c (record_store): Likewise.
1936 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1937 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1938 * read-rtl-function.c (test_loading_mem): Likewise.
1939 * rtlanal.c (may_trap_p_1): Likewise.
1940 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1941 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1942 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1943 (mem_attrs::mem_attrs): New function.
1944 (set_mem_attributes_minus_bitpos): Change bitpos from a
1945 HOST_WIDE_INT to poly_int64.
1946 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1947 (clear_mem_offset, clear_mem_size, change_address)
1948 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1949 initialize mem_attrs.
1950 (set_mem_offset, set_mem_size, adjust_address_1)
1951 (adjust_automodify_address_1, offset_address, widen_memory_access):
1952 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1954 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1955 Alan Hayward <alan.hayward@arm.com>
1956 David Sherwood <david.sherwood@arm.com>
1958 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1959 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1960 rather than 0 to represent an unknown size. Assert that the size
1961 is known when the mode isn't BLKmode.
1962 (may_trap_p_1): Use -1 for unknown sizes.
1963 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1965 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1966 Alan Hayward <alan.hayward@arm.com>
1967 David Sherwood <david.sherwood@arm.com>
1969 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1970 to poly_int64. Update commentary for positions_needed.large.
1971 (read_info_type): Change offset and width from HOST_WIDE_INT
1973 (set_usage_bits): Likewise.
1974 (canon_address): Return the offset as a poly_int64 rather than
1975 a HOST_WIDE_INT. Use strip_offset_and_add.
1976 (set_all_positions_unneeded, any_positions_needed_p): Use
1977 positions_needed.large to track stores with non-constant widths.
1978 (all_positions_needed_p): Likewise. Take the offset and width
1979 as poly_int64s rather than ints. Assert that rhs is nonnull.
1980 (record_store): Cope with non-constant offsets and widths.
1981 Nullify the rhs of an earlier store if we can't tell which bytes
1983 (find_shift_sequence): Take the access_size and shift as poly_int64s
1985 (get_stored_val): Take the read_offset and read_width as poly_int64s
1986 rather than HOST_WIDE_INTs.
1987 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1988 non-constant offsets and widths.
1990 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1991 Alan Hayward <alan.hayward@arm.com>
1992 David Sherwood <david.sherwood@arm.com>
1994 * inchash.h (inchash::hash::add_poly_int): New function.
1995 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1996 Use poly_int64 rather than HOST_WIDE_INT.
1997 (ao_ref::max_size_known_p): New function.
1998 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1999 rather than HOST_WIDE_INT.
2000 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2001 to temporaries until its interface is adjusted to match.
2002 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2003 (aliasing_component_refs_p, decl_refs_may_alias_p)
2004 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2005 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2006 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2008 * alias.c (ao_ref_from_mem): Likewise.
2009 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2010 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2011 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2012 (maybe_trim_complex_store, maybe_trim_constructor_store)
2013 (live_bytes_read, dse_classify_store): Likewise.
2014 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2015 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2016 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2017 (vn_reference_lookup_3): Likewise.
2018 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
2020 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2021 Alan Hayward <alan.hayward@arm.com>
2022 David Sherwood <david.sherwood@arm.com>
2024 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2025 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2026 instead of ranges_overlap_p.
2028 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2029 Alan Hayward <alan.hayward@arm.com>
2030 David Sherwood <david.sherwood@arm.com>
2032 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2033 sizes as poly_int64s rather than HOST_WIDE_INTs.
2035 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2036 Alan Hayward <alan.hayward@arm.com>
2037 David Sherwood <david.sherwood@arm.com>
2039 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2040 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2041 * gimple-fold.c (fold_ctor_reference): Likewise.
2043 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2044 Alan Hayward <alan.hayward@arm.com>
2045 David Sherwood <david.sherwood@arm.com>
2047 * target.def (dwarf_poly_indeterminate_value): New hook.
2048 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2049 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2050 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2051 * doc/tm.texi: Regenerate.
2052 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2053 offset as a poly_int64.
2054 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2055 offset as a poly_int64.
2056 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2057 Take the offset as a poly_int64.
2058 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2059 (frame_pointer_fb_offset): Change to a poly_int64.
2060 (int_loc_descriptor): Take the offset as a poly_int64. Use
2061 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2062 (based_loc_descr): Take the offset as a poly_int64.
2063 Use strip_offset_and_add to handle (plus X (const)).
2064 Use new_reg_loc_descr instead of an open-coded version of the
2065 previous implementation.
2066 (mem_loc_descriptor): Handle CONST_POLY_INT.
2067 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2068 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2070 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2071 Alan Hayward <alan.hayward@arm.com>
2072 David Sherwood <david.sherwood@arm.com>
2074 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2075 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2076 * inchash.h (inchash::hash::add_poly_hwi): New function.
2077 * gengtype.c (main): Register poly_int64.
2078 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2079 offset as a poly_int.
2080 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2081 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2082 offset as a poly_int64.
2083 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2084 * print-rtl.c (print_poly_int): New function.
2085 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2087 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2089 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2090 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2091 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2092 Enforce track_offset_p for parts of a PARALLEL.
2093 (vt_add_function_parameter): Use const_offset for the final
2094 offset to track. Use get_tracked_reg_offset for the parts
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 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2102 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2103 * doc/tm.texi: Regenerate.
2104 * hooks.h (hook_bool_uint_uint_true): Delete.
2105 (hook_bool_puint64_puint64_true): Declare.
2106 * hooks.c (hook_bool_uint_uint_true): Delete.
2107 (hook_bool_puint64_puint64_true): New function.
2108 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2109 instead of unsigned ints.
2110 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2111 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2113 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2114 Alan Hayward <alan.hayward@arm.com>
2115 David Sherwood <david.sherwood@arm.com>
2117 * optabs.h (expand_operand): Add an int_value field.
2118 (create_expand_operand): Add an int_value parameter and use it
2119 to initialize the new expand_operand field.
2120 (create_integer_operand): Replace with a declaration of a function
2121 that accepts poly_int64s. Move the implementation to...
2122 * optabs.c (create_integer_operand): ...here.
2123 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2124 the mode preserves the value of int_value, instead of calling
2125 const_int_operand on the rtx. Use gen_int_mode to generate
2128 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2129 Alan Hayward <alan.hayward@arm.com>
2130 David Sherwood <david.sherwood@arm.com>
2132 * dumpfile.h (dump_dec): Declare.
2133 * dumpfile.c (dump_dec): New function.
2134 * pretty-print.h (pp_wide_integer): Turn into a function and
2135 declare a poly_int version.
2136 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2138 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2139 Alan Hayward <alan.hayward@arm.com>
2140 David Sherwood <david.sherwood@arm.com>
2142 * doc/generic.texi (POLY_INT_CST): Document.
2143 * tree.def (POLY_INT_CST): New tree code.
2144 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2145 * tree-core.h (tree_poly_int_cst): New struct.
2146 (tree_node): Add a poly_int_cst field.
2147 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2148 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2149 instead of a wide_int_ref.
2150 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2152 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2153 instead of an unsigned HOST_WIDE_INT.
2154 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2155 (ptrdiff_tree_p): Declare.
2156 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2157 extern inline implementations if the target doesn't use POLY_INT_CST.
2158 (poly_int_tree_p): New function.
2159 (wi::unextended_tree): New class.
2160 (wi::int_traits <unextended_tree>): New override.
2161 (wi::extended_tree): Add a default constructor.
2162 (wi::extended_tree::get_tree): New function.
2163 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2164 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2165 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2166 (wi::tree_to_poly_wide_ref): New typedefs.
2167 (wi::ints_for): Provide overloads for extended_tree and
2169 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2170 (wi::to_wide): New functions.
2171 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2172 * tree.c (poly_int_cst_hasher): New struct.
2173 (poly_int_cst_hash_table): New variable.
2174 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2175 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2177 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2178 (init_ttree): Initialize poly_int_cst_hash_table.
2179 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2180 a poly_int64 instead of a HOST_WIDE_INT.
2181 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2182 instead of an unsigned HOST_WIDE_INT.
2183 (wide_int_to_tree): Rename to...
2184 (wide_int_to_tree_1): ...this.
2185 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2186 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2187 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2188 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2189 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2190 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2192 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2193 (streamer_read_tree_body): Likewise.
2194 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2195 (streamer_write_tree_body): Likewise.
2196 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2197 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2198 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2199 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2200 * gimple-expr.h (is_gimple_constant): Likewise.
2201 * gimplify.c (maybe_with_size_expr): Likewise.
2202 * print-tree.c (print_node): Likewise.
2203 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2204 * tree-pretty-print.c (dump_generic_node): Likewise.
2205 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2206 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2207 * tree-vrp.c (compare_values_warnv): Likewise.
2208 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2209 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2210 (force_expr_to_var_cost): Likewise.
2211 * tree-ssa-loop.c (for_each_index): Likewise.
2212 * fold-const.h (build_invariant_address, size_int_kind): Take a
2213 poly_int64 instead of a HOST_WIDE_INT.
2214 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2215 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2217 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2218 (int_const_binop_2): New function, split out from...
2219 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2220 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2221 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2222 * gimple-ssa-strength-reduction.c (slsr_process_add)
2223 (slsr_process_mul): Check for INTEGER_CSTs before using them
2225 * stor-layout.c (bits_from_bytes): New function.
2226 (bit_from_pos): Use it.
2227 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2228 by BITS_PER_UNIT to get the TYPE_SIZE.
2229 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2230 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2232 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2233 Alan Hayward <alan.hayward@arm.com>
2234 David Sherwood <david.sherwood@arm.com>
2236 * doc/rtl.texi (const_poly_int): Document. Also document the
2237 rtl sharing behavior.
2238 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2239 * rtl.h (const_poly_int_def): New struct.
2240 (rtx_def::u): Add a cpi field.
2241 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2242 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2243 (wi::rtx_to_poly_wide_ref): New typedef
2244 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2245 (poly_int_rtx_p): New functions.
2246 (trunc_int_for_mode): Declare a poly_int64 version.
2247 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2248 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2250 (strip_offset): Declare.
2251 (strip_offset_and_add): New function.
2252 * rtl.def (CONST_POLY_INT): New rtx code.
2253 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2254 (shared_const_p): Use poly_int_rtx_p.
2255 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2257 (gen_int_shift_amount): Likewise.
2258 * emit-rtl.c (const_poly_int_hasher): New class.
2259 (const_poly_int_htab): New variable.
2260 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2261 (const_poly_int_hasher::hash): New function.
2262 (const_poly_int_hasher::equal): Likewise.
2263 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2264 (immed_wide_int_const): Rename to...
2265 (immed_wide_int_const_1): ...this and make static.
2266 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2267 instead of a wide_int_ref.
2268 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2269 (gen_lowpart_common): Handle CONST_POLY_INT.
2270 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2271 * cselib.c (cselib_hash_rtx): Likewise.
2272 * dwarf2out.c (const_ok_for_output_1): Likewise.
2273 * expr.c (convert_modes): Likewise.
2274 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2275 * rtlhash.c (add_rtx): Likewise.
2276 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2277 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2278 Handle existing CONST_POLY_INT rtxes.
2279 * expmed.h (expand_shift): Take a poly_int64 instead of a
2281 * expmed.c (expand_shift): Likewise.
2282 * rtlanal.c (strip_offset): New function.
2283 (commutative_operand_precedence): Give CONST_POLY_INT the same
2284 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2286 * rtl-tests.c (const_poly_int_tests): New struct.
2287 (rtl_tests_c_tests): Use it.
2288 * simplify-rtx.c (simplify_const_unary_operation): Handle
2290 (simplify_const_binary_operation): Likewise.
2291 (simplify_binary_operation_1): Fold additions of symbolic constants
2292 and CONST_POLY_INTs.
2293 (simplify_subreg): Handle extensions and truncations of
2295 (simplify_const_poly_int_tests): New struct.
2296 (simplify_rtx_c_tests): Use it.
2297 * wide-int.h (storage_ref): Add default constructor.
2298 (wide_int_ref_storage): Likewise.
2299 (trailing_wide_ints): Use GTY((user)).
2300 (trailing_wide_ints::operator[]): Add a const version.
2301 (trailing_wide_ints::get_precision): New function.
2302 (trailing_wide_ints::extra_size): Likewise.
2304 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2305 Alan Hayward <alan.hayward@arm.com>
2306 David Sherwood <david.sherwood@arm.com>
2308 * emit-rtl.h (gen_int_shift_amount): Declare.
2309 * emit-rtl.c (gen_int_shift_amount): New function.
2310 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2312 * calls.c (shift_return_value): Likewise.
2313 * cse.c (fold_rtx): Likewise.
2314 * dse.c (find_shift_sequence): Likewise.
2315 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2316 (expand_shift, expand_smod_pow2): Likewise.
2317 * lower-subreg.c (shift_cost): Likewise.
2318 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2319 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2320 (expand_vec_perm_var): Likewise.
2321 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2322 (simplify_binary_operation_1): Likewise.
2323 * combine.c (try_combine, find_split_point, force_int_to_mode)
2324 (simplify_shift_const_1, simplify_shift_const): Likewise.
2325 (change_zero_ext): Likewise. Use simplify_gen_binary.
2327 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2329 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2331 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2333 * doc/invoke.texi (ARM Options): Document accepted extension options
2334 for -march=armv8.3-a.
2336 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2339 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2340 or --with-float=softfp, set the default CPU to arm10e.
2342 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2344 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2345 * config/visium/predicates.md (const_shift_operand): Likewise.
2346 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2347 (visium_legitimize_reload_address): Likewise.
2349 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2352 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2354 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2357 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2359 2017-12-20 Julia Koval <julia.koval@intel.com>
2361 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2362 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2363 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2364 * config.gcc: Include vpclmulqdqintrin.h.
2365 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2366 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2368 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2369 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2370 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2371 * config/i386/i386.c (isa2_opts): Add -mcx16.
2372 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2373 (ix86_option_override_internal): Move mcx16 to flags2.
2374 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2375 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2376 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2377 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2378 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2379 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2380 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2381 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2382 * doc/invoke.texi: Add -mvpclmulqdq.
2384 2017-12-20 Tom de Vries <tom@codesourcery.com>
2387 * config/i386/i386.c (ix86_static_chain): Move
2388 DECL_STATIC_CHAIN test ...
2389 * calls.c (rtx_for_static_chain): ... here. New function.
2390 * calls.h (rtx_for_static_chain): Declare.
2391 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2392 instead of targetm.calls.static_chain.
2393 * df-scan.c (df_get_entry_block_def_set): Same.
2395 2017-12-19 Tom de Vries <tom@codesourcery.com>
2397 PR tree-optimization/83493
2398 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2401 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2403 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2404 inform with hardcoded english plural handling.
2406 2017-12-18 Jeff Law <law@redhat.com>
2408 PR tree-optimization/83477
2409 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2410 a non-virtual PHI, always push a new range.
2412 2017-12-19 Martin Sebor <msebor@redhat.com>
2415 * builtins.c (compute_objsize): Handle non-constant offsets.
2417 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2419 PR tree-optimization/83444
2420 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2421 character load case, if get_stridx on MEM_REF's operand doesn't
2422 look usable, retry with get_addr_stridx.
2424 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2427 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2428 (variable_tracking_main_1): Keep markers even when VTA fails.
2431 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2432 even if there are markers before it.
2433 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2435 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2437 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2438 typical order conditions.
2439 * sel-sched.c (extract_new_fences_from): Likewise.
2440 * config/visium/constraints.md (J, K, L): Likewise.
2441 * config/visium/predicates.md (const_shift_operand): Likewise.
2442 * config/visium/visium.c (visium_legitimize_address,
2443 visium_legitimize_reload_address): Likewise.
2444 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2445 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2446 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2447 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2448 SET_DUMP_DETAIL): Likewise.
2449 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2450 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2451 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2452 avr_set_core_architecture, avr_set_current_function,
2453 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2454 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2455 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2456 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2457 avr_map_decompose, avr_fold_builtin): Likewise.
2458 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2459 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2460 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2461 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2462 * config/m32c/m32c.c (m32c_conditional_register_usage,
2463 m32c_address_cost): Likewise.
2464 * config/m32c/predicates.md (shiftcount_operand,
2465 longshiftcount_operand): Likewise.
2466 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2467 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2468 can_use_cdx_ldstw): Likewise.
2469 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2470 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2472 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2473 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2474 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2475 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2476 * config/vax/vax.c (adjacent_operands_p): Likewise.
2477 * config/ft32/constraints.md (L, b, KA): Likewise.
2478 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2480 * cfgexpand.c (expand_stack_alignment): Likewise.
2481 * gcse.c (insert_expr_in_table): Likewise.
2482 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2483 * cgraphunit.c (cgraph_node::expand): Likewise.
2484 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2485 * emit-rtl.c (add_insn): Likewise.
2486 * input.c (dump_location_info): Likewise.
2487 * passes.c (NEXT_PASS): Likewise.
2488 * read-rtl-function.c (parse_note_insn_name,
2489 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2491 * sched-rgn.c (sched_rgn_init): Likewise.
2492 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2493 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2494 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2496 * explow.c (eliminate_constant_term): Likewise.
2497 * final.c (leaf_renumber_regs_insn): Likewise.
2498 * cfgrtl.c (print_rtl_with_bb): Likewise.
2499 * genhooks.c (emit_init_macros): Likewise.
2500 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2501 * tree-data-ref.c (conflict_fn): Likewise.
2502 * selftest.c (assert_streq): Likewise.
2503 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2504 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2505 fold_binary_loc, multiple_of_p): Likewise.
2506 * reload.c (push_reload, find_equiv_reg): Likewise.
2507 * et-forest.c (et_nca, et_below): Likewise.
2508 * dbxout.c (dbxout_symbol_location): Likewise.
2509 * reorg.c (relax_delay_slots): Likewise.
2510 * dojump.c (do_compare_rtx_and_jump): Likewise.
2511 * gengtype-parse.c (type): Likewise.
2512 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2513 simplify_const_relational_operation): Likewise.
2514 * reload1.c (do_output_reload): Likewise.
2515 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2516 * gengtype.c (type_for_name): Likewise.
2517 * gimple-ssa-sprintf.c (format_directive): Likewise.
2519 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2522 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2523 accessing it. Adjust comment.
2525 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2528 * predict.c (zero_one_minusone): New function.
2529 (apply_return_prediction): Avoid return prediction for functions
2530 returning only -1, 0 and 1 values, unless they only return -1 and 0
2533 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2535 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2536 fall-through warning.
2537 (arc_compute_frame_size): Remove unused variables.
2538 (arc_print_operand): Fix fprintif format.
2539 (arc_can_follow_jump): Clean fall-through warning.
2541 2017-12-19 Marek Polacek <polacek@redhat.com>
2544 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2547 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2549 * config/arc/arc.c (overriderregs): New variable.
2550 (arc_override_options): Track fixed/call saved/call options.
2551 (arc_conditional_register_usage): Check against overrideregs
2552 variable whenever we change register properties.
2554 2017-12-19 Nathan Sidwell <nathan@acm.org>
2556 * opts.c (finish_options): Don't prefix dump_base_name if it
2557 already contains directories.
2559 2017-12-19 Martin Liska <mliska@suse.cz>
2561 PR rtl-optimization/82675
2562 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2563 more element in sbitmap.
2565 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2567 * gimplify.c (gimplify_expr): Use error_operand_p.
2569 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2572 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2575 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2577 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2578 the jump is to a label.
2580 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2582 PR tree-optimization/83444
2583 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2588 * ipa-inline.c (flatten_remove_node_hook): New function.
2589 (ipa_inline): Keep only nodes with flatten attribute at the end of
2590 the array in the order from ipa_reverse_postorder, only walk that
2591 portion of array for flattening, if there is more than one such
2592 node, temporarily register a removal hook and ignore removed nodes.
2594 PR tree-optimization/80631
2595 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2596 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2597 IFN_REDUC_MAX or IFN_REDUC_MIN.
2599 2017-12-18 Jeff Law <law@redhat.com>
2601 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2602 record symbolic equivalences from backedges in the CFG.
2605 2017-11-19 Jeff Law <law@redhat.com>
2607 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2608 of degenerates resulting from ignoring an edge.
2610 2017-12-18 Martin Sebor <msebor@redhat.com>
2613 PR tree-optimization/78450
2614 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2615 (handle_builtin_strlen): Call it.
2617 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2619 PR rtl-optimization/83424
2620 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2622 2017-12-18 Marek Polacek <polacek@redhat.com>
2625 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2626 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2629 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2631 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2632 (umaddsidi4, umaddsidi_split): Likewise.
2634 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2636 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2639 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2642 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2644 2017-12-18 Richard Biener <rguenther@suse.de>
2646 PR tree-optimization/81877
2647 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2648 (outermost_indep_loop): Adjust.
2649 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2650 (can_sm_ref_p): Adjust.
2652 2017-12-18 Richard Biener <rguenther@suse.de>
2655 * tree.c (array_at_struct_end_p): Return true if the underlying
2656 object has space for at least one element in excess of what
2657 the array domain specifies.
2659 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2661 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2662 example to avoid overfull hbox.
2663 * doc/invoke.texi (Option Summary): Add missing @gol.
2664 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2667 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2668 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2670 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2671 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2672 (AArch64 Options): Document -mverbose-cost-dump.
2673 (ARM Options): Likewise, plus -mflip-thumb.
2675 2017-12-17 Martin Sebor <msebor@redhat.com>
2678 * gimple-ssa-warn-restrict.c
2679 (builtin_memref::offset_out_of_bounds): Correct the handling of
2682 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2684 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2686 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2688 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2690 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2692 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2695 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2697 * doc/invoke.texi: Fix some typos.
2699 2017-12-16 Martin Sebor <msebor@redhat.com>
2701 PR tree-optimization/78918
2702 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2703 * builtins.c (check_sizes): Rename...
2704 (check_access): ...to this. Rename function arguments for clarity.
2705 (check_memop_sizes): Adjust names.
2706 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2707 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2708 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2709 (check_strncat_sizes, expand_builtin_strncat): Same.
2710 (expand_builtin_strncpy, expand_builtin_memset): Same.
2711 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2712 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
2713 (maybe_emit_sprintf_chk_warning): Same.
2714 (expand_builtin_strcpy): Adjust.
2715 (expand_builtin_stpcpy): Same.
2716 (expand_builtin_with_bounds): Detect out-of-bounds accesses
2717 in pointer-checking forms of memcpy, memmove, and mempcpy.
2718 (gcall_to_tree_minimal, max_object_size): Define new functions.
2719 * builtins.h (max_object_size): Declare.
2720 * calls.c (alloc_max_size): Call max_object_size instead of
2721 hardcoding ssizetype limit.
2722 (get_size_range): Handle new argument.
2723 * calls.h (get_size_range): Add a new argument.
2724 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
2725 * doc/invoke.texi (-Wrestrict): Adjust, add example.
2726 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
2728 (gimple_fold_builtin_memory_chk): Same.
2729 (gimple_fold_builtin_stxcpy_chk): New function.
2730 * gimple-ssa-warn-restrict.c: New source.
2731 * gimple-ssa-warn-restrict.h: New header.
2732 * gimple.c (gimple_build_call_from_tree): Propagate location.
2733 * passes.def (pass_warn_restrict): Add new pass.
2734 * tree-pass.h (make_pass_warn_restrict): Declare.
2735 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2737 (handle_builtin_strcat): Same.
2738 (strlen_optimize_stmt): Rename...
2739 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2740 stpncpy, strncpy, and their checking forms.
2742 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2744 PR rtl-optimization/82849
2745 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2746 and get_max_loop_iterations_int.
2748 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2749 Alan Hayward <alan.hayward@arm.com>
2750 David Sherwood <david.sherwood@arm.com>
2752 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2753 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2754 size as a poly_uint64.
2755 (mode_for_vector, mode_for_int_vector): Take the number of vector
2756 elements as a poly_uint64.
2757 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2758 size as a poly_uint64.
2759 (mode_for_vector, mode_for_int_vector): Take the number of vector
2760 elements as a poly_uint64.
2762 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2763 Alan Hayward <alan.hayward@arm.com>
2764 David Sherwood <david.sherwood@arm.com>
2766 * machmode.h (MACRO_MODE): New macro.
2767 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2768 * targhooks.c (default_libcall_value, default_secondary_reload)
2769 (default_memory_move_cost, default_register_move_cost)
2770 (default_class_max_nregs): Likewise.
2772 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2773 Alan Hayward <alan.hayward@arm.com>
2774 David Sherwood <david.sherwood@arm.com>
2776 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2777 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
2779 * genautomata.c (main): Likewise.
2780 * genconditions.c (write_header): Likewise.
2781 * genemit.c (main): Likewise.
2782 * genextract.c (print_header): Likewise.
2783 * genopinit.c (main): Likewise.
2784 * genoutput.c (output_prologue): Likewise.
2785 * genpeep.c (main): Likewise.
2786 * genpreds.c (write_insn_preds_c): Likewise.
2787 * genrecog.c (writer_header): Likewise.
2788 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
2789 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
2790 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2791 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2792 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2793 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2794 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2795 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2796 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2797 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2798 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2799 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2800 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2801 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2802 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2803 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2804 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2805 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2806 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2807 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2808 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2809 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2810 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2811 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2812 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2813 * config/darwin.c (IN_TARGET_CODE): Likewise.
2814 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2815 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2816 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2817 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2818 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2819 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2820 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2821 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2822 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2823 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2824 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2825 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2826 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2827 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2828 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2829 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2830 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2831 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2832 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2833 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2834 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2835 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2836 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2837 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2838 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2839 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2840 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2841 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2842 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2843 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2844 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2845 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2846 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2847 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2848 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2849 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2850 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2851 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2852 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2853 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2854 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2855 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2856 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2857 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2858 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2859 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2860 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2861 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2862 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2863 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2864 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2865 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2866 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2867 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2868 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2869 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2870 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2871 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2872 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2873 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2874 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2875 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2876 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2877 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2878 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2879 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2880 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2881 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2882 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2883 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2884 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2885 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2886 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2887 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2888 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2889 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2890 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2891 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2892 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2893 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2894 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2895 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2896 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2897 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2898 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2899 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2900 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2901 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2902 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2903 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2904 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2905 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2906 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2907 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2908 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2909 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2910 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2911 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2912 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2913 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2914 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2915 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2916 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2917 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2918 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2919 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2921 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2922 Alan Hayward <alan.hayward@arm.com>
2923 David Sherwood <david.sherwood@arm.com>
2925 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2928 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2929 Alan Hayward <alan.hayward@arm.com>
2930 David Sherwood <david.sherwood@arm.com>
2932 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2933 * doc/md.texi (vec_series@var{m}): Document.
2934 * tree.def (VEC_SERIES_EXPR): New tree code.
2935 * tree.h (build_vec_series): Declare.
2936 * tree.c (build_vec_series): New function.
2937 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2938 * tree-pretty-print.c (dump_generic_node): Likewise.
2939 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2940 * tree-inline.c (estimate_operator_cost): Likewise.
2941 * expr.c (expand_expr_real_2): Likewise.
2942 * optabs-tree.c (optab_for_tree_code): Likewise.
2943 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2944 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2945 * expmed.c (make_tree): Handle VEC_SERIES.
2946 * optabs.def (vec_series_optab): New optab.
2947 * optabs.h (expand_vec_series_expr): Declare.
2948 * optabs.c (expand_vec_series_expr): New function.
2949 * tree-vect-generic.c (expand_vector_operations_1): Check that
2950 the operands also have vector type.
2952 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2953 Alan Hayward <alan.hawyard@arm.com>
2954 David Sherwood <david.sherwood@arm.com>
2956 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2957 (VEC_COND_EXPR): Add missing @tindex.
2958 * doc/md.texi (vec_duplicate@var{m}): Document.
2959 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2960 * tree.c (build_vector_from_val): Add stubbed-out handling of
2961 variable-length vectors, using VEC_DUPLICATE_EXPR.
2962 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2963 * cfgexpand.c (expand_debug_expr): Likewise.
2964 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2965 * tree-inline.c (estimate_operator_cost): Likewise.
2966 * tree-pretty-print.c (dump_generic_node): Likewise.
2967 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2968 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2969 (test_vec_duplicate_folding): New function.
2970 (fold_const_c_tests): Call it.
2971 * optabs.def (vec_duplicate_optab): New optab.
2972 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2973 * optabs.h (expand_vector_broadcast): Declare.
2974 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2975 vec_duplicate_optab.
2976 * expr.c (store_constructor): Try using vec_duplicate_optab for
2978 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2980 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2983 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2984 div/mod latencies a bit.
2986 2017-12-15 Jeff Law <law@redhat.com>
2988 PR tree-optimization/36550
2989 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2990 (mark_threaded_blocks): Rewrite code to avoid block copying when
2991 optimizing for size. Don't pessimize blocks which will be
2992 copied, but all the statements will be dead.
2994 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2996 PR tree-optimization/81165
2997 * tree-ssa-threadupdate.c (uses_in_bb): New.
2998 (estimate_threading_killed_stmts): New.
2999 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3000 * tree-ssa-threadedge.c
3001 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3004 2017-12-15 Jeff Law <law@redhat.com>
3006 PR tree-optimization/83410
3007 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3008 threads when parallelizing loops.
3010 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3012 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3014 * config/alpha/alpha.c (vms_attribute_table): Swap
3015 affects_type_identity and handler fields, adjust comments.
3016 * config/mips/mips.c (mips_attribute_table): Likewise.
3017 * config/visium/visium.c (visium_attribute_table): Likewise.
3018 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3019 * config/microblaze/microblaze.c (microblaze_attribute_table):
3021 * config/spu/spu.c (spu_attribute_table): Likewise.
3022 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3023 * config/arc/arc.c (arc_attribute_table): Likewise.
3024 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3025 * config/v850/v850.c (v850_handle_interrupt_attribute,
3026 v850_handle_data_area_attribute): Formatting fixes.
3027 (v850_attribute_table): Swap affects_type_identity and handler
3028 fields, adjust comments.
3029 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3030 * config/arm/arm.c (arm_attribute_table): Likewise.
3031 * config/avr/avr.c (avr_attribute_table): Likewise.
3032 * config/s390/s390.c (s390_attribute_table): Likewise.
3033 * config/sh/sh.c (sh_attribute_table): Likewise.
3034 * config/i386/i386.c (ix86_handle_cconv_attribute,
3035 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3036 (ix86_attribute_table): Swap affects_type_identity and handler
3037 fields, adjust comments.
3038 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3039 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3040 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3041 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3042 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3043 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3044 * config/rx/rx.c (rx_attribute_table): Likewise.
3045 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3046 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3047 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3048 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3049 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3050 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3051 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3052 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3053 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3054 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3055 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3057 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3059 (register_scoped_attributes, decl_attributes): Formatting fixes.
3061 PR tree-optimization/83269
3062 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3063 subtraction in arg0's type if type is signed and arg0 is unsigned.
3067 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3068 convert? on @0 instead of convert. Check type of @1, not @0.
3069 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3070 convert? on @0 instead of convert. Check type of @1, not @0.
3071 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3072 only optimize if either both @1 and @2 types are narrower
3073 precision, or both are wider or equal precision, and in the former
3074 case only if both have undefined overflow.
3076 2017-12-15 Richard Biener <rguenther@suse.de>
3079 * internal-fn.def (IFN_NOP): Add.
3080 * internal-fn.c (expand_NOP): Do nothing.
3081 * lto-streamer-in.c (input_function): Instead of removing
3082 sanitizer calls replace them with IFN_NOP calls.
3084 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3085 Alan Hayward <alan.hayward@arm.com>
3086 David Sherwood <david.sherwood@arm.com>
3088 * dse.c (store_info, read_info_type): Replace begin and end with
3090 (print_range): New function.
3091 (set_all_positions_unneeded, any_positions_needed_p)
3092 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3094 (record_store): Likewise. Optimize the case in which all positions
3096 (get_stored_val): Replace read_begin and read_end with read_offset
3098 (replace_read): Update call accordingly.
3100 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3102 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3103 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3104 (loop_cand::loop_cand): Initialize above members.
3105 (loop_cand::supported_operations): Delete.
3106 (loop_cand::can_interchange_p): Inline above function.
3107 (loop_cand::classify_simple_reduction): Record number of constant
3108 initialized simple reductions.
3109 (should_interchange_loops): New parameters. Check stmt cost of loops
3111 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3112 Update call to should_interchange_loops.
3113 (should_interchange_loop_nest): Update call to
3114 should_interchange_loops.
3116 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3119 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3120 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3121 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3123 2017-12-15 Julia Koval <julia.koval@intel.com>
3125 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3126 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3128 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3129 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3130 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3132 2017-12-15 Julia Koval <julia.koval@intel.com>
3134 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3135 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3136 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3137 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3138 _mm_aesenc_epi128): New intrinsics.
3140 2017-12-15 Julia Koval <julia.koval@intel.com>
3142 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3143 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3145 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3146 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3147 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3149 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3151 * gimple-ssa-strength-reduction.c (analyze_increments):
3152 Distinguish replacement costs for constant strides from those for
3155 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3157 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3159 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3161 * doc/invoke.texi: Document -Wcast-function-type.
3162 * recog.h (stored_funcptr): Change signature.
3163 * tree-dump.c (dump_node): Avoid warning.
3164 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3166 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3169 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3172 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3174 PR tree-optimization/83312
3175 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3176 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3177 "val" param, and to cope with arbitrary basic blocks.
3178 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3179 handle NULL_TREE for "val", dropping "bb" param.
3180 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3181 drop "bb" param. Handle NULL_TREE for "val".
3182 (find_case_label_for_value): Make "switch_stmt" param const.
3183 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3185 (vrp_prop::check_all_array_refs): Reimplement as...
3186 (check_array_bounds_dom_walker::before_dom_children): ...this new
3187 vfunc. Replace linear search through BB block list, excluding
3188 those with non-executable in-edges via dominator walk.
3190 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3192 * config/arm/arm.opt (mverbose-cost-dump): New option.
3193 * config/arm/arm.c (arm_rtx_costs): Use it.
3195 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3198 * reload1.c (emit_input_reload_insns): Skip debug markers.
3200 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3202 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3206 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3207 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3208 * config/sh/sh.c: Skip debug insns besides notes.
3209 * config/sh/sh.md: Likewise.
3210 * config/sh/sh_treg_combine.cc: Likewise.
3211 * config/sh/sync.md: Likewise.
3213 2017-12-14 Tom de Vries <tom@codesourcery.com>
3215 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3216 Add item for weak_undefined.
3218 2017-12-14 Richard Biener <rguenther@suse.de>
3220 PR tree-optimization/67842
3221 * sese.h (bb_in_region): Remove #if 0'ed code.
3223 2017-12-14 Richard Biener <rguenther@suse.de>
3225 PR tree-optimization/83326
3226 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3227 may_be_zero parameter and handle it by not marking the first
3228 peeled copy as not exiting the loop.
3229 (try_peel_loop): Likewise.
3230 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3231 to handle the case of constant or zero iterations and perform
3232 loop header copying on-the-fly.
3234 2017-12-14 Richard Biener <rguenther@suse.de>
3236 PR tree-optimization/83418
3237 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3238 Instead of asserting we don't get unfolded comparisons deal with them.
3240 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3243 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3244 basic blocks. Assert debug bind insns don't appear outside of bbs,
3245 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3248 PR tree-optimization/83198
3249 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3250 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3251 value if arg is a REAL_CST with incompatible type.
3253 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3254 Bin Cheng <bin.cheng@arm.com>
3257 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3259 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3262 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3264 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3265 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3267 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3268 Alan Hayward <alan.hayward@arm.com>
3269 David Sherwood <david.sherwood@arm.com>
3271 * poly-int.h: New file.
3272 * poly-int-types.h: Likewise.
3273 * coretypes.h: Include them.
3274 (POLY_INT_CONVERSION): Define.
3275 * target.def (estimated_poly_value): New hook.
3276 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3277 * doc/tm.texi: Regenerate.
3278 * doc/poly-int.texi: New file.
3279 * doc/gccint.texi: Include it.
3280 * doc/rtl.texi: Describe restrictions on subreg modes.
3281 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3282 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3283 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3284 * targhooks.h (default_estimated_poly_value): Declare.
3285 * targhooks.c (default_estimated_poly_value): New function.
3286 * target.h (estimated_poly_value): Likewise.
3287 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3288 (wi::unary_traits): Delete.
3289 (wi::binary_traits::signed_shift_result_type): Define for
3290 offset_int << HOST_WIDE_INT, etc.
3291 (generic_wide_int::operator <<=): Define for all types and use
3292 wi::lshift instead of <<.
3293 (wi::hwi_with_prec): Add a default constructor.
3294 (wi::ints_for): New class.
3295 (operator <<): Define for all wide-int types.
3296 (operator /): New function.
3297 (operator %): Likewise.
3298 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3299 (ASSERT_MAYBE_NE_AT): New macros.
3301 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3302 Dominik Vogt <vogt@linux.vnet.ibm.com>
3305 * emit-rtl.c (init_emit): Remove ??? comment.
3306 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3307 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3308 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3309 alignment of 3 virtual registers to BITS_PER_WORD.
3311 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3313 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3315 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3316 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3318 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3321 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3323 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3325 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3326 with the exception of debug begin stmt markers appear before
3330 * final.c (rest_of_handle_final): Call variable_tracking_main only
3331 if !flag_var_tracking.
3333 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3334 Jakub Jelinek <jakub@redhat.com>
3338 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3339 labels debug stmts that can only appear after labels.
3341 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3343 PR rtl-optimization/82398
3344 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3345 EXPR_USEFULNESS in priority comparison.
3347 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3349 PR rtl-optimization/83393
3350 * combine.c (move_deaths): If reg_stat points to a too new insn in
3351 last_death, do not use it: find the proper insn instead.
3353 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3356 * config/rs6000/rs6000-string.c (expand_block_compare,
3357 expand_strn_compare): Fix set_mem_size() calls.
3359 2017-12-12 Jeff Law <law@redhat.com>
3361 PR tree-optimization/83298
3362 PR tree-optimization/83362
3363 PR tree-optimization/83383
3364 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3365 push_value_range a public interface. Add new argument to
3366 record_ranges_from_stmt.
3367 * gimple-ssa-evrp-analyze.c
3368 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3369 Update comments. Handle recording temporary equivalences.
3370 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3371 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3372 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3373 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3374 gimple-ssa-evrp-analyze.h.
3375 (record_temporary_equivalences_from_phis): Add new argument. When
3376 the PHI arg is an SSA_NAME, set the result's range to the range
3378 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3379 from statements too.
3380 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3381 Pass it down to children as needed.
3382 (thread_outgoing_edges): Likewise.
3383 (thread_across_edge): Likewise. Push/pop range state as needed.
3384 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3386 2017-12-12 Julia Koval <julia.koval@intel.com>
3388 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3389 (PTA_CANNONLAKE): Remove PTA_CLWB.
3391 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3394 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3395 (vcondv2div2df): Likewise.
3396 (vconduv2dfv2di): Likewise.
3398 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3400 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3401 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3402 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3403 ATTR_MATHFN_FPROUNDING_ERRNO.
3405 2017-12-12 Richard Biener <rguenther@suse.de>
3407 PR tree-optimization/83385
3408 * graphite-scop-detection.c (get_order, order): Remove.
3409 (bb_to_rpo): New global.
3411 (build_scops): Sort pbbs in RPO order.
3413 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3415 * combine.c (simplify_set): Do not transform subregs to zero_extends
3416 if the destination mode is a vector mode.
3418 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3420 PR tree-optimization/83359
3421 * tree-cfg.h (fold_loop_internal_call): Declare.
3422 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3423 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3424 (find_loop_dist_alias): New function.
3425 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3426 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3429 PR tree-optimization/80631
3430 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3431 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3432 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3433 hardcoding zero as the value if COND_EXPR is never true. For
3434 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3435 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3436 hardcoding MAX_EXPR as the reduction operation.
3437 (is_nonwrapping_integer_induction): Allow negative step.
3438 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3439 vect_create_epilog_for_reduction, if no value is suitable, don't
3440 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3442 2017-12-12 Richard Biener <rguenther@suse.de>
3444 PR tree-optimization/81889
3445 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3446 range info from the non-wrapping IV instead of just the range
3449 2017-12-12 Julia Koval <julia.koval@intel.com>
3451 * config.gcc: Add vaesintrin.h.
3452 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3454 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3455 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3457 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3458 * config/i386/immintrin.h: Include vaesintrin.h.
3459 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3460 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3461 _mm_aesdec_epi128): New intrinsics.
3463 2017-12-12 Julia Koval <julia.koval@intel.com>
3465 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3466 OPTION_MASK_ISA_VAES_UNSET): New.
3467 (ix86_handle_option): Handle -mvaes.
3468 * config/i386/cpuid.h: Define bit_VAES.
3469 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3470 * config/i386/i386-c.c (__VAES__): New.
3471 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3472 (ix86_valid_target_attribute_inner_p): Ditto.
3473 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3474 * config/i386/i386.opt: Add -mvaes.
3475 * doc/invoke.texi: Ditto.
3477 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3479 * debug.h (gcc_debug_hooks): Add inline_entry.
3480 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3481 * debug.c (do_nothing_debug_hooks): Likewise.
3482 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3483 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3484 (dwarf2_lineno_debug_hooks): Likewise.
3486 * common.opt (gstatement-frontiers): New, setting
3487 debug_nonbind_markers_p.
3488 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3489 * toplev.c (process_options): Autodetect value for debug statement
3491 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3492 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3494 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3495 markers. Integrate source bind into debug stmt expand loop.
3496 (pass_expand::execute): Check debug marker limit. Avoid deep
3497 TER and expand debug locations for debug bind insns only.
3498 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3500 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3501 * final.c (reemit_insn_block_notes): Take current block from
3502 nonbind markers. Declare note where it's first set.
3503 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3504 begin stmt markers if enabled.
3505 (notice_source_line): Handle nonbind markers. Fail if their
3506 location is unknown or that of builtins.
3507 (rest_of_handle_final): Convert begin stmt markers to notes if
3508 var-tracking didn't run.
3509 (rest_of_clean_state): Skip begin stmt markers.
3510 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3512 * function.c (allocate_struct_function): Set begin_stmt_markers.
3513 * function.h (struct function): Add debug_marker_count counter
3514 and debug_nonbind_markers flag.
3515 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3516 * gimple-low.c (lower_function_body): Adjust
3517 debug_nonbind_markers.
3518 (lower_stmt): Drop or skip gimple debug stmts.
3519 (lower_try_catch): Skip debug stmts.
3520 * gimple.c (gimple_build_debug_begin_stmt): New.
3521 (gimple_copy): Increment debug_marker_count if copying one.
3522 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3523 * gimplify.c (rexpr_location): New.
3524 (rexpr_has_location): New.
3525 (warn_switch_unreachable_r): Handle gimple debug stmts.
3526 (shortcut_cond_r): Call expr_location.
3528 (find_goto_label): New.
3529 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3531 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3533 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3534 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3535 (LANG_HOOKS_INITIALIZER): ... this.
3536 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3537 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3539 (update_ebb_live_info): Skip debug insn markers.
3540 * lra.c (debug_insn_static_data): Rename to...
3541 (debug_bind_static_data): ... this.
3542 (debug_marker_static_data): New.
3543 (lra_set_insn_recog_data): Select one of the above depending
3545 (lra_update_isn_regno_info): Don't assume debug insns have
3547 (push_insns): Skip debug insns.
3548 * lto-streamer-in.c (input_function): Drop debug stmts
3549 depending on active options. Adjust debug_nonbind_markers.
3550 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3551 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3552 begin stmt marker notes.
3553 (print_insn): Likewise.
3554 * recog.c (extract_insn): Recognize rtl for debug markers.
3555 * rtl.def (DEBUG_MARKER): New.
3556 * tree-inline.c: Include params.h.
3557 (remap_gimple_stmt): Handle nonbind markers.
3558 (maybe_move_debug_stmts_to_successors): Likewise.
3559 (copy_debug_stmt): Likewise.
3560 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3561 markers regardless of no side effects.
3562 (tsi_link_before): Don't update container's side effects when adding
3563 a begin stmt marker.
3564 (tsi_link_after): Likewise.
3565 (expr_first): Skip begin stmt markers.
3566 (expr_last): Likewise.
3567 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3568 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3569 Disregard nonbind markers.
3570 * tree.c (make_node_stat): Don't set side effects for begin stmt
3572 (build1_stat): Likewise.
3573 * tree.def (DEBUG_BEGIN_STMT): New.
3574 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3575 * var-tracking.c (delete_debug_insns): Renamed to...
3576 (delete_vta_debug_insns): ... this.
3577 (reemit_marker_as_note): New.
3578 (vt_initialize): Reemit markers.
3579 (delete_vta_debug_insns): Likewise.
3580 (vt_debug_insns_local): Reemit or delete markers.
3581 (variable_tracking_main_1): Likewise.
3582 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3583 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3584 (gimple_debug_nonbind_marker_p): New.
3585 (gimple_build_debug_bind): Adjust.
3586 (gimple_build_debug_begin_stmt): New.
3587 * doc/invoke.texi (max-debug-marker-count): New param.
3588 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3589 (debug_parameter_ref, debug_marker): New.
3590 (NOTE_INSN_BEGIN_STMT): New.
3591 (DEBUG_INSN): Describe begin stmt markers.
3593 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3594 without debug insns, we wouldn't, but clean up debug insns
3595 after a control flow insn nevertheless.
3597 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3598 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3599 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3600 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3601 (rtl_tidy_fallthru_edge): Likewise.
3602 (rtl_verify_fallthru): Likewise.
3603 (rtl_verify_bb_layout): Likewise.
3604 (skip_insns_after_block): Likewise.
3605 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3606 * dwarf2out.c: Include print-rtl.h.
3607 (dwarf2out_next_real_insn): New.
3608 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3609 Dump debug binds in asm comments.
3610 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3611 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3612 callers to use gsi_start_nondebug_bb instead.
3613 (gsi_after_labels): Skip gimple debug stmts.
3614 (gsi_start_nondebug): New.
3615 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3616 (proper_loop_form_for_interchange): Adjust.
3617 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3618 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3619 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3620 (collect_fallthrough_labels): Likewise.
3621 (should_warn_for_implicit_fallthrough): Likewise.
3622 (warn_implicit_fallthrough_r): Likewise.
3623 (expand_FALLTHROUGH_r): Likewise.
3624 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3625 (graphite_copy_stmts_from_block): Skip nonbind markers.
3626 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3627 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3628 * jump.c (clean_barriers): Skip debug insns.
3629 * omp-expand.c (expand_parallel_call): Skip debug insns.
3630 (expand_task_call): Likewise.
3631 (remove_exit_barrier): Likewise.
3632 (expand_omp_taskreg): Likewise.
3633 (expand_omp_for_init_counts): Likewise.
3634 (expand_omp_for_generic): Likewise.
3635 (expand_omp_for_static_nochunk): Likewise.
3636 (expand_omp_for_static_chunk): Likewise.
3637 (expand_omp_simd): Likewise.
3638 (expand_omp_taskloop_for_outer): Likewise.
3639 (expand_omp_taskloop_for_inner): Likewise.
3640 (expand_oacc_for): Likewise.
3641 (expand_omp_sections): Likewise.
3642 (expand_omp_single): Likewise.
3643 (expand_omp_synch): Likewise.
3644 (expand_omp_atomic_load): Likewise.
3645 (expand_omp_atomic_store): Likewise.
3646 (expand_omp_atomic_fetch_op): Likewise.
3647 (expand_omp_atomic_pipeline): Likewise.
3648 (expand_omp_atomic_mutex): Likewise.
3649 (expand_omp_target): Likewise.
3650 (grid_expand_omp_for_loop): Likewise.
3651 (grid_expand_target_grid_body): Likewise.
3652 (build_omp_regions_1): Likewise.
3653 * omp-low.c (check_combined_parallel): Skip debug stmts.
3654 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3655 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3656 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3658 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3659 (make_edges): Likewise.
3660 (cleanup_dead_labels): Likewise.
3661 (gimple_can_merge_blocks_p): Likewise.
3662 (stmt_starts_bb_p): Likewise.
3663 (gimple_block_label): Likewise.
3664 (gimple_redirect_edge_and_branch): Likewise.
3665 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3667 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3669 * tree-pretty-print (print_declaration): Omit initializer in slim
3671 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3673 (eliminate_unnecessary_stmts): Stabilize block removal order.
3674 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3675 * var-tracking.c (get_first_insn): New.
3676 (vt_emit_notes): Call it.
3677 (vt_initialize): Walk any insns before the first BB.
3678 (delete_debug_insns): Likewise.
3680 * gimple.h (enum gimple_debug_subcode): Add
3681 GIMPLE_DEBUG_BEGIN_STMT.
3682 (gimple_debug_begin_stmt_p): New.
3683 (gimple_debug_nonbind_marker_p): New.
3684 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3685 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3686 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3687 * insn-notes.def (BEGIN_STMT): New.
3688 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3689 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3690 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3691 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3692 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3693 (INSN_DEBUG_MARKER_KIND): New.
3694 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3695 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3696 (INSN_VAR_LOCATION_PTR): New.
3697 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3699 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3700 deps only if debug bind insns are enabled.
3701 (pass_expand::execute): Avoid deep TER and expand
3702 debug locations for debug bind insns only.
3703 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3704 debug stmts special handling down to debug bind stmts.
3705 * combine.c (try_combine): Narrow debug insns special handling
3706 down to debug bind insns.
3707 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3708 Narrow debug insns preexisting special handling down to debug
3710 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3711 handling down to debug bind insns.
3712 * function.c (instantiate_virtual_regs): Skip debug markers,
3713 adjust handling of debug binds.
3714 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
3715 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
3716 * haifa-sched.c (schedule_insn): Narrow special handling of debug
3717 insns to debug bind insns.
3718 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
3719 special handling of debug stmts to debug bind stmts.
3720 * ipa-split.c (split_function): Likewise.
3721 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
3722 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
3724 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
3725 * regrename.c (build_def_use): Likewise.
3726 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
3727 (pass_cprop_hardreg): Narrow special casing of debug insns to
3729 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
3730 * reload1.c (reload): Likewise.
3731 * sese.c (sese_insert_phis_for_liveouts): Narrow special
3732 casing of debug stmts to debug bind stmts.
3733 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
3734 * ssa-iterators.h (num_imm_uses): Likewise.
3735 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
3736 debug stmts to debug bind stmts.
3737 * tree-inline.c (tree_function_versioning): Narrow special casing
3738 of debug stmts to debug bind stmts.
3739 * tree-loop-distribution.c (generate_loops_for_partition):
3740 Narrow special casing of debug stmts to debug bind stmts.
3741 * tree-sra.c (analyze_access_subtree): Narrow special casing
3742 of debug stmts to debug bind stmts.
3743 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3744 stmts to debug bind stmts.
3745 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3746 casing of debug stmts to debug bind stmts.
3747 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3748 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3749 casing of debug stmts to debug bind stmts.
3750 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3752 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3753 debug stmts to debug bind stmts.
3754 (gimple_replace_ssa_lhs): Likewise.
3755 (insert_debug_temp_for_var_def): Likewise.
3756 (insert_debug_temps_for_defs): Likewise.
3757 (reset_debug_uses): Likewise.
3758 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3759 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3760 (adjust_debug_stmts): Likewise.
3761 (adjust_phi_and_debug_stmts): Likewise.
3762 (vect_do_peeling): Likewise.
3763 * tree-vect-loop.c (vect_transform_loop): Likewise.
3764 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3765 * var-tracking.c (adjust_mems): Narrow special casing of debug
3766 insns to debug bind insns.
3767 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
3768 (compute_bb_dataflow, vt_find_locations): Likewise.
3769 (vt_expand_loc, emit_notes_for_changes): Likewise.
3770 (vt_init_cfa_base): Likewise.
3771 (vt_emit_notes): Likewise.
3772 (vt_initialize): Likewise.
3773 (vt_finalize): Likewise.
3775 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3776 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3777 (next_nonnote_nondebug_insn_bb): New.
3778 (prev_nonnote_nondebug_insn_bb): New.
3779 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
3780 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
3781 (prev_nonnote_nondebug_insn_bb): Declare.
3782 (next_nonnote_nondebug_insn_bb): Declare.
3783 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
3784 * cfgrtl.c (get_last_bb_insn): Likewise.
3785 * lra.c (push_insns): Likewise.
3787 2017-12-11 David Malcolm <dmalcolm@redhat.com>
3790 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
3792 * selftest.c (selftest::test_runner::test_runner): New ctor.
3793 (selftest::test_runner::~test_runner): New dtor.
3794 * selftest.h (class selftest::test_runner): New class.
3796 2017-12-11 Carl Love <cel@us.ibm.com>
3798 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3799 vec_extract_fp32_from_shortl]): Add #defines.
3800 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3801 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3802 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3803 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3804 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3805 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3806 * doc/extend.texi: Add documentation for the added builtins.
3808 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3810 PR rtl-optimization/80693
3811 PR rtl-optimization/81019
3812 PR rtl-optimization/81020
3813 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3814 are not mentioned in i3. Place the REG_UNUSED note on i2,
3815 possibly modified to REG_DEAD, if it did not originate in i3.
3817 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3819 * recog.c (store_data_bypass_p_1): New function.
3820 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3821 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3823 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3825 PR rtl-optimization/83361
3826 * ifcvt.c (if_convert): Call fixup_partitions.
3828 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3830 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3831 early folding of splat_u{8,16,32}.
3833 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3835 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3836 output_operand_lossage first argument with capital letter.
3837 (aarch64_override_options): Don't start error and sorry first argument
3838 with capital letter.
3840 2017-12-11 Andi Kleen <ak@linux.intel.com>
3842 PR gcov-profile/83355
3843 * auto-profile.c (string_table::get_index_by_decl): Don't
3844 recurse when abstract origin points to itself.
3846 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3848 PR tree-optimization/83320
3849 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3850 (prune_datarefs_not_in_loop): Ditto.
3852 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3854 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3856 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3858 PR tree-optimization/83337
3859 * gimple-loop-interchange.cc (compute_access_stride): Handle
3860 bitfield DRs properly.
3862 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3864 PR tree-optimization/83338
3865 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3866 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3867 vector of unsigned integers to vector of signed integers.
3869 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3871 PR rtl-optimization/83317
3872 * lra-constraints.c (process_address_1): Add insn code check.
3874 2017-12-08 Michael Matz <matz@suse.de>
3876 Fix PR tree-optimization/83323
3877 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3878 head-controlled loops and loop BBs.
3879 * common.opt (funroll-and-jam): Remove, instead ...
3880 (floop-unroll-and-jam): ... reuse this option.
3881 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3882 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3883 (-floop-unroll-and-jam): ... this option.
3885 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3887 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3888 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3889 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3891 2017-12-08 Richard Biener <rguenther@suse.de>
3894 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3895 handle accesses outside of zero-sized vars.
3897 2017-12-08 Martin Jambor <mjambor@suse.cz>
3899 PR tree-optimization/83141
3900 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3901 test for MEM_REFs implicitely changing types with padding. Remove
3903 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3905 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3907 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3909 * config/avr/avr.c (avr_attribute_table): Likewise.
3910 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3911 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3912 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3913 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3914 of struct attribute_spec.
3915 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3917 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3920 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3922 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3925 * profile-count.c (profile_count::from_gcov_type): Move from
3926 profile-count.h; handle overflow.
3927 * profile-count.h (profile_count::from_gcov_type): Move offline.
3929 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3931 PR rtl-optimization/83304
3932 * combine.c (move_deaths): If we do not know where a register died,
3935 2017-12-08 Richard Biener <rguenther@suse.de>
3937 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3938 Provide -fopt-info-loop feedback when we interchange in a nest.
3940 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3942 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3943 for armv6 ARM CPU IDs.
3945 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3947 * common/config/arm/arm-common.c: Include <algorithm>.
3948 (INCLUDE_VECTOR): Define.
3949 (compare_opt_names): New function.
3950 (arm_rewrite_selected_arch): Only strip out extensions that can be
3951 expressed through -mfpu. Sort the remaining extensions
3954 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3956 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3957 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3958 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3959 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3961 2017-06-08 Tristan Gingold <gindold@adacore.com>
3964 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3965 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3967 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3969 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3970 of struct attribute_spec.
3972 2017-12-08 Julia Koval <julia.koval@intel.com>
3974 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3975 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3976 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3977 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3978 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3979 _mm_maskz_dpwssds_epi32): Ditto.
3981 2017-12-08 Richard Biener <rguenther@suse.de>
3983 PR tree-optimization/81303
3984 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3985 conditions try to create a comparison vector type matching
3986 the data vector type.
3987 (vectorizable_condition): Adjust.
3988 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3989 Leave invariant conditions alone in case we can vectorize those.
3991 2017-12-08 Julia Koval <julia.koval@intel.com>
3993 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3994 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3995 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3996 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3997 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3999 2017-12-08 Julia Koval <julia.koval@intel.com>
4001 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4002 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4003 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4004 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4005 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4006 _mm_maskz_dpbusds_epi32): New intrinsics.
4008 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4010 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4012 (ld<bh>io_signed): Likewise.
4013 (st<bhw>io): Likewise.
4014 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4015 SMALL_INT12 constant integer operand.
4017 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4019 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4021 (nios2_large_constant_allowed): ...this. Adjust uses.
4022 (nios2_plus_symbolic_constant_p): Rename to...
4023 (nios2_plus_large_constant_p): ...this. Adjust uses.
4024 (nios2_legitimate_address_p): Correct CONST_INT handling.
4025 (nios2_symbolic_memory_operand_p): Rename to...
4026 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4027 (nios2_large_constant_p): Check for large constant integers too.
4028 (nios2_split_large_constant): Handle constant integers.
4029 (nios2_split_symbolic_memory_operand): Rename to...
4030 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4031 (nios2_legitimize_constant_address): Handle constant integers.
4032 (r0rel_constant_p): Handle small constant integers.
4033 (nios2_print_operand_address): Handle r0-relative integer addresses.
4034 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4035 * config/nios2/nios2.md: Adjust for renamed functions.
4037 2017-12-07 Andrew Waterman <andrew@sifive.com>
4039 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4040 (TARGET_HAVE_SRODATA_SECTION): New define.
4041 (riscv_select_section): New function.
4043 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4044 Alexander Monakov <amonakov@ispras.ru>
4045 Jakub Jelinek <jakub@redhat.com>
4048 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4050 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4052 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4053 a bool success value. Don't call output_operand_lossage here.
4054 (aarch64_print_ldpstp_address): Return a bool success value.
4055 (aarch64_print_operand_address): Call output_addr_const if
4056 aarch64_print_address_internal fails.
4057 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4058 'y'; call output_operand_lossage instead. Call output_operand_lossage
4059 if aarch64_print_ldpstp_address fails.
4061 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4063 * tree-vector-builder.h
4064 (tree_vector_builder::binary_encoded_nelts): Declare.
4065 * tree-vector-builder.c
4066 (tree_vector_builder::binary_encoded_nelts): New function.
4067 * fold-const.c (negate_expr_p): Likewise.
4068 (operand_equal_p, fold_checksum_tree): Likewise.
4069 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4070 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4071 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4072 (uniform_vector_p): Likewise.
4073 * varasm.c (const_hash_1, compare_constant): Likewise.
4074 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4075 (valid_lattice_transition): Operate directly on the VECTOR_CST
4077 * ipa-icf.c: Include tree-vector-builder.h.
4078 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4079 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4081 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4083 * tree.c (build_vector): Delete.
4084 * tree.h (build_vector): Make static and move into the self-testing
4087 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4089 * vector-builder.h (vector_builder::derived): New const overload.
4090 (vector_builder::elt): New function.
4091 * tree-vector-builder.h (tree_vector_builder::type): New function.
4092 (tree_vector_builder::apply_step): Declare.
4093 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4095 * gimple-fold.h (tree_vector_builder): Declare.
4096 (gimple_build_vector): Take a tree_vector_builder instead of a
4097 type and vector of elements.
4098 * gimple-fold.c (gimple_build_vector): Likewise.
4099 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4101 (get_initial_defs_for_reduction): Likewise.
4102 (vectorizable_induction): Likewise.
4104 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4106 * tree-vector-builder.h
4107 (tree_vector_builder::new_binary_operation): Declare.
4108 * tree-vector-builder.c
4109 (tree_vector_builder::new_binary_operation): New function.
4110 * fold-const.c (fold_relational_const): Use it.
4111 (const_binop): Likewise. Check that both input vectors have
4112 the same number of elements, thus excluding things like WIDEN_SUM.
4113 Check whether it is possible to operate directly on the encodings
4116 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4118 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4119 new_unary_operation, operating only on the encoded elements.
4120 (const_unop): Likewise.
4121 (exact_inverse): Likewise.
4122 (distributes_over_addition_p): New function.
4123 (const_binop): Use tree_vector_builder and new_unary_operation
4124 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4125 on the encoded elements unless the encoding is strided and the
4126 operation does not distribute over addition.
4127 (fold_convert_const): Use tree_vector_builder and
4128 new_unary_operation. Operate only on the encoded elements
4129 for truncating integer conversions, or for non-stepped encodings.
4131 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4133 * config/sparc/sparc.c: Include tree-vector-builder.h.
4134 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4135 * expmed.c: Include tree-vector-builder.h.
4136 (make_tree): Use tree_vector_builder instead of build_vector.
4137 * fold-const.c: Include tree-vector-builder.h.
4138 (const_binop): Use tree_vector_builder instead of build_vector.
4139 (const_unop): Likewise.
4140 (native_interpret_vector): Likewise.
4141 (fold_vec_perm): Likewise.
4142 (fold_ternary_loc): Likewise.
4143 * gimple-fold.c: Include tree-vector-builder.h.
4144 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4146 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4147 (simplify_vector_constructor): Use tree_vector_builder instead
4149 * tree-vect-generic.c: Include tree-vector-builder.h.
4150 (add_rshift): Use tree_vector_builder instead of build_vector.
4151 (expand_vector_divmod): Likewise.
4152 (optimize_vector_constructor): Likewise.
4153 * tree-vect-loop.c: Include tree-vector-builder.h.
4154 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4155 of build_vector. Explicitly use a stepped encoding for
4157 * tree-vect-slp.c: Include tree-vector-builder.h.
4158 (vect_get_constant_vectors): Use tree_vector_builder instead
4160 (vect_transform_slp_perm_load): Likewise.
4161 (vect_schedule_slp_instance): Likewise.
4162 * tree-vect-stmts.c: Include tree-vector-builder.h.
4163 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4164 (vect_gen_perm_mask_any): Likewise.
4165 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4166 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4168 (build_vector_from_val): Likewise. Explicitly use a duplicate
4171 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4173 * doc/generic.texi (VECTOR_CST): Describe new representation of
4175 * vector-builder.h: New file.
4176 * tree-vector-builder.h: Likewise.
4177 * tree-vector-builder.c: Likewise.
4178 * Makefile.in (OBJS): Add tree-vector-builder.o.
4179 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4180 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4181 (tree_vector): Change the number of elements to
4182 vector_cst_encoded_nelts.
4183 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4184 (VECTOR_CST_ELTS): Delete.
4185 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4186 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4187 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4188 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4189 (VECTOR_CST_ENCODED_ELT): Likewise.
4190 (vector_cst_encoded_nelts): New function.
4191 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4192 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4193 (vector_cst_int_elt, vector_cst_elt): Declare.
4194 * tree.c: Include tree-vector-builder.h.
4195 (tree_code_size): Abort if passed VECTOR_CST.
4196 (tree_size): Update for new VECTOR_CST layout.
4197 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4198 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4199 (build_vector): Use tree_vector_builder.
4200 (vector_cst_int_elt, vector_cst_elt): New functions.
4201 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4202 encoded elements and then create the vector in the canonical form.
4203 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4204 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4205 (tree_c_tests): Call test_vector_cst_patterns.
4206 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4208 (hash_tree): Likewise.
4209 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4210 (streamer_write_tree_header): Likewise.
4211 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4212 (streamer_alloc_tree): Likewise. Update call to make_vector.
4213 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4215 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4217 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4218 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4219 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4221 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4222 Richard Biener <rguenther@suse.de>
4224 PR tree-optimization/81303
4225 * Makefile.in (gimple-loop-interchange.o): New object file.
4226 * common.opt (floop-interchange): Reuse the option from graphite.
4227 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4228 -floop-interchange and mention it for -O3.
4229 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4230 * gimple-loop-interchange.cc: New file.
4231 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4232 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4233 * passes.def (pass_linterchange): New pass.
4234 * timevar.def (TV_LINTERCHANGE): New time var.
4235 * tree-pass.h (make_pass_linterchange): New declaration.
4236 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4237 interchange. Record IV before/after increment in new parameters.
4238 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4239 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4241 (check_reduction_path): ...New function here.
4242 * tree-vectorizer.h (check_reduction_path): New declaration.
4244 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4247 PR rtl-optimization/80818
4248 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4249 always early clobbered.
4250 * lra-lives.c (process_bb_lives): Check input hard regs for early
4251 clobbered non-operand hard reg.
4253 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4256 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4257 types_compatible_p, just that TYPE_MODE is the same.
4259 2017-12-07 Martin Sebor <msebor@redhat.com>
4262 * attribs.c (empty_attribute_table): Initialize new member of
4263 struct attribute_spec.
4264 (decl_attributes): Add argument. Handle mutually exclusive
4265 combinations of attributes.
4266 (selftests::test_attribute_exclusions): New function.
4267 (selftests::attribute_c_tests): Ditto.
4268 * attribs.h (decl_attributes): Add default argument.
4269 * selftest.h (attribute_c_tests): Declare.
4270 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4271 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4273 * doc/extend.texi (Common Function Attributes): Update const and pure.
4274 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4275 of struct attribute_spec.
4276 * config/arc/arc.c (arc_attribute_table): Same.
4277 * config/arm/arm.c (arm_attribute_table): Same.
4278 * config/avr/avr.c ( avr_attribute_table): Same.
4279 * config/bfin/bfin.c (bfin_attribute_table): Same.
4280 * config/cr16/cr16.c (cr16_attribute_table): Same.
4281 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4282 * config/h8300/h8300.c (h8300_attribute_table): Same.
4283 * config/i386/i386.c (ix86_attribute_table): Same.
4284 * config/ia64/ia64.c (ia64_attribute_table): Same.
4285 * config/m32c/m32c.c (m32c_attribute_table): Same.
4286 * config/m32r/m32r.c (m32r_attribute_table): Same.
4287 * config/m68k/m68k.c (m68k_attribute_table): Same.
4288 * config/mcore/mcore.c (mcore_attribute_table): Same.
4289 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4290 * config/mips/mips.c (mips_attribute_table): Same.
4291 * config/msp430/msp430.c (msp430_attribute_table): Same.
4292 * config/nds32/nds32.c (nds32_attribute_table): Same.
4293 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4294 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4295 * config/rl78/rl78.c (rl78__attribute_table): Same.
4296 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4297 * onfig/rx/rx.c (rx_attribute_table): Same.
4298 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4299 * config/sh/sh.c (sh_attribute_table): Same.
4300 * config/sparc/sparc.c (sparc_attribute_table): Same.
4301 * config/spu/spu.c (spu_attribute_table): Same.
4302 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4303 * config/v850/v850.c (v850_attribute_table): Same.
4304 * config/visium/visium.c (visium_attribute_table): Same.
4306 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4309 * config/arm/arm.c (INCLUDE_STRING): Define.
4310 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4311 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4314 2017-12-07 Michael Matz <matz@suse.de>
4316 Add unroll and jam pass
4318 * gimple-loop-jam.c: New file.
4319 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4320 * common.opt (funroll-and-jam): New option.
4321 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4322 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4323 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4324 * passes.def: Add pass_loop_jam.
4325 * timevar.def (TV_LOOP_JAM): Add.
4326 * tree-pass.h (make_pass_loop_jam): Declare.
4327 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4328 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4329 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4330 to flow_loop_tree_node_add.
4331 (duplicate_subloops, copy_loops_to): Append to sibling list.
4332 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4333 * doc/invoke.texi (-funroll-and-jam): Document new option.
4334 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4336 2017-12-07 Richard Biener <rguenther@suse.de>
4338 PR tree-optimization/83296
4339 PR tree-optimization/67769
4340 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4341 flow sensitive info in an unrelated BB.
4342 (value_replacement): Use reset_flow_sensitive_info.
4343 (minmax_replacement): Reset flow sensitive info on the def
4344 we move. Do not reset flow sensitive info in the whole BB
4345 we move the stmt to.
4346 (abs_replacement): Likewise.
4348 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4351 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4352 rs6000_cpu to the given -mcpu=, or to the default processor.
4354 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4356 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4357 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4358 instead of rs6000_cpu_attr.
4359 (rs6000_adjust_cost): Ditto.
4360 (is_microcoded_insn): Ditto.
4361 (rs6000_adjust_priority): Ditto.
4362 (rs6000_issue_rate): Ditto.
4363 (rs6000_use_sched_lookahead): Ditto.
4364 (rs6000_use_sched_lookahead_guard): Ditto.
4365 (rs6000_sched_reorder): Ditto.
4366 (force_new_group): Ditto.
4367 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4368 (group_ending_nop): Ditto.
4370 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4372 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4373 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4374 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4375 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4377 (rs6000_reassociation_width): Ditto.
4378 (rs6000_emit_epilogue): Ditto.
4379 (rs6000_adjust_cost): Ditto.
4380 (is_microcoded_insn): Ditto.
4381 (is_cracked_insn): Ditto.
4382 (rs6000_adjust_priority): Ditto.
4383 (rs6000_sched_reorder): Ditto.
4384 (rs6000_sched_reorder2): Ditto.
4385 (insn_must_be_first_in_group): Ditto.
4386 (insn_must_be_last_in_group): Ditto.
4387 (rs6000_register_move_cost): Ditto.
4388 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4391 2017-12-07 Julia Koval <julia.koval@intel.com>
4393 * config.gcc: Add vaesintrin.h.
4394 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4396 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4397 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4399 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4400 * config/i386/immintrin.h: Include vaesintrin.h.
4401 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4402 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4403 _mm_aesdec_epi128): New intrinsics.
4405 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4407 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4408 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4409 c-family/c-spellcheck.cc.
4410 (best_macro_match::best_macro_match): Likewise.
4412 (struct edit_distance_traits<cpp_hashnode *>): Move to
4413 c-family/c-spellcheck.h.
4414 (class best_macro_match): Likewise.
4416 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4418 PR tree-optimization/83293
4419 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4420 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4421 might insert into empty bb.
4424 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4425 simplify for plus with :c added, and pointer_plus without that.
4426 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4427 with undefined overflow and the conversion is not widening,
4428 perform negation in utype and only convert to type afterwards.
4429 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4430 simplify for plus with :c added, and pointer_plus without that.
4431 If type is integral with undefined overflow and the conversion is
4432 not widening, perform minus in utype and only convert to type
4433 afterwards. Move the last pointer_diff_expr simplify into the
4436 2017-12-06 Martin Sebor <msebor@redhat.com>
4438 PR tree-optimization/82646
4439 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4440 strncpy, not maxlen.
4442 2017-12-06 Martin Sebor <msebor@redhat.com>
4444 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4447 PR tree-optimization/83075
4448 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4449 strncat/strncpy don't change length of source string.
4451 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4454 2017-11-29 Martin Aberg <maberg@gaisler.com>
4456 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4457 to prevent b2bst errata sequence.
4458 (sqrtdf2_fix): Likewise.
4460 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4462 PR tree-optimization/81945
4463 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4464 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4465 to dest_cfun has orig_loop_num set, either remap it to the new
4466 loop number if the loop got moved too, or clear it.
4468 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4470 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4472 (thunderx2t99_nothing): New insn reservation.
4473 (thunderx2t99_mrs): New insn reservation.
4474 (thunderx2t99_multiple): New insn reservation.
4475 (thunderx2t99_alu_basi): Add bfx to reservation.
4476 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4478 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4481 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4483 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4485 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4486 tree-ssa-pre.c::remove_dead_inserted_code.
4487 * tree-ssa-dce.h: New file.
4488 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4489 (remove_dead_inserted_code): Move and rename to function
4490 tree-ssa-dce.c::simple_dce_from_worklist.
4491 (pass_pre::execute): Update use.
4493 2017-12-05 Richard Biener <rguenther@suse.de>
4495 PR tree-optimization/83277
4496 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4497 to code-gen liveout vars.
4499 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4501 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4502 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4503 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4506 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4508 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4510 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4511 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4514 2017-12-05 Richard Biener <rguenther@suse.de>
4516 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4517 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4518 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4519 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4521 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4523 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4525 2017-12-05 Martin Liska <mliska@suse.cz>
4526 Jakub Jelinek <jakub@redhat.com>
4528 * doc/invoke.texi: Document the options.
4529 * flag-types.h (enum sanitize_code): Add
4530 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4531 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4532 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4533 * opts.c: Define new sanitizer options.
4534 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4535 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4537 2017-12-05 Julia Koval <julia.koval@intel.com>
4539 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4540 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4541 (ix86_handle_option): Handle -mavx512vnni.
4542 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4543 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4544 * config/i386/i386-c (__AVX512VNNI__): New.
4545 * config/i386/i386.c (ix86_target_string): Handle new option.
4546 (ix86_valid_target_attribute_inner_p): Handle new option.
4547 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4548 * config/i386/i386.opt (mavx512vnni): New option.
4550 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4553 * athlon.md: Disable for generic.
4554 * haswell.md: Enable for generic.
4555 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4556 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4558 (ix86_adjust_cost): Move generic to haswell path.
4560 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4562 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4563 instead of MEM_P in a couple more places. Fix formatting issues.
4565 2017-12-04 Jim Wilson <jimw@sifive.com>
4567 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4568 instead of GP_REG_LAST-1.
4569 (riscv_adjust_libcall_cfi_prologue): Likewise.
4570 (riscv_adjust_libcall_cri_epilogue): Likewise.
4571 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4574 2017-12-04 Luis Machado <luis.machado@linaro.org>
4576 * ipa-pure-const.c (check_decl): Add missing newline.
4577 (state_from_flags): Likewise.
4579 2017-12-04 Jeff Law <law@redhat.com>
4581 PR tree-optimizatin/78496
4582 * gimple-ssa-evrp-analyze.h
4583 (evrp_range_analyzer::get_vr_values): Simplify.
4584 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4585 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4586 and gimple-ssa-evrp-analyze.h.
4587 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4588 (simplify_stmt_for_jump_threading): Copy a blob of code from
4589 tree-vrp.c to use ranges to simplify statements.
4590 (dom_opt_dom_walker::before_dom_children): Call
4591 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4592 (dom_opt_dom_walker::after_dom_children): Similarly for
4593 evrp_range_analyzer::leave.
4594 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4597 * gimple-ssa-evrp-analyze.c
4598 (evrp_range_analyzer::extract_range_from_stmt): Always use
4599 vr_values::update_value_range so preexisting range info is
4600 medged with new range info, even if the new range is VR_VARYING.
4602 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4604 * combine.c: Adjust comment.
4605 (use_crosses_set_p): Delete.
4606 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4607 (try_combine): Ditto.
4609 2017-12-04 Richard Biener <rguenther@suse.de>
4611 PR tree-optimization/83255
4612 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4613 Re-add zero-iteration check.
4615 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4617 PR rtl-optimization/83245
4618 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4619 hard registers as earlyclobber, also if not in an asm.
4621 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4625 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4628 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4629 rs6000_cpu based on cpu_index, not tune_index.
4631 2017-12-04 Richard Biener <rguenther@suse.de>
4633 PR tree-optimization/83238
4634 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4635 code match comment, rejecting invalid SESE regions.
4637 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4639 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4640 require base operand is a REG_POINTER prior to reload on targets
4641 with non-equivalent space registers.
4643 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4645 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4646 (update_specialized_profile): Fix updating of counts.
4647 (perhaps_add_new_callers): Likewise.
4649 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4652 * x86-tune.def: Remove obsolette FIXMEs.
4653 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4654 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4655 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4657 (X86_TUNE_PAD_RETURNS): Disable for generic.
4659 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4661 PR tree-optimization/83170
4662 PR tree-optimization/83241
4663 * gimple-ssa-store-merging.c
4664 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4665 gimple_vuse (ins_stmt) in case it has changed.
4666 (imm_store_chain_info::output_merged_store): Likewise.
4668 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4672 * tree-cfg.c (pass_warn_function_return::execute): Handle
4673 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4674 with BUILTINS_LOCATION.
4678 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4679 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4680 instead of DECL_MODE.
4682 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4683 last argument to rtx pointer.
4684 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4685 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4686 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4687 depending on the chosen ISAs.
4688 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4689 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4690 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4692 * config/i386/sse.md (mov<mode>_internal): Likewise.
4693 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4695 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4697 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4699 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4701 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4702 parameters from prototype.
4703 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4704 parameters. Don't print a comment.
4705 (emit_fusion_gpr_load): Adjust.
4706 (emit_fusion_load_store): Adjust.
4707 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4708 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4709 comment on the second line.
4711 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4714 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4715 rs6000_cpu based on cpu_index, not tune_index.
4717 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4719 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
4720 which_alternative instead of which_alternative + 1.
4721 (output_asm_insn): Print an extra tab if the template is short.
4723 2017-12-01 Jim Wilson <jimw@sifive.com>
4725 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
4727 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
4728 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
4730 * doc/tm.texi: Regenerate.
4732 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4735 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
4736 whether we can allocate pseudos before trying to fix an address.
4737 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
4738 memory address is indexed or indirect.
4739 (floatuns_<mode>si2_hw2): Likewise.
4741 2017-12-01 Jason Merrill <jason@redhat.com>
4743 * Makefile.in (TAGS): Add c-family/*.cc.
4745 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4747 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4748 (call_value_insn): Likewise.
4749 (sibcall_insn): Likewise.
4750 (sibcall_value_insn): Likewise.
4751 (movsi_aarch64): Likewise.
4752 (movdi_aarch64): Likewise.
4753 (add_losym_): Likewise.
4754 (ldr_got_small_): Likewise.
4755 (ldr_got_small_sidi): Likewise.
4756 (ldr_got_small_28k_): Likewise.
4757 (ldr_got_small_28k_sidi): Likewise.
4758 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4759 Move output_addr_const to symbolic case. Add error check.
4761 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4763 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4764 * config/s390/s390.md: Disable MVC merging peephole if it would
4765 disable operand forwarding.
4766 (new peephole2): Split MVCs if it would turn them into up to 2
4769 2017-12-01 Richard Biener <rguenther@suse.de>
4771 PR tree-optimization/83232
4772 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
4773 detection of same access. Instead of breaking the group here
4774 do not consider the duplicate. Add comment explaining real fix.
4776 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4778 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
4780 2017-12-01 Sudakshina Das <sudi.das@arm.com>
4782 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
4784 2017-12-01 Jakub Jelinek <jakub@redhat.com>
4786 * function.h (struct function): Remove cilk_frame_decl,
4787 is_cilk_function and calls_cilk_spawn fields.
4788 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
4790 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
4791 cilk_elemental field.
4792 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
4793 * target.def: Adjust comment.
4794 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
4795 Don't test cilk_elemental.
4797 PR tree-optimization/83233
4798 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4799 bswap_stat name for the struct.
4802 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4803 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4804 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4805 SWITCH_BREAK_LABEL_P set on the label.
4806 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4807 added for default case if it was missing and not all cases covered.
4808 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4809 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4811 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4812 to avoid -Wimplicit-fallthrough warning.
4813 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4814 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4816 PR tree-optimization/83221
4817 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4819 (init_reassoc): Formatting fix.
4822 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4823 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4827 * tree-cfg.c: Include asan.h.
4828 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4830 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4832 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4834 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4836 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4837 vpcmpeqd instruction.
4839 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4841 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4842 registers type for 128bit mode.
4844 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4846 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4848 * gengtype-state.c (read_state_pair): Likewise.
4849 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4850 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4851 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4852 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4853 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4854 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4855 * builtins.c (fold_builtin_3): Likewise.
4856 * graphite-scop-detection.c
4857 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4858 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4861 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4864 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4865 parameter. Return true if ignore_decl_rtl_set_p is true and other
4866 conditions are satisfied.
4867 * asan.h (asan_protect_global): Add new parameter.
4868 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4869 to asan_protect_global calls.
4871 2017-11-30 Jim Wilson <jimw@sifive.com>
4873 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4874 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4875 -mstrict-align, add info on default value. Delete redundant lines for
4876 -mabi. Add missing -mexplicit-relocs docs.
4878 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4880 * config/arc/arc.md (trap): New pattern.
4882 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4884 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4885 instruction to end into a delay slot.
4886 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4887 can be placed into a delay slot against reg_note.
4889 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4891 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4892 labels number of usages.
4894 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4896 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4898 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4900 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4903 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4904 multiplication by power of 2 constant into two shifts + comparison.
4906 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4909 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4911 2017-11-30 Richard Biener <rguenther@suse.de>
4913 PR tree-optimization/83202
4914 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4915 allow_peel argument and guard peeling.
4916 (canonicalize_loop_induction_variables): Likewise.
4917 (canonicalize_induction_variables): Pass false.
4918 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4919 peeling from cunrolli.
4921 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4923 * combine.c (try_combine): Print a message to dump file whenever
4924 I0, I1, or I2 cannot be combined into I3.
4926 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4928 PR rtl-optimization/83156
4929 PR rtl-optimization/82621
4930 * combine.c (try_combine): Don't split an I2 if one of the dests is
4931 set again before I3. Allow unused dests.
4933 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4935 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4937 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4939 PR rtl-optimization/80818
4940 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4941 recursively. Use insn code for clobber.
4942 (lra_set_insn_recog_data): Pass the new arg to
4943 collect_non_operand_hard_regs.
4944 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4946 (lra_update_insn_regno_info): Pass insn to
4947 add_regs_to_insn_regno_info.
4949 2017-11-29 Jim Wilson <jimw@sifive.com>
4950 Andrew Waterman <andrew@sifive.com>
4952 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4953 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4954 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4955 use SINGLE_SHIFT_COST.
4956 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4957 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4959 2017-11-29 Julia Koval <julia.koval@intel.com>
4961 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4962 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4963 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4964 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4965 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4966 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4967 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4968 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4969 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4970 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4971 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4972 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4973 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4974 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4975 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4976 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4977 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4978 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4979 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4980 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4981 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4982 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4983 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4984 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4985 __builtin_ia32_vpshldv_v2di_mask,
4986 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4987 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4988 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4990 2017-11-29 Julia Koval <julia.koval@intel.com>
4992 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4993 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4994 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4995 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4996 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4997 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4998 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4999 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5000 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5001 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5002 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5003 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5004 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5005 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5006 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5007 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5008 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5009 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5010 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5011 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5012 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5014 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5016 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5017 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5020 2017-11-29 Martin Aberg <maberg@gaisler.com>
5022 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5023 to prevent b2bst errata sequence.
5024 (sqrtdf2_fix): Likewise.
5026 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5028 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5029 (div_sqrt_insn_p): New function.
5030 (sparc_do_work_around_errata): Insert NOP instructions to
5031 prevent sequences that could trigger the TN-0013 errata for
5032 certain LEON3 processors.
5033 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5034 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5035 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5036 (in_branch_delay): Prevent div and sqrt in delay slot if
5038 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5040 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5042 * config/sparc/sparc.c (atomic_insn_p): New function.
5043 (sparc_do_work_around_errata): Insert NOP instructions to
5044 prevent sequences that could trigger the TN-0010 errata for
5046 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5047 instruction referable in atomic_insns_p.
5049 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5051 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5052 (atomic_compare_and_swap_leon3_1): Likewise.
5055 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5057 * config/sparc/sparc.c (fpop_insn_p): New function.
5058 (sparc_do_work_around_errata): Insert NOP instructions to
5059 prevent sequences that could trigger the TN-0012 errata for
5061 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5062 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5063 (in_branch_annul_delay): Prevent floating-point instructions
5064 in delay slot of annulled integer branch.
5066 2017-11-29 Richard Biener <rguenther@suse.de>
5068 PR tree-optimization/83202
5069 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5071 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5072 nodes vectorized to the same stmts multiple times.
5073 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5074 (vect_analyze_slp_instance): Adjust.
5075 (scalar_stmts_to_slp_tree_map_t): New typedef.
5076 (vect_schedule_slp_instance): Add a map recording the SLP node
5077 representing the vectorized stmts for a set of scalar stmts.
5078 Avoid code-generating redundancies.
5079 (vect_schedule_slp): Allocate map and pass it down.
5081 2017-11-29 Nathan Sidwell <nathan@acm.org>
5084 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5087 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5089 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5090 cases for printing LDP/STP memory addresses.
5091 (aarch64_print_address_internal): Renamed from
5092 aarch64_print_operand_address, added parameter, add Pmode check.
5093 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5094 (aarch64_print_operand_address): Indirect to
5095 aarch64_print_address_internal.
5096 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5097 'y' operand output specifier.
5099 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5102 * tree.c (build_simple_mem_ref_loc): Handle
5103 get_addr_base_and_unit_offset returning a MEM_REF.
5106 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5109 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5112 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5114 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5116 (layout::move_to_column): Likewise.
5118 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5121 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5122 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5124 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5125 Martin Jambor <mjambor@suse.cz>
5128 * tree.h (expr_type_first_operand_type_p): Declare
5129 * tree.c (expr_type_first_operand_type_p): New function.
5130 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5131 (ipa_value_from_jfunc): Adjust declaration.
5132 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5133 Use it as result type for arithmetics, unless it is NULL in which case
5134 be more conservative.
5135 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5136 ipa_get_jf_pass_through_result.
5137 (propagate_vals_across_pass_through): Likewise.
5138 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5139 is to propagate_vals_across_pass_through.
5140 (propagate_constants_across_call): Pass PARM_TYPE to
5141 propagate_scalar_across_jump_function.
5142 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5143 ipa_value_from_jfunc.
5144 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5145 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5146 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5147 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5148 target_type, pass it to ipa_value_from_jfunc.
5149 (update_indirect_edges_after_inlining): Pass parameter type to
5150 try_make_edge_direct_simple_call.
5152 2017-11-28 Jeff Law <law@redhat.com>
5154 * gimple-ssa-evrp-analyze.c
5155 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5156 refine ranges if scev_initialized_p returns true.
5157 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5159 2017-11-28 Julia Koval <julia.koval@intel.com>
5161 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5162 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5163 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5164 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5165 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5166 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5167 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5168 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5169 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5170 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5171 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5172 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5173 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5174 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5175 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5176 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5177 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5178 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5179 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5180 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5181 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5182 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5184 2017-11-28 Julia Koval <julia.koval@intel.com>
5186 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5187 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5188 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5189 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5190 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5191 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5192 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5193 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5194 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5195 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5196 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5197 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5198 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5199 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5200 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5201 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5202 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5203 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5204 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5205 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5206 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5207 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5208 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5209 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5210 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5211 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5212 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5213 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5214 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5215 __builtin_ia32_vpshld_v2di_mask): New builtins.
5216 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5217 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5219 2017-11-28 Richard Biener <rguenther@suse.de>
5221 PR tree-optimization/80776
5222 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5224 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5226 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5227 If the incoming edge is an effective fallthru because the other
5228 edge only reaches a __builtin_unreachable () then record ranges
5229 derived from the controlling condition in SSA info.
5230 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5231 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5233 2017-11-28 Olivier Hainque <hainque@adacore.com>
5235 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5236 on mingw build hosts.
5238 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5241 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5243 2017-11-28 Richard Biener <rguenther@suse.de>
5246 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5247 copies generated from memcpy use a character array as reference
5250 2017-11-28 Julia Koval <julia.koval@intel.com>
5251 Sebastian Peryt <sebastian.peryt@intel.com>
5253 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5254 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5255 cilk-common.o, cilk.h, cilk-common.c): Remove.
5257 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5258 * builtins.c (is_builtin_name): Remove cilkplus condition.
5259 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5260 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5261 cilk-builtins.def, cilkplus.def): Remove.
5262 * cif-code.def (CILK_SPAWN): Remove.
5263 * cilk-builtins.def: Delete.
5264 * cilk-common.c: Ditto.
5266 * cilkplus.def: Ditto.
5267 * config/darwin.h (fcilkplus): Delete.
5268 * cppbuiltin.c: Ditto.
5269 * doc/extend.texi: Remove cilkplus doc.
5270 * doc/generic.texi: Ditto.
5271 * doc/invoke.texi: Ditto.
5272 * doc/passes.texi: Ditto.
5273 * gcc.c (fcilkplus): Remove.
5274 * gengtype.c (cilk.h): Remove.
5275 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5277 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5279 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5281 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5282 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5283 cilkplus conditions.
5284 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5285 inline_read_section): Ditto.
5286 * ipa-inline-analysis.c (cilk.h): Remove.
5287 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5288 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5289 append_linker_options): Remove condition for fcilkplus.
5290 * lto/lto-lang.c (cilk.h): Remove.
5291 (lto_init): Remove condition for fcilkplus.
5292 * omp-expand.c (expand_cilk_for_call): Delete.
5293 (expand_omp_taskreg, expand_omp_for_static_chunk,
5294 expand_omp_for): Remove cilkplus
5296 (expand_cilk_for): Delete.
5297 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5298 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5299 execute_lower_omp, diagnose_sb_0): Ditto.
5300 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5301 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5302 * tree-nested.c: Ditto.
5303 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5304 (dump_generic_node): Ditto.
5305 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5306 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5308 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5310 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5312 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5313 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5314 for vector mode and !TARGET_SIMD.
5316 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5318 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5320 * tree.h (SWITCH_LABELS): Remove.
5321 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5322 assert SWITCH_BODY is non-NULL.
5323 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5325 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5327 PR tree-optimization/80788
5328 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5329 has TREE_OVERFLOW set, call drop_tree_overflow.
5331 2017-11-28 Richard Biener <rguenther@suse.de>
5333 PR tree-optimization/83158
5334 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5336 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5339 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5340 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5342 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5344 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5345 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5347 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5350 * varasm.c (compare_constant): Take the mode of the constants into
5351 account when comparing floating point constants.
5353 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5355 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5356 from explicit instantiation of debug_helper.
5357 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5359 2017-11-27 Richard Biener <rguenther@suse.de>
5361 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5364 2017-11-27 Richard Biener <rguenther@suse.de>
5366 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5367 (type_hash_canon): Also clear int_cst_hash_table entry for
5369 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5371 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5373 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5374 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5375 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5377 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5379 * hash-map.h (gt_cleare_cache): Avoid UB.
5381 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5383 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5384 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5385 (decide_unroll_runtime_iterations): Likewise.
5386 (decide_unroll_stupid): Likewise.
5388 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5391 * config/i386/i386.md: Add a loop with incssp.
5393 2017-11-27 Martin Jambor <mjambor@suse.cz>
5395 PR tree-optimization/81248
5396 * tree-sra.c (splice_param_accesses): Remove size check.
5397 (decide_one_param_reduction): Fix size check.
5398 * gimple-pretty-print.c (dump_profile): Silence warning.
5399 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5401 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5404 * dbxout.c (lastlineno): New variable.
5405 (dbx_debug_hooks): Use dbxout_switch_text_section as
5406 switch_text_section debug hook.
5407 (dbxout_function_end): Switch to current_function_section
5408 rather than function_section. If crtl->has_bb_partition,
5409 output just one N_FUN, depending on in_cold_section_p.
5410 (dbxout_source_line): Remember last lineno in lastlineno.
5411 (dbxout_switch_text_section): New function.
5412 (dbxout_function_decl): Adjust dbxout_block caller.
5413 (dbx_block_with_cold_children): New function.
5414 (dbxout_block): Return true if any LBRAC/RBRAC have been
5415 emitted. Use dbx_block_with_cold_children at depth == 0
5416 in second partition. Add PARENT_BLOCKNUM argument, pass
5417 it optionally adjusted to children. Output LBRAC/RBRAC
5418 around recursive call only if the block is in the current
5419 partition, if not and anything was output, emit empty
5421 * final.c (final_scan_insn): Compute cold_function_name
5422 before calling switch_text_section debug hook. Call
5423 that hook even if dwarf2out_do_frame if not emitting
5427 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5428 TREE_READONLY decls.
5430 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5432 PR rtl-optimization/82488
5433 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5435 2017-11-26 Julia Koval <julia.koval@intel.com>
5437 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5439 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5442 2017-11-26 Julia Koval <julia.koval@intel.com>
5444 * config/i386/driver-i386.c (host_detect_local_cpu):
5445 Detect skylake-avx512.
5447 2017-11-26 Julia Koval <julia.koval@intel.com>
5449 * config.gcc: Add -march=cannonlake.
5450 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5451 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5452 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5453 (PTA_CANNONLAKE): New.
5454 (processor_target_table): Add cannonlake.
5455 (ix86_option_override_internal): Ditto.
5456 (fold_builtin_cpu): Ditto.
5457 (get_builtin_code_for_version): Handle cannonlake.
5458 (M_INTEL_COREI7_CANNONLAKE): New.
5459 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5460 * doc/invoke.texi: Add -march=cannonlake.
5462 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5464 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5465 (try_init_one_plugin): Alternative implementation for MinGW.
5466 * Makefile.in (plugin_implib): New.
5467 (gengtype-lex.c): Fix broken AIX workaround.
5468 * configure: Regenerate.
5469 * doc/plugins.texi: Document support for MinGW.
5471 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5473 PR rtl-optimization/81553
5474 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5475 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5476 is a shift where C1 has different mode than the whole shift, use C1's
5477 mode for MULT rather than the shift's mode.
5480 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5481 builtins not enabled in the currently selected ISA.
5483 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5485 PR tree-optimization/71026
5486 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5487 (insert_reciprocals): Change to insert reciprocals before a division
5488 by a square and to insert the square of a reciprocal.
5489 (execute_cse_reciprocals_1): Change to consider division by a square.
5490 (register_division_in): Add importance parameter.
5492 2017-11-24 Richard Biener <rguenther@suse.de>
5494 PR tree-optimization/82402
5495 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5496 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5498 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5500 * match.pd (0-ptr): New transformation.
5502 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5505 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5506 when skipping checking.
5508 2017-11-24 Richard Biener <rguenther@suse.de>
5510 PR tree-optimization/83128
5511 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5512 (vn_reference_lookup_3): Likewise.
5514 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5517 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5518 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5521 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5522 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5523 check for the latter with gimple_call_builtin_p. Do not handle
5524 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5526 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5528 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5531 2017-11-23 Julia Koval <julia.koval@intel.com>
5533 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5534 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5535 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5536 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5537 _mm512_maskz_expandloadu_epi16): New intrinsics.
5538 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5539 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5540 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5541 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5542 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5543 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5544 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5545 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5546 _mm256_maskz_expandloadu_epi8): New intrinsics.
5547 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5548 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5549 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5550 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5551 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5552 * config/i386/sse.md (VI248_VLBW): New iterator.
5553 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5555 2017-11-23 Julia Koval <julia.koval@intel.com>
5557 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5558 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5559 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5560 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5561 _mm512_mask_compressstoreu_epi16): New.
5562 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5563 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5564 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5565 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5566 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5567 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5568 _mm256_mask_compressstoreu_epi8): New.
5569 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5570 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5571 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5572 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5573 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5574 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5575 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5576 __builtin_ia32_compresshi128_mask,
5577 __builtin_ia32_compressstoreuqi512_mask,
5578 __builtin_ia32_compressstoreuhi512_mask,
5579 __builtin_ia32_compressstoreuqi256_mask,
5580 __builtin_ia32_compressstoreuqi128_mask,
5581 __builtin_ia32_compressstoreuhi256_mask,
5582 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5583 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5585 (ix86_expand_special_args_builtin): Handle new types.
5586 (s4fma_expand): Handle new builtin array.
5587 * config/i386/immintrin.h: Include new headers.
5588 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5589 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5591 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5593 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5594 fixes. Declare temp and g variables at the top in order to avoid
5595 {} in most of the cases.
5597 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5599 * match.pd (ptr-0): New transformation.
5601 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5603 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5604 (struct addr_mode_cost_table): New.
5605 (struct tune_params): Add field addr_mode_costs.
5606 * config/arm/arm.c (generic_addr_mode_costs): New.
5607 (arm_slowmul_tune): Initialise addr_mode_costs field.
5608 (arm_fastmul_tune): Likewise.
5609 (arm_strongarm_tune): Likewise.
5610 (arm_xscale_tune): Likewise.
5611 (arm_9e_tune): Likewise.
5612 (arm_marvell_pj4_tune): Likewise.
5613 (arm_v6t2_tune): Likewise.
5614 (arm_cortex_tune): Likewise.
5615 (arm_cortex_a8_tune): Likewise.
5616 (arm_cortex_a7_tune): Likewise.
5617 (arm_cortex_a15_tune): Likewise.
5618 (arm_cortex_a35_tune): Likewise.
5619 (arm_cortex_a53_tune): Likewise.
5620 (arm_cortex_a57_tune): Likewise.
5621 (arm_exynosm1_tune): Likewise.
5622 (arm_xgene1_tune): Likewise.
5623 (arm_cortex_a5_tune): Likewise.
5624 (arm_cortex_a9_tune): Likewise.
5625 (arm_cortex_a12_tune): Likewise.
5626 (arm_cortex_a73_tune): Likewise.
5627 (arm_v7m_tune): Likewise.
5628 (arm_cortex_m7_tune): Likewise.
5629 (arm_v6m_tune): Likewise.
5630 (arm_fa726te_tune): Likewise.
5631 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5633 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5635 * config/arm/arm.c (arm_mem_costs): New function.
5636 (arm_rtx_costs_internal): Use arm_mem_costs.
5638 2017-11-23 Mark Wielaard <mark@klomp.org>
5640 * dwarf2out.c (init_sections_and_labels): Use generation to create
5641 unique ranges_section_label and ranges_base_label. Return generation.
5642 (output_rnglists): Add generation argument. Use generation to create
5643 unique ranges labels.
5644 (dwarf2out_finish): Get generation from init_sections_and_labels
5645 and pass generation to output_rnglists.
5647 2017-11-23 Mike Stump <mikestump@comcast.net>
5648 Eric Botcazou <ebotcazou@adacore.com>
5650 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5651 * cfgloop.h (struct loop): Add unroll field.
5652 * function.h (struct function): Add has_unroll bitfield.
5653 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5654 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5655 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5656 (pass_rtl_unroll_loops::gate): Likewise.
5657 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5658 for which loop->unroll==1.
5659 (decide_unroll_constant_iterations): Use note for consistency and
5660 take loop->unroll into account. Return early if loop->unroll is set.
5661 Fix thinko in existing test.
5662 (decide_unroll_runtime_iterations): Use note for consistency and
5663 take loop->unroll into account.
5664 (decide_unroll_stupid): Likewise.
5665 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5666 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5667 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5669 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5670 (print_loop): Print loop->unroll if set.
5671 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5672 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5673 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5675 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5676 loop->unroll is set and smaller than the trip count. Otherwise bypass
5677 entirely the heuristics if loop->unroll is set. Remove dead note.
5678 Fix off-by-one bug in other note.
5679 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5680 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5682 (tree_unroll_loops_completely): Make static.
5683 (pass_complete_unroll::execute): Use correct type for variable.
5684 (pass_complete_unrolli::execute): Fix formatting.
5685 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5687 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5689 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5690 enable when TARGET_PREFER_AVX128 is set.
5692 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5694 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5696 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5697 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5699 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5701 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5702 frequencies. Fix estimation of aggregate parameters.
5704 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5706 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5709 2017-11-23 Tom de Vries <tom@codesourcery.com>
5711 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
5713 2017-11-23 Tom de Vries <tom@codesourcery.com>
5715 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
5718 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
5721 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
5722 sibcall_value_pcrel_fdpic): Use local variable instead of
5724 (calli_tbr_rel): Add missing operand 2.
5725 (call_valuei_tbr_rel): Add missing operand 3.
5727 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5730 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
5731 bitpos/bitsize covering the whole destination, use store_expr only if
5732 the complex mode is the same. Otherwise, use expand_normal and if
5733 it returns CONCAT, subreg each part separately instead of trying to
5734 subreg the whole result.
5736 2017-11-23 Richard Biener <rguenther@suse.de>
5738 PR tree-optimization/23094
5739 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
5740 come here from walking over backedges in the first iteration.
5741 (vn_reference_lookup_3): Skip clobbers that store the same value.
5743 2017-11-23 Richard Biener <rguenther@suse.de>
5745 PR tree-optimization/81403
5746 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5747 a block we need a leader relative to.
5748 (phi_translate_1): For nary processing require a leader from
5749 get_representative_for given we run expression simplification
5750 using match-and-simplify. Remove previous fix.
5752 2017-11-22 Jeff Law <law@redhat.com>
5754 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5755 Use new method allocate_value_range rather than accessing the
5756 vrp_value_range_pool data member directly.
5757 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5758 to use extract_range_from_stmt method to avoid need for
5759 extract_range_from_assignment method.
5760 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5761 method rather than setting values_propgated data member directly.
5762 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5763 and values propagated data members and extract_range_from_assignment
5764 method. Reorder private data members to conform to standards.
5765 Add new methods set_lattice_propagation_complete and
5766 allocate_value_range.
5768 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
5770 PR rtl-optimization/83030
5771 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
5772 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
5773 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5774 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5775 CROSSING_JUMP_P flag is already set before setting it.
5776 * cfgrtl.c (fixup_partition_crossing): Likewise.
5777 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
5780 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5782 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
5783 Handle the case where both arguments are using gen_const_vec_series.
5785 2017-11-22 David Malcolm <dmalcolm@redhat.com>
5788 * pretty-print.c (pp_format): Move quoting implementation to
5789 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
5790 to pass address of "quote" local.
5791 (pp_begin_quote): New function.
5792 (pp_end_quote): New function.
5793 * pretty-print.h (printer_fn): Convert penultimate param from bool
5795 (pp_begin_quote): New decl.
5796 (pp_end_quote): New decl.
5797 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5798 param from bool to bool *.
5799 * tree-diagnostic.h (default_tree_printer): Likewise.
5801 2017-11-22 Jeff Law <law@redhat.com>
5803 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5804 Initialize vr_values.
5805 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5806 vr_values via vr_values class instance rather than delegators.
5807 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5808 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5809 (evrp_range_analyzer::push_value_range): Likewise.
5810 (evrp_range_analyzer::pop_value_range): Likewise.
5811 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5812 most delegators. Those remaining are exposed as public interfaces.
5813 Make vr_values a pointer and private.
5814 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5816 (evrp_range_analyzer::get_vr_value): New method.
5817 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5818 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5819 (class evrp_dom_walker): Attach evrp_folder class, initialize
5820 it in the ctor. Remove temporary delegators.
5821 (evrp_dom_walker::before_dom_children): Call methods in attached
5822 evrp_range_analyzer class via class instance pointer. Use
5823 free value_range_constant_singleton to remove need for
5824 op_with_constant_singleton_value delegator method. Do not
5825 create a vrp_prop class instance for every call! Narrow
5826 scope of a couple variables.
5827 (evrp_dom_walker::cleanup): Call methods in attached
5828 evrp_range_analyzer class via class instance pointer.
5829 * vr-values.h (class vr_values): Privatize many methods and
5832 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5834 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5835 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5836 * expr.c (expand_expr_real_2): Likewise.
5837 * fold-const.c (const_unop): Likewise.
5838 * optabs-tree.c (optab_for_tree_code): Likewise.
5839 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5840 * tree-inline.c (estimate_operator_cost): Likewise.
5841 * tree-pretty-print.c (dump_generic_node): Likewise.
5842 (op_code_prio): Likewise.
5843 (op_symbol_code): Likewise.
5844 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5845 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5846 * internal-fn.c (direct_internal_fn_optab): New function.
5847 (direct_internal_fn_array, direct_internal_fn_supported_p
5848 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5849 * fold-const-call.c (fold_const_reduction): New function.
5850 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5852 * tree-vect-loop.c: Include internal-fn.h.
5853 (reduction_code_for_scalar_code): Rename to...
5854 (reduction_fn_for_scalar_code): ...this and return an internal
5856 (vect_model_reduction_cost): Take an internal_fn rather than
5858 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5860 (vectorizable_reduction): Use internal functions rather than tree
5861 codes for the reduction operation. Update calls to the functions
5863 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5864 Use calls to internal functions rather than REDUC tree codes.
5865 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5867 2017-11-22 Olivier Hainque <hainque@adacore.com>
5869 * config/vxworks.c (vxworks_override_options): Pick default
5870 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5871 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5872 DWARF_GNAT_ENCODINGS_DEFAULT.
5873 * config/vxworksae.h: Likewise.
5875 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5877 PR tree-optimization/83104
5878 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5881 2017-11-22 Marek Polacek <polacek@redhat.com>
5882 H.J. Lu <hongjiu.lu@intel.com>
5883 Jason Merrill <jason@redhat.com>
5888 * calls.c (initialize_argument_information): Call
5889 warn_parameter_passing_abi target hook.
5890 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5891 argument onto stack.
5892 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5893 * common.opt: Update -fabi-version description.
5894 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5895 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5897 (ix86_is_empty_record): New function.
5898 (ix86_warn_parameter_passing_abi): New function.
5899 (TARGET_EMPTY_RECORD_P): Redefine.
5900 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5901 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5902 * doc/tm.texi: Regenerated.
5903 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5904 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5905 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5906 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5907 instead of int_size_in_bytes.
5908 * expr.c (copy_blkmode_to_reg): Likewise.
5909 * function.c (aggregate_value_p): Return 0 for empty types.
5910 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5912 (locate_and_pad_parm): Call arg size_in_bytes instead
5914 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5915 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5916 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5918 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5919 (std_gimplify_va_arg_expr): Skip empty records. Call
5920 arg_size_in_bytes instead size_in_bytes.
5921 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5922 * tree-core.h (tree_type_common): Add empty_flag.
5923 (tree_decl_common): Update comments.
5924 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5926 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5927 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5929 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5930 * tree.c (default_is_empty_type): New function.
5931 (default_is_empty_record): New function.
5932 (arg_int_size_in_bytes): New function.
5933 (arg_size_in_bytes): New function.
5934 (get_ultimate_context): New function.
5935 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5936 TRANSLATION_UNIT_WARN_EMPTY_P.
5937 (default_is_empty_record, arg_int_size_in_bytes,
5938 arg_size_in_bytes, get_ultimate_context): Declare.
5940 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5942 * config/arm/arm.c (cmse_clear_registers): New function.
5943 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5944 code by call to cmse_clear_registers.
5945 (cmse_nonsecure_entry_clear_before_return): Likewise.
5947 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5949 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5950 (vdot_s32, vdotq_s32): New.
5951 (vdot_lane_u32, vdotq_lane_u32): New.
5952 (vdot_lane_s32, vdotq_lane_s32): New.
5955 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5958 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5959 for unsigned values with fewer HWIs than the precision.
5960 (test_overflow): New function.
5961 (wide_int_cc_tests): Call it.
5963 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5965 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5966 has the same mode class as Pmode.
5968 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5970 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5971 padding_bits_to_clear_ptr.
5972 (cmse_nonsecure_entry_clear_before_return): Likewise.
5974 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5976 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5977 auto_sbitap instead of integer bitfield to control register needing
5980 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5982 PR tree-optimization/83044
5983 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5984 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5985 that need the upper bound. Subtract offset from
5986 get_addr_base_and_unit_offset only if positive and subtract it
5987 before division by eltsize rather than after it.
5990 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5991 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5992 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5996 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5997 expand_binop with *mul_widen_optab, make sure at least one of the
5998 operands doesn't have VOIDmode.
6001 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6003 PR rtl-optimization/82044
6004 PR tree-optimization/82042
6005 * dse.c (record_store): Check for overflow.
6006 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6007 add_wild_read instead of clear_rhs_from_active_local_stores on
6008 overflow and log it into dump_file.
6010 2017-11-22 Richard Biener <rguenther@suse.de>
6012 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6013 fake edges to exit when looking for a place to insert.
6014 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6016 (insert_into_preds_of_block): Commit edge insertion immediately,
6017 assert that doesn't require new BBs.
6018 (fini_pre): Release expressions.
6019 (pass_pre::execute): Shuffle things around a bit, if the fn
6020 is too large do not compute AVAIL either as this is really the
6023 2017-11-22 Richard Biener <rguenther@suse.de>
6025 PR tree-optimization/83089
6026 * tree-if-conv.c (pass_if_conversion::execute): If anything
6027 changed reset SCEV and free the number of iteration estimates.
6029 2017-11-21 Martin Sebor <msebor@redhat.com>
6031 PR tree-optimization/82945
6032 * calls.h (warn_nonstring_bound): Remove unused function.
6034 2017-11-21 Martin Sebor <msebor@redhat.com>
6036 PR tree-optimization/82945
6037 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6038 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6039 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6041 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6042 * calls.h (get_attr_nonstring_decl): Declare new function.
6043 * doc/extend.texi (attribute nonstring): Update.
6044 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6045 get_attr_nonstring_decl and handle it.
6046 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6047 detection of nul-termination.
6048 (strlen_to_stridx): Change to a pointer.
6049 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6050 (pass_strlen::execute): Same.
6052 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6054 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6055 for the new option -mprefer-vector-width=[none|128|256|512].
6056 * config/i386/i386.c (ix86_target_string): remove old style options
6057 -mprefer-avx256 and make -mprefer-avx128 as alias.
6058 (ix86_option_override_internal): Apply defaults for the
6059 -mprefer-vector-width=[128|256] option.
6060 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6061 Implement macros to work with -mprefer-vector-width=.
6062 * config/i386/i386.opt: Implemented option
6063 -mprefer-vector-width=[none|128|256|512].
6064 * doc/invoke.texi: Documentation for
6065 -mprefer-vector-width=[none|128|256|512].
6067 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6069 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6071 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6073 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6075 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6077 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6079 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6080 * expr.c (expand_expr_real_2): Likewise.
6081 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6082 fold_binary_loc): Likewise.
6083 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6084 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6085 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6086 MINUS_EXPR transformations.
6087 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6088 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6089 * tree-inline.c (estimate_operator_cost): Likewise.
6090 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6091 op_symbol_code): Likewise.
6092 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6093 * vr-values.c (extract_range_from_binary_expr): Likewise.
6094 * varasm.c (initializer_constant_valid_p_1): Likewise.
6095 * tree.def: New tree code POINTER_DIFF_EXPR.
6097 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6099 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6100 integer suffix to movbe mnemonic.
6101 (*bswaphi2_movbe): Ditto.
6102 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6104 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6107 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6108 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6109 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6111 2017-11-21 Martin Liska <mliska@suse.cz>
6113 * tree-inline.c (expand_call_inline): Remove not needed
6116 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6117 Jakub Jelinek <jakub@redhat.com>
6120 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6121 Remove static keyword from f variable.
6123 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6125 PR tree-optimization/83086
6126 * gimple-ssa-store-merging.c
6127 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6128 rather than n.base_addr.
6130 2017-11-21 Martin Liska <mliska@suse.cz>
6132 PR rtl-optimization/82044
6133 PR tree-optimization/82042
6134 * dse.c (check_mem_read_rtx): Check for overflow.
6136 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6138 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6141 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6143 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6144 and its corresponding call arg location note.
6146 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6147 Andrew Burgess <andrew.burgess@embecosm.com>
6149 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6151 (arc_return_slot_offset): Likewise.
6152 (arc_eh_return_address_location): New declaration.
6153 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6154 (MUST_SAVE_REGISTER): Add exception handler case.
6155 (MUST_SAVE_RETURN_ADDR): Likewise.
6156 (arc_frame_pointer_required): Likewise.
6157 (arc_frame_pointer_needed): New function.
6158 (arc_compute_frame_size): Changed.
6159 (arc_expand_prologue): Likewise.
6160 (arc_expand_epilogue): Likewise.
6161 (arc_initial_elimination_offset): Likewise.
6162 (arc_return_slot_offset): Delete.
6163 (arc_eh_return_address_location): New function.
6164 (arc_builtin_setjmp_frame_value): Likewise.
6165 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6166 (EH_RETURN_STACKADJ_RTX): Define.
6167 (EH_RETURN_HANDLER_RTX): Likewise.
6168 * config/arc/arc.md (eh_return): Delete.
6170 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6172 * print-tree.h (debug_vec_tree): Remove prototype.
6173 * gdbinit.in (pvt): Remove macro.
6175 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6177 PR tree-optimization/83047
6178 * gimple-ssa-store-merging.c
6179 (imm_store_chain_info::output_merged_store): If the loads with the
6180 same vuse are in different basic blocks, for load_gsi pick a load
6181 location that is dominated by the other loads.
6184 * config/i386/i386.c (ix86_memmodel_check): Start
6185 -Winvalid-memory-model diagnostics with lowercase letter.
6188 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6189 set in_cold_section_p to the partition containing loc_list->first.
6190 When seeing loc_list->last_before_switch node, update secname and
6191 perform range_across_switch second partition handling only after that.
6194 * run-rtl-passes.c: Include debug.h.
6195 (run_rtl_passes): Call debug_hooks->assembly_start.
6196 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6200 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6201 OPTAB_DIRECT in calls to expand_simple_binop.
6203 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6206 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6208 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6211 * spellcheck.h (best_match::blithely_get_best_candidate): New
6214 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6216 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6217 with lower case letter.
6219 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6221 * config/i386/i386.md (bswaphi2): New expander.
6222 (*bswaphi2_movbe): New insn pattern.
6223 (bswaphi -> rorhi pepehole2): New peephole pattern.
6225 2017-11-20 Jeff Law <law@redhat.com>
6227 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6228 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6229 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6230 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6231 gimple-ssa-evrp-analyze.h.
6233 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6234 set BB_VISITED here.
6235 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6237 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6239 * tree-predcom.c: Add general comment on Store-Store chains.
6240 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6242 (get_chain_last_ref_at): Rename into...
6243 (get_chain_last_write_at): ...this.
6244 (get_chain_last_write_before_load): New function.
6245 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6246 CT_STORE_STORE when write reference is added.
6247 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6248 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6249 (initialize_root_vars_store_elim_1): Ditto.
6250 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6251 definition is created.
6252 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6253 chain by replacing it with dominant stored value.
6255 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6257 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6259 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6261 * vr-values.c (extract_range_from_binary_expr): Use a full range
6264 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6266 * config/arm/arm.md (R4_REGNUM): Define constant.
6267 (nonsecure_call_internal): Remove r4 clobber.
6268 (nonsecure_call_value_internal): Likewise.
6269 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6270 clobber and resequence match_operands.
6271 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6272 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6273 (nonsecure_call_value_reg_thumb2): Likewise.
6275 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6277 PR tree-optimization/78821
6278 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6279 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6280 to the address of the base rather than the base itself.
6281 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6282 (find_bswap_or_nop_finalize): New function.
6283 (find_bswap_or_nop): Use it.
6284 (bswap_replace): Return a tree rather than bool, change first
6285 argument from gimple * to gimple_stmt_iterator, allow inserting
6286 into an empty sequence, allow ins_stmt to be NULL - then emit
6287 all stmts into gsi. Fix up MEM_REF address gimplification.
6288 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6289 (struct store_immediate_info): Add N and INS_STMT non-static
6291 (store_immediate_info::store_immediate_info): Initialize them
6292 from newly added ctor args.
6293 (merged_store_group::apply_stores): Formatting fixes. Sort by
6295 (stmts_may_clobber_ref_p): For stores call also
6296 refs_anti_dependent_p.
6297 (gather_bswap_load_refs): New function.
6298 (imm_store_chain_info::try_coalesce_bswap): New method.
6299 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6300 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6301 (imm_store_chain_info::output_merged_store): Fail if number of
6302 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6303 and NOP_EXPR rhs_code.
6304 (pass_store_merging::process_store): Compute n and ins_stmt, if
6305 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6306 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6308 (pass_store_merging::execute): Calculate dominators.
6310 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6311 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6312 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6313 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6314 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6315 class pass_optimize_bswap, bswap_replace,
6316 pass_optimize_bswap::execute): Moved to ...
6317 * gimple-ssa-store-merging.c: ... this file.
6318 Include optabs-tree.h.
6319 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6320 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6321 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6322 anonymous namespace, remove static keywords.
6323 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6324 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6326 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6329 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6331 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6333 * vec.h (debug_helper): New function.
6334 (DEFINE_DEBUG_VEC): New macro.
6335 * hash-set.h (debug_helper): New function.
6336 (DEFINE_DEBUG_HASH_SET): New macro.
6337 * cfg.c (debug_slim (edge)): New function.
6338 Call DEFINE_DEBUG_VEC for edges.
6339 Call DEFINE_DEBUG_HASH_SET for edges.
6340 * cfghooks.c (debug_slim (basic_block)): New function.
6341 Call DEFINE_DEBUG_VEC for basic blocks.
6342 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6343 * print-tree.c (debug_slim): New function to handle trees.
6344 Call DEFINE_DEBUG_VEC for trees.
6345 Call DEFINE_DEBUG_HASH_SET for trees.
6346 (debug (vec<tree, va_gc>) &): Remove.
6347 (debug (<vec<tree, va_gc>) *): Remove.
6348 * print-rtl.c (debug_slim): New function to handle const_rtx.
6349 Call DEFINE_DEBUG_VEC for rtx_def.
6350 Call DEFINE_DEBUG_VEC for rtx_insn.
6351 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6352 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6353 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6354 (debug (vec<rtx_insn *> *ptr): Remove.
6355 (debug_insn_vector): Remove.
6356 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6358 2017-11-20 Tom de Vries <tom@codesourcery.com>
6360 PR rtl-optimization/82020
6361 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6362 IF_THEN_ELSE condition.
6364 2017-11-19 Jeff Law <law@redhat.com>
6366 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6367 of degenerates resulting from ignoring an edge.
6369 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6372 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6374 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6377 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6378 for uninitialied counts.
6380 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6383 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6385 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6388 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6390 2017-11-19 Tom de Vries <tom@codesourcery.com>
6392 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6395 2017-11-19 Tom de Vries <tom@codesourcery.com>
6397 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6400 2017-11-19 Tom de Vries <tom@codesourcery.com>
6402 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6405 2017-11-19 Tom de Vries <tom@codesourcery.com>
6407 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6409 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6411 2017-11-19 Tom de Vries <tom@codesourcery.com>
6413 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6415 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6416 semicolon after MCORE_EXPORT_NAME call.
6418 2017-11-19 Tom de Vries <tom@codesourcery.com>
6421 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6422 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6424 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6426 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6427 (cgraph_edge::clone): Cleanup updating of profile.
6428 * ipa-cp.c (update_profiling_info): Likewise.
6429 * ipa-inline-transform.c (inline_transform): Likewise.
6430 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6431 * ipa-split.c (execute_split_functions): Do not split when function
6433 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6434 * profile-count.c (profile_count::dump): Add global0.
6435 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6437 (profile_count::to_sreal_scale): Likewise.
6438 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6439 (profile_count::combine_with_ipa_count): New function.
6440 * profile-count.h (profile_guessed_global0adjusted): New.
6441 (profile_count::adjusted_zero): New.
6442 (profile_count::global0adjusted): New.
6443 (profile_count::combine_with_ipa_count): New.
6444 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6445 correct profile of return block of split functions.
6446 (copy_cfg_body): Remove unused profile_count.
6447 (copy_body): Likewise.
6448 (expand_call_inline): Update.
6449 (tree_function_versioning): Update.
6451 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6453 * hash-set.h (hash_set::empty): New.
6454 * tree-ssa-threadbackward.h: Delete.
6455 * tree-ssa-threadbackward.c (class thread_jumps): New.
6456 Move max_threaded_paths into class.
6457 (fsm_find_thread_path): Remove arguments that are now in class.
6458 (profitable_jump_thread_path): Rename to...
6459 (thread_jumps::profitable_jump_thread_path): ...this.
6460 (convert_and_register_jump_thread_path): Rename to...
6461 (thread_jumps::convert_and_register_current_path): ...this.
6462 (check_subpath_and_update_thread_path): Rename to...
6463 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6464 (register_jump_thread_path_if_profitable): Rename to...
6465 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6466 (handle_phi): Rename to...
6467 (thread_jumps::handle_phi): ...this.
6468 (handle_assignment): Rename to...
6469 (thread_jumps::handle_assignment): ...this.
6470 (fsm_find_control_statement_thread_paths): Rename to...
6471 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6472 (find_jump_threads_backwards): Rename to...
6473 (thread_jumps::find_jump_threads_backwards): ...this.
6474 Initialize path local data.
6475 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6476 from within thread_jumps class.
6477 (pass_early_thread_jumps::execute): Same.
6479 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6481 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6483 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6485 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6486 to canonical location.
6488 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6491 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6493 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6495 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6497 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6498 rcrt1.o%s/grcrt1.o%s for -static-pie.
6500 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6502 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6503 ix86_shift_rotate_cost): Break out from ...
6504 (ix86_rtx_costs): ... here.
6505 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6508 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6510 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6511 when entry block was promoted unlikely.
6512 (estimate_bb_frequencies): Increase frequency scale.
6513 * profile-count.h (profile_count): Export precision info.
6515 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6517 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6518 disturbin profile of entry block.
6520 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6522 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6525 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6527 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6528 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6530 2017-11-17 Jeff Law <law@redhat.com>
6532 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6533 from evrp_dom_walker class. Various methods moved into new class.
6534 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6535 (evrp_range_analyzer::enter): New method.
6536 (evrp_range_analyzer::leave): New method.
6537 (evrp_dom_walker): Remove delegators no longer needed by this class.
6538 Replace vr_values data member with evrp_range_analyzer
6540 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6541 method extracted from evrp_dom_walker::before_dom_children.
6542 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6543 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6545 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6546 Add private copy constructor and move assignment operators.
6547 Privatize methods and class data where trivially possible.
6548 (evrp_dom_walker::cleanup): New function, extracted from
6549 execute_early_vrp. Simplify access to class data.
6551 * vr-values.h (get_output_for_vrp): Prototype.
6552 * vr-values.c (get_output_for_vrp): New function extracted from
6553 vrp_visit_assignment_or_call and extract_range_from_stmt.
6554 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6556 2017-11-17 Luis Machado <luis.machado@linaro.org>
6558 * config/aarch64/aarch64.c
6559 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6560 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6561 tune_params::AUTOPREFETCHER_WEAK.
6563 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6566 * config/arm/arm.c (arm_valid_target_attribute_rec):
6567 Parse "arch=" and "+<ext>".
6568 (arm_valid_target_attribute_tree): Re-init global options.
6569 (arm_option_override): Make non-static.
6570 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6571 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6572 (__ARM_FEATURE_CRC32): Support undef.
6573 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6574 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6576 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6578 * gdbinit.in (break-on-diagnostic): New command.
6580 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6582 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6583 return to indirect jump for EH return if control-flow
6584 protection is enabled. Change explicit 'false' argument in
6585 pro_epilogue_adjust_stack with a value of
6587 * config/i386/i386.md (simple_return_indirect_internal):
6588 Remove SImode restriction to support 64-bit.
6590 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6592 * combine.c (added_notes_insn): New.
6593 (try_combine): Handle added_notes_insn like added_links_insn.
6594 Rewrite return value code.
6595 (distribute_notes): Set added_notes_insn to the earliest insn we added
6598 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6600 PR rtl-optimization/82621
6601 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6602 dest of one of those SETs is unused.
6604 2017-11-17 Richard Biener <rguenther@suse.de>
6607 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6608 * tree-pretty-print.c (dump_generic_node): Handle
6609 annot_expr_parallel_kind.
6610 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6611 * gimplify.c (gimple_boolify): Likewise.
6613 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6615 * config.gcc (extra_headers): Add cet.h for x86 targets.
6616 * config/i386/cet.h: New file.
6617 * doc/install.texi: Add --enable-cet/--disable-cet.
6619 2017-11-17 Richard Biener <rguenther@suse.de>
6621 PR tree-optimization/83017
6622 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6623 (gen_parallel_loop): Properly count iterations.
6624 (parallelize_loops): Handle loop->can_be_parallel independent
6625 of flag_loop_parallelize_all. Make static profitability test match
6627 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6628 * invoke.texi (parloops-min-per-thread): Document.
6630 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6632 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6633 upstreaming review comments.
6635 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6637 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6638 with fast unaligned access.
6639 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6641 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6643 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6644 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6646 2017-11-17 Richard Biener <rguenther@suse.de>
6648 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6649 folding of references.
6651 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6654 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6655 of replacing call to strncmp with corresponding call to strcmp when
6658 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6660 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6661 option prefer-avx256 for skylake-avx512 configuration.
6662 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6663 (get_builtin_code_for_version): Ditto.
6665 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6666 Monk Chiang <sh.chiang04@gmail.com>
6668 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6669 (FIXED_REGISTERS): Reserve more register numbers.
6670 (CALL_USED_REGISTERS): Likewise.
6671 (REG_ALLOC_ORDER): Likewise.
6672 (REG_CLASS_CONTENTS): Likewise.
6673 (REGISTER_NAMES): Likewise.
6675 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6676 Kito Cheng <kito.cheng@gmail.com>
6678 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6680 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6682 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6684 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6685 parameter list for vec_splats.
6687 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6689 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6691 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6693 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6696 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6697 with the required version of automake 1.11.6.
6699 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6701 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6703 (DU_C2_3_power9): Correct reservation combinations.
6704 (FP_DIV_power9, VEC_DIV_power9): New.
6705 (power9-alu): Split out rotate/shift...
6706 (power9-rot): ...to here, correct dispatch resource.
6707 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6709 (power9-fp): Correct latency.
6710 (power9-sdiv): Add div/sqrt resource.
6711 (power9-ddiv): Correct latency, add div/sqrt resource.
6712 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
6713 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
6715 (power9-qpdiv, power9-qpmul): Adjust resource usage.
6717 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
6719 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
6720 switch statement mapping KF built-ins to TF built-ins if we don't
6721 have the proper ISA 3.0 assembler support.
6723 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6725 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
6726 (lower_emutls_function_body): Do not compute it.
6728 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6730 * ipa-split.c (split_bb_info): Turn time to sreal.
6731 (split_point): Likewise.
6732 (dump_split_point): Likewise.
6733 (fine_split_points): Likewise.
6734 (execute_split_functions): Only zero split_bbs; turn time to sreals.
6736 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6738 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6741 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6743 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6745 (expensive_function_p): Remove useless assert.
6746 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6748 2017-11-16 Martin Sebor <msebor@redhat.com>
6750 PR tree-optimization/82588
6751 PR tree-optimization/82583
6752 * tree-vrp.c (check_array_ref): Handle flexible array members,
6753 string literals, and inner indices.
6754 (search_for_addr_array): Add detail to diagnostics.
6756 2017-11-16 Nathan Sidwell <nathan@acm.org>
6760 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6761 (SET_DECL_ASSEMBLER_NAME): Forward to
6762 overwrite_decl_assembler_name.
6763 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6764 (overwrite_decl_assembler_name): Declare.
6765 * tree.c (overwrite_decl_assembler_name): New.
6766 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
6767 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
6768 (LANG_HOOKS_INITIALIZER): Add it.
6769 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
6770 * langhooks.c (lhd_set_decl_assembler_name): Use
6771 SET_DECL_ASSEMBLER_NAME.
6772 (lhd_overwrite_decl_assembler_name): Default implementation.
6774 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6775 Jackson Woodruff <jackson.woodruff@arm.com>
6777 PR tree-optimization/71026
6778 * match.pd: Canonicalize constant multiplies in division.
6780 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6782 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
6783 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
6784 Remove OPT_fomit_frame_pointer entry.
6785 * common/config/arc/arc-common.c: Likewise.
6786 * common/config/arm/arm-common.c: Likewise.
6787 * common/config/avr/avr-common.c: Likewise.
6788 * common/config/c6x/c6x-common.c: Likewise.
6789 * common/config/cr16/cr16-common.c: Likewise.
6790 * common/config/cris/cris-common.c: Likewise.
6791 * common/config/epiphany/epiphany-common.c: Likewise.
6792 * common/config/fr30/fr30-common.c: Likewise.
6793 * common/config/frv/frv-common.c: Likewise.
6794 * common/config/ia64/ia64-common.c: Likewise.
6795 * common/config/iq2000/iq2000-common.c: Likewise.
6796 * common/config/lm32/lm32-common.c: Likewise.
6797 * common/config/m32r/m32r-common.c: Likewise.
6798 * common/config/mcore/mcore-common.c: Likewise.
6799 * common/config/microblaze/microblaze-common.c: Likewise.
6800 * common/config/mips/mips-common.c: Likewise.
6801 * common/config/mmix/mmix-common.c: Likewise.
6802 * common/config/mn10300/mn10300-common.c: Likewise.
6803 * common/config/nios2/nios2-common.c: Likewise.
6804 * common/config/pa/pa-common.c: Likewise.
6805 * common/config/pdp11/pdp11-common.c: Likewise.
6806 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6807 * common/config/riscv/riscv-common.c: Likewise.
6808 * common/config/rs6000/rs6000-common.c: Likewise.
6809 * common/config/rx/rx-common.c: Likewise.
6810 * common/config/s390/s390-common.c: Likewise.
6811 * common/config/sh/sh-common.c: Likewise.
6812 * common/config/sparc/sparc-common.c: Likewise.
6813 * common/config/tilegx/tilegx-common.c: Likewise.
6814 * common/config/tilepro/tilepro-common.c: Likewise.
6815 * common/config/v850/v850-common.c: Likewise.
6816 * common/config/visium/visium-common.c: Likewise.
6817 * common/config/xstormy16/xstormy16-common.c: Likewise.
6818 * common/config/xtensa/xtensa-common.c: Likewise.
6819 * invoke.texi (-fomit-frame-pointer): Update documentation.
6821 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6823 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6825 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6827 * cfg.c (scale_bbs_frequencies_int,
6828 cale_bbs_frequencies_gcov_type): Remove.
6829 * cfg.h (scale_bbs_frequencies_int,
6830 cale_bbs_frequencies_gcov_type): Remove.
6832 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6834 * tree-ssa-loop-manip.c
6835 (scale_dominated_blocks_in_loop): Update to profile counts.
6836 (tree_transform_and_unroll_loop): Likewise.
6838 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6840 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6841 scale_bbs_frequencies_int.
6843 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6845 * final.c (compute_alignments): Use counts rather than frequencies.
6847 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6849 * cfgloopanal.c: Include sreal.h
6850 (average_num_loop_insns): Use counts and sreal for accounting.
6852 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6854 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6857 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6858 Kito Cheng <kito.cheng@gmail.com>
6860 * config/nds32/constraints.md: Provide more constraints.
6861 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6862 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6863 support constraints usage.
6865 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6867 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6869 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6870 Kito Cheng <kito.cheng@gmail.com>
6872 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6873 * config/nds32/nds32.opt: Refine the layout.
6874 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6875 TARGET_EXT_STRING): Support new options.
6876 * config/nds32/nds32.h: Likewise.
6877 * config/nds32/nds32.md: Likewise.
6878 * config/nds32/nds32-predicates.c: Likewise.
6879 * config/nds32/constraints.md: Likewise.
6880 * common/config/nds32/nds32-common.c: Likewise.
6882 2017-11-16 Julia Koval <julia.koval@intel.com>
6885 * config/i386/gfniintrin.h: Add sse check.
6886 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6888 2017-11-16 Julia Koval <julia.koval@intel.com>
6890 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6891 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6892 (ix86_handle_option): Handle -mavx512vbmi2.
6893 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6894 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6895 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6896 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6897 (ix86_valid_target_attribute_inner_p): Ditto.
6898 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6899 * config/i386/i386.opt (mavx512vbmi2): New option.
6900 * doc/invoke.texi: Add new option.
6902 2017-11-16 Julia Koval <julia.koval@intel.com>
6904 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6905 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6906 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6907 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6908 _mm512_gf2p8mul_epi8): New intrinsics.
6909 * config/i386/i386-builtin-types.def
6910 (V64QI_FTYPE_V64QI_V64QI): New type.
6911 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6912 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6913 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6914 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6915 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6916 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6918 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6920 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6921 explicitly as a stream of bytes.
6923 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6925 * config/rs6000/altivec.h (vec_xst_be): New #define.
6926 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6927 and externalize from *altivec_vperm_<mode>_internal.
6928 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6930 (XL_BE_V8HI): Likewise.
6931 (XL_BE_V4SI): Likewise.
6932 (XL_BE_V4SI): Likewise.
6933 (XL_BE_V2DI): Likewise.
6934 (XL_BE_V4SF): Likewise.
6935 (XL_BE_V2DF): Likewise.
6936 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6937 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6938 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6939 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6940 VSX_BUILTIN_VEC_XST_BE.
6941 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6942 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6944 (altivec_init_builtins): Replace conditional calls to def_builtin
6945 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6946 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6947 with unconditional calls. Remove calls to def_builtin for
6948 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6949 __builtin_vec_xst_be.
6950 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6951 to define_expand, and add alternate RTL generation for P8.
6952 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6953 vsx_ld_elemrev_v8hi.
6954 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6955 add alternate RTL generation for P8.
6956 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6957 vsx_ld_elemrev_v16qi.
6958 (vsx_st_elemrev_v8hi): Convert define_insn
6959 to define_expand, and add alternate RTL generation for P8.
6960 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6961 vsx_st_elemrev_v8hi.
6962 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6963 add alternate RTL generation for P8.
6964 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6965 vsx_st_elemrev_v16qi.
6967 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6970 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6971 TARGET_AVX512ER check.
6972 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6973 neither -mzeroupper nor -mno-zeroupper is used and
6974 TARGET_EMIT_VZEROUPPER is set.
6975 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6976 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6978 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6980 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6981 folding of vector compares.
6982 (fold_build_vec_cmp): New helper function.
6983 (fold_compare_helper): New helper function.
6984 (builtin_function_type): Add compare builtins to the list of functions
6985 having unsigned arguments. Cosmetic updates to comment indentation.
6986 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6987 the not+eq combination.
6989 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6991 PR tree-optimization/82726
6992 PR tree-optimization/70754
6993 * tree-predcom.c (order_drefs_by_pos): New function.
6994 (combine_chains): Move code setting has_max_use_after to...
6995 (try_combine_chains): ...here. New parameter. Sort combined chains
6996 according to position information.
6997 (tree_predictive_commoning_loop): Update call to above function.
6998 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7000 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7002 PR tree-optimization/82726
7004 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7006 PR tree-optimization/70754
7007 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7008 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7009 combined stmt before it if not NULL.
7010 (combine_chains): Process refs reversely and compute dominance point
7014 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7016 PR tree-optimization/79663
7017 * tree-predcom.c (combine_chains): Process refs in reverse order
7018 only for ZERO length chains, and add explaining comment.
7020 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7022 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7023 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7024 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7026 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7028 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7030 2017-11-15 Martin Liska <mliska@suse.cz>
7032 * tree-cfg.c (pass_warn_function_return::execute):
7033 Compare warn_return_type for greater than zero.
7035 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7039 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7040 to return true on Xeon and not on Xeon Phi.
7041 (ix86_check_avx256_register): Changed to ...
7042 (ix86_check_avx_upper_register): ... this. Add extra check for
7043 VALID_AVX512F_REG_OR_XI_MODE.
7044 (ix86_avx_u128_mode_needed): Changed
7045 ix86_check_avx256_register to ix86_check_avx_upper_register.
7046 (ix86_check_avx256_stores): Changed to ...
7047 (ix86_check_avx_upper_stores): ... this. Changed
7048 ix86_check_avx256_register to ix86_check_avx_upper_register.
7049 (ix86_avx_u128_mode_after): Changed
7050 avx_reg256_found to avx_upper_reg_found. Changed
7051 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7052 (ix86_avx_u128_mode_entry): Changed
7053 ix86_check_avx256_register to ix86_check_avx_upper_register.
7054 (ix86_avx_u128_mode_exit): Ditto.
7055 * config/i386/i386.h: (host_detect_local_cpu): New define.
7057 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7059 * config/arm/xgene1.md (xgene1): Split into automatons
7060 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7061 (xgene1_f_load): Adjust reservations and/or types.
7062 (xgene1_f_store): Likewise.
7063 (xgene1_load_pair): Likewise.
7064 (xgene1_store_pair): Likewise.
7065 (xgene1_fp_load1): Likewise.
7066 (xgene1_load1): Likewise.
7067 (xgene1_store1): Likewise.
7068 (xgene1_move): Likewise.
7069 (xgene1_alu): Likewise.
7070 (xgene1_simd): Likewise.
7071 (xgene1_bfm): Likewise.
7072 (xgene1_neon_load1): Likewise.
7073 (xgene1_neon_store1): Likewise.
7074 (xgene1_neon_logic): Likewise.
7075 (xgene1_neon_st1): Likewise.
7076 (xgene1_neon_ld1r): Likewise.
7077 (xgene1_alu_cond): Added.
7078 (xgene1_shift_reg): Likwise.
7079 (xgene1_bfx): Likewise.
7080 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7082 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7085 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7087 (can_widen_mult_without_libcall): New function.
7088 (expand_mul_overflow): If only checking unsigned mul overflow,
7089 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7090 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7091 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7092 (expand_DIVMOD): Formatting fix.
7093 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7094 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7095 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7097 PR tree-optimization/82977
7098 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7099 constructed temporary to strlen_to_stridx.put.
7101 2017-11-15 Martin Liska <mliska@suse.cz>
7103 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7104 * configure: Regenerate.
7106 2017-11-15 Martin Liska <mliska@suse.cz>
7109 * config/sh/sh-mem.cc: Use proper probability for
7112 2017-11-14 Jeff Law <law@redhat.com>
7114 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7115 the red zone for stack_clash_protection_final_dynamic_probe targets
7116 when the total dynamic stack size is zero bytes.
7118 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7119 blocks is post order.
7121 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7123 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7124 * final.c (rest_of_clean_state): Set it for the
7125 -fcompare-debug dump.
7126 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7127 class when TDF_COMPARE_DEBUG is set.
7129 * dwarf2out.c (gen_producer_string): Discard
7132 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7135 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7136 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7137 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7138 (__TGMATH_REAL_2_3): Remove macros.
7139 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7140 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7142 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7144 (remquo): Define using __TGMATH_REAL_3.
7146 2017-11-14 Jeff Law <law@redhat.com>
7148 * vr-values.c: New file with contents extracted from tree-vrp.c.
7149 * Makefile.in (OBJS): Add vr-values.o
7150 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7151 (set_value_range, set_and_canonicalize_value_range): Likewise.
7152 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7153 (value_range_constant_singleton, symbolic_range_p): Likewise.
7154 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7155 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7156 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7157 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7158 (find_case_label_range, find_case_label_index): Likewise.
7159 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7160 (range_int_cst_singleton_p, value_inside_range): Likewise.
7161 (get_single_symbol): Likewise.
7162 (switch_update): Move structure definition here.
7163 (to_remove_edges, to_update_switch_stmts): Provide externs.
7164 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7165 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7166 (vrp_val_is_min, set_value_range): Likewise.
7167 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7168 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7169 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7170 (range_is_nonnull, range_int_cst_p): Likewwise.
7171 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7172 (get_single_symbol, operand_less_p): Likewise
7173 (compare_values_warnv, compare_values): Likewise.
7174 (value_inside_range, value_range_constant_singleton): Likewise.
7175 (zero_nonzero_bitgs_from_vr): Likewise.
7176 (extract_range_from_binary_expr_1): Likewise.
7177 (overflow_comparison_p): Likewise.
7178 (to_remove_edges, to_update_switch_stmts): Likewise.
7179 (find_case_label-index, find_case_label_range): Likewise.
7180 (switch_update, set_value_range_to_nonnegative): Remove.
7181 (set_value_range_to_truthvalue): Likewise.
7182 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7183 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7184 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7185 (find_case_label_ranges, test_for_singularity): Likewise.
7186 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7187 (x_vr_values): Move to its remaining use site.
7189 2017-11-10 Jeff Law <law@redhat.com>
7191 * vr-values.h (VR_INITIALIZER): Move #define here.
7192 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7193 * Makefile.in (OBJS): Add tree-evrp.o
7194 * tree-vrp.h (assert_info): Move structure definition here.
7195 (set_value_range_to_varying): Prototype.
7196 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7197 (infer_value_range, register_edge_assert_for): Likewise.
7198 (stmt_interesting_for_vrp): Likewise.
7199 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7200 (set_value_range_to_varying): No longer static.
7201 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7202 (infer_value_range, register_edge_assert_for): Likewise.
7204 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7206 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7207 generate the XXBRD instruction.
7209 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7210 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7211 as being compatible if -mabi=ieeelongdouble.
7212 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7213 to setup float128 built-ins with hardware support.
7214 (BU_FLOAT128_HW_2): Likewise.
7215 (BU_FLOAT128_HW_3): Likewise.
7216 (BU_FLOAT128_HW_VSX_1): Likewise.
7217 (BU_FLOAT128_HW_VSX_2): Likewise.
7218 (scalar_extract_expq): Change float128 built-in functions to
7219 accommodate having both KFmode and TFmode functions. Use the
7220 KFmode variant as the default.
7221 (scalar_extract_sigq): Likewise.
7222 (scalar_test_neg_qp): Likewise.
7223 (scalar_insert_exp_q): Likewise.
7224 (scalar_insert_exp_qp): Likewise.
7225 (scalar_test_data_class_qp): Likewise.
7226 (sqrtf128_round_to_odd): Delete processing the round to odd
7227 built-in functions as special built-in functions, and define them
7228 as float128 built-ins. Use the KFmode variant as the default.
7229 (truncf128_round_to_odd): Likewise.
7230 (addf128_round_to_odd): Likewise.
7231 (subf128_round_to_odd): Likewise.
7232 (mulf128_round_to_odd): Likewise.
7233 (divf128_round_to_odd): Likewise.
7234 (fmaf128_round_to_odd): Likewise.
7235 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7236 support for KFmode and TFmode xststdcqp calls.
7237 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7238 point, switch the built-in handlers for the get/set float128
7239 exponent, get float128 mantissa, float128 test built-ins, and the
7240 float128 round to odd built-in functions. Eliminate creating the
7241 float128 round to odd built-in functions as special built-ins.
7242 (rs6000_init_builtins): Eliminate special creation of the float128
7243 round to odd built-in functions.
7244 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7245 function insns to support both TFmode and KFmode variants.
7246 (xsxsigqp_<mode>): Likewise.
7247 (xsiexpqpf_<mode>): Likewise.
7248 (xsiexpqp_<mode>): Likewise.
7249 (xststdcqp_<mode>): Likewise.
7250 (xststdcnegqp_<mode>): Likewise.
7251 (xststdcqp_<mode>): Likewise.
7253 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7255 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7256 unused path_in_freq_ptr parameter.
7257 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7259 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7261 * ipa-inline.c (edge_badness): Dump sreal frequency.
7262 (compute_inlined_call_time): Match natural implementaiton ...
7263 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7264 forgotten division by CGRAPH_FREQ_BASE.
7266 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7268 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7269 Solaris 11. Update comment.
7270 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7272 * config/sol2.h (STARTFILE_SPEC): Likewise.
7273 * configure: Regenerate.
7275 2017-11-14 Carl Love <cel@us.ibm.com>
7277 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7278 le_ and be_ prefixes to swap* variables. Remove
7279 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7282 2017-11-14 Jason Merrill <jason@redhat.com>
7284 Support GTY((cache)) on hash_map.
7285 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7286 (ggc_cache_remove): Override it instead of ggc_mx.
7287 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7288 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7289 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7290 (simple_cache_map_traits): Override maybe_mx.
7291 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7292 (hash_map): Friend gt_cleare_cache.
7293 (gt_cleare_cache): New.
7294 * tree.h (tree_cache_traits): New hash_map traits class.
7295 (tree_cache_map): New typedef.
7297 2017-11-14 Richard Biener <rguenther@suse.de>
7299 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7300 paramter and handling.
7301 (cleanup_control_flow_bb): Likewise.
7302 (cleanup_control_flow_pre): New helper performing a DFS walk
7303 to call cleanup_control_flow_bb in PRE order.
7304 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7305 via cleanup_control_flow_pre.
7307 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7309 * config/aarch64/aarch64-simd.md
7310 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7311 (*aarch64_simd_bsl<mode>_alt): Likewise.
7312 (aarch64_simd_bsldi_internal): New.
7313 (aarch64_simd_bsldi_alt): Likewise.
7315 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7317 * tracer.c (better_p): Do not compare frequencies.
7318 * reg-stack.c (better_edge): Likewise.
7319 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7322 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7324 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7325 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7326 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7327 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7328 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7329 * predict.c (maybe_hot_frequency_p): Inline to ...
7330 (maybe_hot_count_p): ... here; rewrite to counts.
7331 (counts_to_freqs): Rename to ...
7332 (update_max_bb_count): ... this one.
7333 (expensive_function_p): Use counts.
7334 (estimate_bb_frequencies): Update.
7335 (rebuild_frequencies): Update.
7336 * predict.h (counts_to_freqs): Rename to ...
7337 (update_max_bb_count): ... this one.
7338 * profile.c (compute_branch_probabilities): Add debug info
7339 * tree-inline.c (expand_call_inline): Update debug info.
7340 (optimize_inline_calls): Use update_max_bb_count..
7341 (tree_function_versioning): Use update_max_bb_count..
7342 * value-prof.c (gimple_value_profile_transformations):
7343 Do not use update_max_bb_count.
7345 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7347 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7348 always use frequencies.
7350 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7352 * bb-reorder.c: Remove frequencies from comments.
7353 (better_edge_p): Use profile counts.
7354 (find_traces): Dump profile counts.
7355 (rotate_loop): Use profile counts.
7356 (find_traces_1_round): Likewise.
7357 (connect_better_edge_p): Use counts instead of probabilities for
7359 (copy_bb_p): Drop early check for non-0 frequency.
7360 (sanitize_hot_paths): Update comments.
7362 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7364 * ipa-split.c (struct split_point): Add count.
7365 (consider_split): Do not compute incoming frequency; compute incoming
7366 count and store it to split_point.
7367 (split_function): Set count of the call to split part correctly.
7369 2017-11-13 Carl Love <cel@us.ibm.com>
7371 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7373 2017-11-13 Tom Tromey <tom@tromey.com>
7375 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7377 2017-11-13 Carl Love <cel@us.ibm.com>
7379 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7380 Add support for builtins:
7381 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7382 vector {un,}signed {char,int,short},
7383 vector {un,}signed {char,int,short}) arguments.
7384 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7385 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7386 Add BU_P9V_AV_2 expansions for the builtins.
7387 * config/rs6000/altivec.h (vec_first_match_index,
7388 vec_first_mismatch_index, vec_first_match_or_eos_index,
7389 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7390 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7391 new extern declaration.
7392 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7393 * config/rs6000/vsx.md (first_match_index_<mode>,
7394 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7395 first_mismatch_or_eos_index_<mode>): Add define expand.
7396 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7397 * doc/extend.texi: Update the built-in documenation file for the new
7400 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7402 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7403 into the min/max operations for _Float<N> and _Float<N>X types.
7405 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7408 * dwarf2out.c (do_eh_frame): New static variable.
7409 (dwarf2out_begin_prologue): Set it.
7410 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7412 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7414 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7416 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7418 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7421 * debug.h (dwarf2out_do_eh_frame): Declare.
7422 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7423 (dwarf2out_do_frame): Use it.
7424 (dwarf2out_do_cfi_asm): Likewise.
7425 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7426 (dwarf2out_assembly_start): Likewise.
7427 (dwarf2out_begin_prologue): Fix comment.
7428 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7429 if the target needs either debug or unwind DWARF2 info.
7430 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7431 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7432 -ftrapping-math, -ftrapv and -fwrapv.
7434 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7436 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7437 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7438 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7439 (estimate_edge_size_and_time): Likewise.
7440 (ipa_merge_fn_summary_after_inlining): Likewise.
7441 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7442 (compute_uninlined_call_time): Use sreal_frequency.
7443 (compute_inlined_call_time): Likewise.
7444 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7445 * profile-count.c: Include sreal.h.
7446 (profile_count::to_sreal_scale): New.
7447 * profile-count.h: Forward declare sreal.
7448 (profile_count::to_sreal_scale): Declare.
7450 2017-11-13 Nathan Sidwell <nathan@acm.org>
7452 * diagnostic.c (maybe_line_and_column): New.
7453 (diagnostic_get_location_text): Use it.
7454 (diagnostic_report_current_module): Likewise.
7455 (test_diagnostic_get_location_text): Add tests.
7457 2017-11-13 Luis Machado <luis.machado@linaro.org>
7459 * doc/md.texi (Specifying processor pipeline description): Fix
7460 incorrect latency for the div instruction example.
7462 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7464 PR tree-optimization/78821
7465 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7466 that bit_not_p is the same.
7467 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7468 (split_group): Count precisely bit_not_p bits in each statement.
7469 (invert_op): New function.
7470 (imm_store_chain_info::output_merged_store): Use invert_op to
7471 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7472 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7474 2017-11-13 Martin Liska <mliska@suse.cz>
7476 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7477 (struct source_info): Likewise.
7478 (add_branch_counts): Likewise.
7479 (add_line_counts): Likewise.
7480 (function_summary): Likewise.
7481 (output_intermediate_line): Likewise.
7482 (generate_results): Likewise.
7484 2017-11-13 Martin Liska <mliska@suse.cz>
7486 * gcov.c (struct block_info): Remove typedef for block_t.
7487 (struct line_info): Likewise.
7488 (line_info::has_block): Likewise.
7489 (EXIT_BLOCK): Likewise.
7490 (unblock): Likewise.
7491 (circuit): Likewise.
7492 (get_cycles_count): Likewise.
7493 (process_file): Likewise.
7494 (read_graph_file): Likewise.
7495 (solve_flow_graph): Likewise.
7496 (find_exception_blocks): Likewise.
7497 (add_line_counts): Likewise.
7498 (accumulate_line_info): Likewise.
7499 (output_line_details): Likewise.
7501 2017-11-13 Martin Liska <mliska@suse.cz>
7503 * gcov.c (struct arc_info): Remove typedef for arc_t.
7504 (struct line_info): Likewise.
7505 (add_branch_counts): Likewise.
7506 (output_branch_count): Likewise.
7507 (function_info::~function_info): Likewise.
7508 (circuit): Likewise.
7509 (output_intermediate_line): Likewise.
7510 (read_graph_file): Likewise.
7511 (solve_flow_graph): Likewise.
7512 (find_exception_blocks): Likewise.
7513 (add_line_counts): Likewise.
7514 (accumulate_line_info): Likewise.
7515 (output_line_details): Likewise.
7516 (output_function_details): Likewise.
7518 2017-11-13 Martin Liska <mliska@suse.cz>
7520 * gcov.c (struct function_info): Remove typedef for function_t.
7521 (struct source_info): Likewise.
7522 (source_info::get_functions_at_location): Likewise.
7523 (solve_flow_graph): Likewise.
7524 (find_exception_blocks): Likewise.
7525 (add_line_counts): Likewise.
7526 (output_intermediate_file): Likewise.
7527 (process_file): Likewise.
7528 (generate_results): Likewise.
7529 (release_structures): Likewise.
7530 (read_graph_file): Likewise.
7531 (read_count_file): Likewise.
7532 (accumulate_line_counts): Likewise.
7533 (output_lines): Likewise.
7535 2017-11-13 Martin Liska <mliska@suse.cz>
7537 * gcov.c (function_info::function_info): Remove num_counts
7538 and add vector<gcov_type>.
7539 (function_info::~function_info): Use the vector.
7540 (process_file): Likewise.
7541 (read_graph_file): Likewise.
7542 (read_count_file): Likewise.
7543 (solve_flow_graph): Likewise.
7545 2017-11-13 Martin Liska <mliska@suse.cz>
7547 * gcov.c (function_info::is_artificial): New function.
7548 (process_file): Erase all artificial early.
7549 (generate_results): Skip as all artificial are already
7552 2017-11-13 Martin Liska <mliska@suse.cz>
7554 * gcov.c (read_graph_file): Store to global vector of functions.
7555 (read_count_file): Iterate the vector.
7556 (process_file): Likewise.
7557 (generate_results): Likewise.
7558 (release_structures): Likewise.
7560 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7562 PR tree-optimization/82954
7563 * gimple-ssa-store-merging.c
7564 (imm_store_chain_info::coalesce_immediate_stores): If
7565 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7567 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7569 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7570 Upddate call to ENDIAN_LANE_N.
7571 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7572 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7573 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7574 and use aarch64_endian_lane_rtx.
7575 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7577 2017-11-12 Tom de Vries <tom@codesourcery.com>
7579 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7581 2017-11-12 Tom de Vries <tom@codesourcery.com>
7583 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7585 2017-11-12 Tom de Vries <tom@codesourcery.com>
7587 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7589 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7590 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7591 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7593 2017-11-11 Martin Sebor <msebor@redhat.com>
7596 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7599 * prefic.c (translate_name): Replace strncpy with memcpy to
7600 avoid -Wstringop-truncation.
7602 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7604 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7606 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7608 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7610 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7612 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7615 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7617 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7618 enable generating XXBRH if the value is in a vector register.
7619 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7620 value is in a vector register.
7621 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7622 register to register bswap64's instead of doing the GPR sequence
7623 used on previous machines.
7624 (bswapdi2_xxbrd): New insn.
7625 (bswapdi2_reg): Disallow on ISA 3.0.
7626 (register to register bswap64 splitter): Do not split the insn on
7627 ISA 3.0 systems that use XXBRD.
7629 2017-11-10 Martin Sebor <msebor@redhat.com>
7632 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7634 (find_subframework_file): Same.
7636 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7638 * auto-profile.c (afdo_indirect_call): Drop frequency.
7639 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7640 (cgraph_node::create_edge): Drop frequency argument.
7641 (cgraph_node::create_indirect_edge): Drop frequency argument.
7642 (cgraph_edge::make_speculative): Drop frequency arguments.
7643 (cgraph_edge::resolve_speculation): Do not update frequencies
7644 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7645 (cgraph_node::dump): Check consistency in IPA mode.
7646 (cgraph_edge::maybe_hot_p): Use IPA counter.
7647 (cgraph_edge::verify_count_and_frequency): Rename to ...
7648 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7649 (cgraph_node::verify_node): Update.
7650 * cgraph.h (struct cgraph_edge): Drop frequency.
7651 (cgraph_edge::frequency): New function.
7652 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7654 (cgraph_edge::rebuild_edges): Likewise.
7655 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7656 (duplicate_thunk_for_node): Do not pass frequency.
7657 (cgraph_node::create_clone): Scale only counts.
7658 (cgraph_node::create_virtual_clone): Do not pass frequency.
7659 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7660 (cgraph_node::create_version_clone): Do not pass frequency.
7661 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7662 (cgraph_node::expand_thunk): Do not pass frequency.
7663 (cgraph_node::create_wrapper): Do not pass frequency.
7664 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7666 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7667 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7668 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7669 (gather_caller_stats): Use frequency function.
7670 (ipcp_cloning_candidate_p): Use frequency function.
7671 (ipcp_propagate_stage): Use frequency function.
7672 (get_info_about_necessary_edges): Use frequency function.
7673 (update_profiling_info): Update only IPA profile.
7674 (update_specialized_profile): Use frequency functoin.
7675 (perhaps_add_new_callers): Update only IPA profile.
7676 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7677 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7678 (dump_ipa_call_summary): Use frequency function.
7679 (estimate_edge_size_and_time): Use frequency function.
7680 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7681 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7682 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7683 (update_noncloned_counts): ... ths one; scale counts only.
7684 (clone_inlined_nodes): Do not scale frequency.
7685 (inline_call): Do not pass frequency.
7686 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7687 (compute_inlined_call_time): Use IPA profile.
7688 (want_inline_small_function_p): Use IPA profile.
7689 (want_inline_self_recursive_call_p): Use IPA profile.
7690 (edge_badness): Use IPA profile.
7691 (lookup_recursive_calls): Use IPA profile.
7692 (recursive_inlining): Do not pass frequency.
7693 (resolve_noninline_speculation): Do not update frequency.
7694 (inline_small_functions): Collect max of IPA profile.
7695 (dump_overall_stats): Dump IPA porfile.
7696 (dump_inline_stats): Dump IPA porfile.
7697 (ipa_inline): Collect IPA stats.
7698 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7699 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7700 (ipa_propagate_frequency): Use frequency function.
7701 (ipa_profile): Cleanup.
7702 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7703 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7704 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7705 (input_node): Do not stream frequency.
7706 (input_edge): Do not stream frequency.
7707 (merge_profile_summaries): Scale only IPA profiles.
7708 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7709 * predict.c (drop_profile): Do not recompute frequency.
7710 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7711 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7712 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
7713 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
7714 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
7715 * tree-inline.c (copy_bb): Do not scale frequency.
7716 (expand_call_inline): Do not scale frequency.
7717 (tree_function_versioning): Do not scale frequency.
7718 * ubsan.c (ubsan_create_edge): Do not pass frequency.
7720 2017-11-10 Julia Koval <julia.koval@intel.com>
7722 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
7723 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
7724 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
7725 (_mm256_maskz_gf2p8affine_epi64_epi8)
7726 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
7727 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
7728 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
7729 (__builtin_ia32_vgf2p8affineqb_v32qi)
7730 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
7731 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
7733 2017-11-10 Tamar Christina <tamar.christina@arm.com>
7737 (arm_option_override): Refactor.
7738 (arm_option_reconfigure_globals): New.
7739 (arm_options_perform_arch_sanity_checks): New.
7740 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7742 (arm_options_perform_arch_sanity_checks): Likewise
7744 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7746 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7747 (power9-qpmul): New.
7748 * rs6000/rs6000.md ("type" attr): Add qmul.
7749 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7750 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7751 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7753 2017-11-10 Martin Sebor <msebor@redhat.com>
7756 * builtins.c (compute_objsize): Handle arrays that
7757 compute_builtin_object_size likes to fail for. Make extern.
7758 * builtins.h (compute_objsize): Declare.
7759 (check_strncpy_sizes): New function.
7760 (expand_builtin_strncpy): Call check_strncpy_sizes.
7761 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7762 -Wstringop-truncation.
7763 (gimple_fold_builtin_strncat): Same.
7764 * gimple.c (gimple_build_call_from_tree): Set call location.
7765 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
7766 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
7767 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
7768 (handle_builtin_strlen): Use strlen_to_stridx.
7769 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
7771 Use strlen_to_stridx.
7772 (pass_strlen::execute): Release strlen_to_stridx.
7773 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7774 (-Wstringop-truncation): Document new option.
7776 2017-11-10 Martin Liska <mliska@suse.cz>
7778 PR gcov-profile/82702
7779 * gcov.c (main): Handle intermediate files in a different
7781 (get_gcov_intermediate_filename): New function.
7782 (output_gcov_file): Remove support of intermediate files.
7783 (generate_results): Allocate intermediate file.
7784 (release_structures): Clean-up properly fn_end.
7785 (output_intermediate_file): Start iterating with line 1.
7787 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7789 PR tree-optimization/82929
7790 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7791 ops_swapped_p non-static data member.
7792 (store_immediate_info::store_immediate_info): Clear it.
7793 (imm_store_chain_info::coalesce_immediate_stores): If swapping
7794 ops set ops_swapped_p.
7795 (count_multiple_uses): Handle ops_swapped_p.
7797 2017-11-10 Martin Liska <mliska@suse.cz>
7799 * coverage.c (coverage_init): Stream information about
7800 support of has_unexecuted_blocks.
7801 * doc/gcov.texi: Document that.
7802 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7803 * gcov.c (read_graph_file): Likewise.
7804 (output_line_beginning): Fix a small issue with
7807 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7809 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7810 reference of trivial component.
7812 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7815 * gimple-ssa-store-merging.c
7816 (pass_store_merging::terminate_all_aliasing_chains): For
7817 gimple_store_p stmts also call refs_output_dependent_p.
7819 PR rtl-optimization/82913
7820 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7823 2017-11-09 Jeff Law <law@redhat.com>
7825 * vr-values.h: New file with vr_values class.
7826 * tree-vrp.c: Include vr-values.h
7827 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7828 data objects into the vr_values class.
7829 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7830 (get_value_range): Make it a member function within vr_values class.
7831 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7832 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7833 (op_with_constant_singleton_value_range): Likewise.
7834 (extract_range_for_var_from_comparison_expr): Likewise.
7835 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7836 (extract_range_from_binary_expr): Likewise.
7837 (extract_range_from_unary_expr): Likewise.
7838 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7839 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7840 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7841 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7842 (compare_name_with_value, compare_names): Likewise.
7843 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7844 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7845 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7846 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7847 (extract_range_from_phi_node): Likewise.
7848 (simplify_truth_ops_using_ranges): Likewise.
7849 (simplify_div_or_mod_using_ranges): Likewise.
7850 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7851 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7852 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7853 (simplify_float_conversion_using_ranges): Likewise.
7854 (simplify_internal_call_using_ranges): Likewise.
7855 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7856 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7857 poitner via x_vr_values for calls into gimple folder.
7858 (vrp_initialize_lattice): Make this the vr_values ctor.
7859 (vrp_free_lattice): Make this the vr_values dtor.
7860 (set_vr_value): New function.
7861 (class vrp_prop): Add vr_values data member. Add various member
7862 functions as well as member functions that delegate to vr_values.
7863 (check_array_ref): Make a member function within vrp_prop class.
7864 (search_for_addr_array, vrp_initialize): Likewise.
7865 (vrp_finalize): Likewise. Revamp to avoid direct access to
7866 vr_value, values_propagated, etc.
7867 (check_array_bounds): Extract vrp_prop class instance pointer from
7868 walk info structure. Use it to call member functions.
7869 (check_all_array_refs): Make a member function within vrp_prop class.
7870 Smuggle class instance pointer via walk info structure.
7871 (x_vr_values): New local static.
7872 (vrp_valueize): Use x_vr_values to get class instance.
7873 (vr_valueize_1): Likewise.
7874 (class vrp_folder): Add vr_values data member. Add various member
7875 functions as well as member functions that delegate to vr_values.
7876 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7877 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7878 class instance from vr_values. Use it to call member functions.
7879 (vrp_dom_walker): Add vr_values data member.
7880 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7881 instance via x_vr_values.
7882 (identify_jump_threads): Accept vr_values as argument. Store
7883 it into the walker structure.
7884 (evrp_dom_walker): Add vr_values class data member. Add various
7886 (evrp_dom_walker::try_find_new_range): Use vr_values data
7887 member to access the memory allocator.
7888 (evrp_dom_walker::before_dom_children): Store vr_values class
7889 instance into the vrp_folder class.
7890 (evrp_dom_walker::push_value_range): Rework to avoid direct
7891 access to num_vr_values and vr_value.
7892 (evrp_dom_walker::pop_value_range): Likewise.
7893 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7894 Use vr_values to get to dump_all_value_ranges member function.
7895 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7896 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7897 Pass vr_values class instance down to identify_jump_threads.
7898 Remove call to vrp_free_lattice.
7899 (debug_all_value_ranges): Remove.
7901 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7902 (vrp_folder): Likewise.
7904 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7905 Get it from the existing bitmap instead.
7906 (vrp_intersect_ranges_1): Likewise.
7908 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7910 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7912 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7913 set bit_not_p to it.
7914 (imm_store_chain_info::coalesce_immediate_stores): Break group
7915 if bit_not_p is different.
7916 (count_multiple_uses, split_group,
7917 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7918 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7919 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7920 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7921 to store_immediate_info ctor.
7923 2017-11-09 Jim Wilson <jimw@sifive.com>
7925 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7926 (scan_prog_file): Likewise.
7928 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7930 * bb-reorder.c (max_entry_frequency): Remove.
7931 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7932 connect_traces, push_to_next_round_p): Remove prototypes.
7933 (find_traces_1_round): Use counts only.
7934 (push_to_next_round_p): Likewise.
7935 (find_traces): Likewise.
7936 (rotate_loop): Likewise.
7937 (find_traces_1_round): Likewise.
7938 (connect_traces): Likewise.
7939 (edge_order): Likewise.
7941 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7943 * config/arm/arm.c (output_return_instruction): Add comments to
7944 indicate requirement for cmse_nonsecure_entry return to account
7945 for the size of clearing instruction output here.
7946 (thumb_exit): Likewise.
7947 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7948 return in hardfloat mode.
7950 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7952 * config/rs6000/rs6000.c (machine_function): Add a bool,
7953 "toc_is_wrapped_separately".
7954 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7955 if it wasn't explicitly set or unset, we are optimizing for speed, and
7956 doing separate shrink-wrapping.
7957 (rs6000_get_separate_components): Enable the TOC component if
7958 saving the TOC register in the prologue.
7959 (rs6000_components_for_bb): Handle the TOC component.
7960 (rs6000_emit_prologue_components): Store the TOC register where needed.
7961 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7962 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7964 2017-11-09 Martin Jambor <mjambor@suse.cz>
7966 * ipa-param-manipulation.c: New file.
7967 * ipa-param-manipulation.h: Likewise.
7968 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7969 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7970 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7971 (ipa_parm_adjustment): Likewise.
7972 (ipa_parm_adjustment_vec): Likewise.
7973 (ipa_get_vector_of_formal_parms): Moved declaration to
7974 ipa-param-manipulation.h.
7975 (ipa_get_vector_of_formal_parm_types): Likewise.
7976 (ipa_modify_formal_parameters): Likewise.
7977 (ipa_modify_call_arguments): Likewise.
7978 (ipa_combine_adjustments): Likewise.
7979 (ipa_dump_param_adjustments): Likewise.
7980 (ipa_modify_expr): Likewise.
7981 (ipa_get_adjustment_candidate): Likewise.
7982 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7983 ipa-param-manipulation.c.
7984 (ipa_get_vector_of_formal_parm_types): Likewise.
7985 (ipa_modify_formal_parameters): Likewise.
7986 (ipa_modify_call_arguments): Likewise.
7987 (ipa_modify_expr): Likewise.
7988 (get_ssa_base_param): Likewise.
7989 (ipa_get_adjustment_candidate): Likewise.
7990 (index_in_adjustments_multiple_times_p): Likewise.
7991 (ipa_combine_adjustments): Likewise.
7992 (ipa_dump_param_adjustments): Likewise.
7993 * tree-sra.c: Also include ipa-param-manipulation.h
7994 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7997 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7998 Alan Hayward <alan.hayward@arm.com>
7999 David Sherwood <david.sherwood@arm.com>
8001 * doc/sourcebuild.texi (vect_masked_store): Document.
8003 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8004 Alan Hayward <alan.hayward@arm.com>
8005 David Sherwood <david.sherwood@arm.com>
8007 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8009 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8010 Alan Hayward <alan.hayward@arm.com>
8011 David Sherwood <david.sherwood@arm.com>
8013 * doc/sourcebuild.texi (vect_variable_length): Document.
8015 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8016 Alan Hayward <alan.hayward@arm.com>
8017 David Sherwood <david.sherwood@arm.com>
8019 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8021 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8022 Alan Hayward <alan.hayward@arm.com>
8023 David Sherwood <david.sherwood@arm.com>
8025 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8027 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8028 Alan Hayward <alan.hayward@arm.com>
8029 David Sherwood <david.sherwood@arm.com>
8031 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8032 previously undocumented selectors.
8033 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8035 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8037 * doc/rtl.texi (const_vector): Say that elements can be
8038 const_wide_ints too.
8039 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8040 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8041 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8042 * optabs.c (expand_vector_broadcast): Likewise.
8044 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8045 Alan Hayward <alan.hayward@arm.com>
8046 David Sherwood <david.sherwood@arm.com>
8048 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8049 scaled index even if the unscaled address was invalid.
8050 Don't increase the complexity of using a scale in that case.
8052 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8053 Alan Hayward <alan.hayward@arm.com>
8054 David Sherwood <david.sherwood@arm.com>
8056 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8057 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8058 UNITS_PER_WORD bytes.
8059 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8060 based on REGMODE_NATURAL_SIZE of the inner mode.
8061 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8062 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8063 * expmed.c (lowpart_bit_field_p): Divide the value up into
8064 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8065 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8066 whether something is likely to occupy more than one register.
8068 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8071 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8072 profile_count::adjust_for_ipa_scaling.
8073 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8074 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8076 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8078 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8080 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8081 (split_group): Add total_orig and total_new arguments, estimate the
8082 number of statements related to the store group without store merging
8083 and with store merging.
8084 (imm_store_chain_info::output_merged_store): Adjust split_group
8085 callers, punt if estimated number of statements with store merging
8086 is not smaller than estimated number of statements without it.
8088 (handled_load): Remove has_single_use checks.
8089 (pass_store_merging::process_store): Likewise.
8091 2017-11-09 Richard Biener <rguenther@suse.de>
8093 PR tree-optimization/82902
8094 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8096 2017-11-09 Martin Liska <mliska@suse.cz>
8099 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8102 2017-11-09 Martin Liska <mliska@suse.cz>
8104 PR tree-optimization/82669
8105 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8107 2017-11-09 Martin Liska <mliska@suse.cz>
8109 PR gcov-profile/48463
8110 * coverage.c (coverage_begin_function): Output also end locus
8111 of a function and information whether the function is
8113 * gcov-dump.c (tag_function): Parse and print the information.
8114 * gcov.c (INCLUDE_MAP): Add include.
8115 (INCLUDE_SET): Likewise.
8116 (struct line_info): Move earlier in the source file because
8117 of vector<line_info> in function_info structure.
8118 (line_info::line_info): Likewise.
8119 (line_info::has_block): Likewise.
8120 (struct source_info): Add new member index.
8121 (source_info::get_functions_at_location): New function.
8122 (function_info::group_line_p): New function.
8123 (output_intermediate_line): New function.
8124 (output_intermediate_file): Use the mentioned function.
8125 (struct function_start): New.
8126 (struct function_start_pair_hash): Likewise.
8127 (process_file): Add code that identifies group functions.
8128 Assign lines either to global or function scope.
8129 (generate_results): Skip artificial functions.
8130 (find_source): Assign index for each source file.
8131 (read_graph_file): Read new flag artificial and end_line.
8132 (add_line_counts): Assign it either to global of function scope.
8133 (accumulate_line_counts): Isolate core of the function to
8134 accumulate_line_info and call it for both function and global
8136 (accumulate_line_info): New function.
8137 (output_line_beginning): Fix GNU coding style.
8138 (print_source_line): New function.
8139 (output_line_details): Likewise.
8140 (output_function_details): Likewise.
8141 (output_lines): Iterate both source (global) scope and function
8143 (struct function_line_start_cmp): New class.
8144 * doc/gcov.texi: Reflect changes in documentation.
8146 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8149 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8150 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8151 and similarly for not instead of neg.
8153 2017-11-08 Andi Kleen <ak@linux.intel.com>
8155 * config/i386/i386.opt: Add -mforce-indirect-call.
8156 * config/i386/predicates.md: Check for flag_force_indirect_call.
8157 * doc/invoke.texi: Document -mforce-indirect-call
8159 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8161 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8163 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8164 config/riscv/riscv.c (predict.h): New include.
8165 (riscv_slow_unaligned_access_p): No longer static.
8166 (riscv_block_move_straight): Add require.
8167 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8169 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8172 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8173 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8174 nonimmediate_operand predicate for operand 1 instead of
8177 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8179 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8181 * config/aarch64/constraints.md (Uml): New constraint.
8182 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8185 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8187 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8188 of two vec_duplicates into a vec_concat.
8190 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8192 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8193 Simplify vec_merge of vec_duplicate and vec_concat.
8194 * config/aarch64/constraints.md (Utq): New constraint.
8195 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8198 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8200 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8201 Simplify vec_merge of vec_duplicate and const_vector.
8202 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8204 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8205 mode iterator. Update predicate on operand 1 to
8206 handle non-const_vec constants. Delete constraints.
8207 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8209 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8211 PR tree-optimization/78821
8212 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8214 (store_operand_info::store_operand_info): Initialize it to false.
8215 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8216 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8217 store in the group, and if chain_info is non-NULL, to ignore altogether
8219 (compatible_load_p): Fail if bit_not_p does not match.
8220 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8221 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8222 (pass_store_merging::process_store): Adjust
8223 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8224 call terminate_all_aliasing_chains newly when adding a store into
8225 a chain with non-NULL chain_info.
8227 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8229 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8231 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8233 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8235 (aarch64_layout_frame): Initialise emit_frame_chain.
8236 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8237 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8239 2017-11-08 Martin Liska <mliska@suse.cz>
8241 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8242 of gimple_call_internal_p.
8244 2017-11-07 Tom de Vries <tom@codesourcery.com>
8246 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8248 2017-11-07 Tom de Vries <tom@codesourcery.com>
8250 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8253 2017-11-08 Martin Liska <mliska@suse.cz>
8256 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8258 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8260 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8262 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8265 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8266 (*cmp<mode>_ccz_1): New insn with $k alternative.
8269 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8270 RTX_COMM_COMPARE as commutative as well.
8271 (ix86_binary_operator_ok): Formatting fix.
8272 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8273 *<code><mode>3<mask_name><round_saeonly_name>,
8274 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8275 *<s>mul<mode>3_highpart<mask_name>,
8276 *vec_widen_umult_even_v16si<mask_name>,
8277 *vec_widen_umult_even_v8si<mask_name>,
8278 *vec_widen_umult_even_v4si<mask_name>,
8279 *vec_widen_smult_even_v16si<mask_name>,
8280 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8281 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8282 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8283 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8284 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8285 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8286 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8287 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8288 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8289 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8290 ix86_binary_operator_ok. Formatting fixes.
8291 (*<plusminus_insn><mode>3<mask_name><round_name>,
8292 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8295 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8297 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8298 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8299 eq and ne if TARGET_ISEL.
8300 (cmp): New code_iterator.
8301 (UNS, UNSU_, UNSIK): New code_attrs.
8302 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8303 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8305 ("eq<mode>3"): ... this.
8306 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8308 ("ne<mode>3"): ... this.
8310 2017-11-07 Julia Koval <julia.koval@intel.com>
8313 * common/config/i386/i386-common.c
8314 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8315 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8316 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8317 * config/i386/i386.opt: Ditto.
8318 * config/i386/i386.c (ix86_target_string): Ditto.
8319 (ix86_option_override_internal): Ditto.
8320 (ix86_init_mpx_builtins): Move MPX to args2.
8321 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8322 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8323 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8324 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8325 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8326 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8327 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8329 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8332 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8333 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8334 (zero-extendsidi peephole2): Remove peephole.
8336 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8339 * stor-layout.c: Include attribs.h.
8340 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8341 explicit lookup of "aligned" attribute.
8343 2017-11-07 Andrew Waterman <andrew@sifive.com>
8345 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8346 (riscv_expand_block_move): Likewise.
8347 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8349 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8350 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8351 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8352 (riscv_adjust_block_mem): Likewise.
8353 (riscv_block_move_loop): Likewise.
8354 (riscv_expand_block_move): Likewise.
8355 * config/riscv/riscv.md (movmemsi): New pattern.
8357 2017-11-07 Michael Clark <michaeljclark@mac.com>
8359 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8360 (MUSL_DYNAMIC_LINKER): Likewise.
8362 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8364 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8365 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8368 2017-11-07 Richard Biener <rguenther@suse.de>
8370 * match.pd: Fix build.
8372 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8373 Jackson Woodruff <jackson.woodruff@arm.com>
8375 PR tree-optimization/71026
8376 * match.pd: Canonicalize negate in division.
8378 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8381 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8383 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8385 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8386 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8388 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8390 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8391 non-scalar integral types.
8392 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8393 (-(A-B), -(~A)): New transformations.
8395 2017-11-07 Tom de Vries <tom@codesourcery.com>
8397 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8398 semicolon after "do {} while (0)".
8399 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8400 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8401 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8402 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8403 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8405 2017-11-07 Tom de Vries <tom@codesourcery.com>
8407 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8408 after "do {} while (0)".
8409 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8410 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8411 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8412 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8413 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8415 2017-11-07 Tom de Vries <tom@codesourcery.com>
8418 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8420 (arm_rtx_costs_internal): Add missing semicolon after
8421 HANDLE_NARROW_SHIFT_ARITH call.
8423 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8425 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8426 disable isel if it was not set explicitly.
8428 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8430 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8431 (add_type_attribute) likewise.
8433 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8435 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8436 of struct ix86_frame.
8437 (ix86_initial_elimination_offset): Likewise.
8438 (ix86_expand_split_stack_prologue): Likewise.
8440 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8442 * tree-vrp.h (enum value_range_type): Update stale comment.
8444 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8445 Alan Hayward <alan.hayward@arm.com>
8446 David Sherwood <david.sherwood@arm.com>
8448 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8449 (aarch64_expand_vec_perm_const): Take the number of units too.
8450 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8451 (aarch64_expand_vec_perm_const): Likewise.
8452 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8453 (vec_perm<mode>): Update accordingly.
8455 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8456 Alan Hayward <alan.hayward@arm.com>
8457 David Sherwood <david.sherwood@arm.com>
8459 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8460 Take the number of units too.
8461 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8462 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8463 but check for a vector mode before rather than after the call.
8464 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8465 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8466 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8467 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8468 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8469 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8470 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8471 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8472 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8473 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8474 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8475 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8476 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8477 (aarch64_sqdmull2_n<mode>): Update accordingly.
8479 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8480 Alan Hayward <alan.hayward@arm.com>
8481 David Sherwood <david.sherwood@arm.com>
8483 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8484 the number of units too.
8485 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8486 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8487 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8488 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8489 (vec_store_lanesxi<mode>): Update accordingly.
8491 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8492 Alan Hayward <alan.hayward@arm.com>
8493 David Sherwood <david.sherwood@arm.com>
8495 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8496 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8497 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8498 of units rather than the mode.
8499 * config/aarch64/iterators.md (nunits): New mode attribute.
8500 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8501 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8502 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8503 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8504 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8505 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8506 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8507 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8508 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8509 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8510 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8511 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8512 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8513 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8514 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8515 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8516 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8517 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8518 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8519 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8520 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8521 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8522 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8523 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8524 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8525 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8526 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8527 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8528 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8529 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8530 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8531 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8532 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8533 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8534 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8535 (aarch64_simd_vec_setv2di): Likewise.
8537 2017-11-06 Carl Love <cel@us.ibm.com>
8539 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8541 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8542 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8544 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8546 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8547 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8548 Add power 8 macro expansions.
8549 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8550 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8551 power 8 instructions. (VSX_XXBR): Add iterator.
8553 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8555 * config/arm/arm.md (predicable_short_it): Change default to "no",
8556 improve documentation, remove uses that are identical to the default.
8557 (enabled_for_depr_it): Rename to enabled_for_short_it.
8558 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8559 Remove default uses.
8560 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8561 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8562 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8563 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8565 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8568 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8569 float128 helper macros, which are no longer used after deleting
8570 the old 'q' built-in functions, and moving the round to odd
8571 built-in functions to being special built-in functions.
8572 (BU_FLOAT128_2): Likewise.
8573 (BU_FLOAT128_1_HW): Likewise.
8574 (BU_FLOAT128_2_HW): Likewise.
8575 (BU_FLOAT128_3_HW): Likewise.
8576 (FABSQ): Delete old 'q' built-in functions.
8577 (COPYSIGNQ): Likewise.
8578 (SQRTF128_ODD): Move round to odd built-in functions to be
8579 special built-in functions, so that we can handle
8580 -mabi=ieeelongdouble.
8581 (TRUNCF128_ODD): Likewise.
8582 (ADDF128_ODD): Likewise.
8583 (SUBF128_ODD): Likewise.
8584 (MULF128_ODD): Likewise.
8585 (DIVF128_ODD): Likewise.
8586 (FMAF128_ODD): Likewise.
8587 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8588 built-in names to 'f128'.
8589 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8590 old 'q' built-in functions, as the machine independent code for
8591 'f128' built-in functions handles this.
8592 (rs6000_expand_builtin): Add expansion for float128 round to odd
8593 functions, keying off on -mabi=ieeelongdouble of whether to use
8594 the KFmode or TFmode variant.
8595 (rs6000_init_builtins): Initialize the _Float128 round to odd
8597 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8598 _Float128 'q' built-in functions are now mapped into the new
8599 'f128' built-in functions.
8601 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8603 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8604 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8606 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8609 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8610 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8611 * toplev.c: Include "ipa-fnsummary.h".
8612 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8614 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8616 PR tree-optimization/82838
8617 * gimple-ssa-store-merging.c
8618 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8619 on a separate gimple_seq which is then appended to seq.
8621 2017-11-06 Jeff Law <law@redhat.com>
8624 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8625 (get_probe_interval): New functions.
8626 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8627 (ix86_adjust_stack_and_probe): Likewise.
8628 (output_adjust_stack_and_probe): Likewise.
8629 (ix86_emit_probe_stack_range): Likewise.
8630 (ix86_expand_prologue): Likewise.
8632 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8634 PR tree-optimization/82816
8635 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8636 if the modes of the two types are the same.
8637 (convert_plusminus_to_widen): Likewise.
8639 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8641 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8643 (usadv16qi): New define_expand.
8644 (usadv8hi): New define_expand.
8646 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8649 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8651 (inline_transform): Likewise.
8652 * predict.c (counts_to_freqs): Remove useless conditional.
8653 * profile-count.h (profile_count::apply_scale): Move sanity check.
8654 * tree-inline.c (copy_bb): Always scale.
8655 (copy_cfg_body): Likewise.
8657 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8660 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8662 (*cmp_ite0): Add enabled_for_depr_it attribute.
8663 (*cmp_ite1): Likewise.
8665 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8667 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8670 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8672 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8673 return the value by pointer.
8674 (extract_range_from_multiplicative_op_1): Update accordingly.
8675 Return as soon as an operation fails.
8677 2017-11-05 Tom de Vries <tom@codesourcery.com>
8680 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8681 (DEF_SANITIZER_BUILTIN): ... here.
8682 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8683 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8685 2017-11-05 Tom de Vries <tom@codesourcery.com>
8688 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8690 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8691 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8692 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8694 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8695 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8696 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8698 2017-11-05 Tom de Vries <tom@codesourcery.com>
8701 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8704 2017-11-04 Michael Clark <michaeljclark@mac.com>
8706 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8707 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8709 (*addsi3_extended): Likewise.
8710 (*addsi3_extended2): Likewise.
8711 (<optab>si3): Likewise.
8712 (<optab>di3): Likewise.
8713 (<optab><mode>3): Likewise.
8714 (<*optabe>si3_internal): Likewise.
8715 (zero_extendqi<SUPERQI:mode>2): Likewise.
8716 (*add<mode>hi3): Likewise.
8717 (*xor<mode>hi3): Likewise.
8718 (<optab>di3): Likewise.
8719 (*<optab>si3_extend): Likewise.
8720 (*sge<u>_<X:mode><GPR:mode>): Likewise.
8721 (*slt<u>_<X:mode><GPR:mode>): Likewise.
8722 (*sle<u>_<X:mode><GPR:mode>): Likewise.
8724 2017-11-04 Andrew Waterman <andrew@sifive.com>
8726 * config/riscv/riscv.c (riscv_option_override): Conditionally set
8727 TARGET_STRICT_ALIGN based upon -mtune argument.
8729 2017-11-04 Andrew Waterman <andrew@sifive.com>
8731 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
8733 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
8735 * config/i386/i386.c (choose_basereg): Use optional scratch
8736 register and add assertion.
8737 (x86_emit_outlined_ms2sysv_save): Use scratch register when
8738 needed, and don't allocate stack.
8739 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
8740 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8741 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8743 2017-11-03 Jeff Law <law@redhat.com>
8745 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8746 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8747 to probe at the start of a noreturn function.
8749 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8751 PR tree-optimization/78821
8752 * gimple-ssa-store-merging.c: Update the file comment.
8753 (MAX_STORE_ALIAS_CHECKS): Define.
8754 (struct store_operand_info): New type.
8755 (store_operand_info::store_operand_info): New constructor.
8756 (struct store_immediate_info): Add rhs_code and ops data members.
8757 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8758 and op1r arguments to the ctor, initialize corresponding data members.
8759 (struct merged_store_group): Add load_align_base and load_align
8761 (merged_store_group::merged_store_group): Initialize them.
8762 (merged_store_group::do_merge): Update them.
8763 (merged_store_group::apply_stores): Pick the constant for
8764 encode_tree_to_bitpos from one of the two operands, or skip
8765 encode_tree_to_bitpos if neither operand is a constant.
8766 (class pass_store_merging): Add process_store method decl. Remove
8767 bool argument from terminate_all_aliasing_chains method decl.
8768 (pass_store_merging::terminate_all_aliasing_chains): Remove
8769 var_offset_p argument and corresponding handling.
8770 (stmts_may_clobber_ref_p): New function.
8771 (compatible_load_p): New function.
8772 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
8773 if there is overlap and rhs_code is not INTEGER_CST. For
8774 non-overlapping stores terminate group if rhs is not mergeable.
8775 (get_alias_type_for_stmts): Change first argument from
8776 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8777 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8778 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
8780 (get_location_for_stmts): Change first argument from
8781 auto_vec<gimple *> & to vec<gimple *> &.
8782 (struct split_store): Remove orig_stmts data member, add orig_stores.
8783 (split_store::split_store): Create orig_stores rather than orig_stmts.
8784 (find_constituent_stmts): Renamed to ...
8785 (find_constituent_stores): ... this. Change second argument from
8786 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
8787 to info structures rather than the statements.
8788 (split_group): Rename ALLOW_UNALIGNED argument to
8789 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
8790 it. Adjust find_constituent_stores caller.
8791 (imm_store_chain_info::output_merged_store): Handle rhs_code other
8792 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
8793 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
8794 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
8795 (mem_valid_for_store_merging): New function.
8796 (handled_load): New function.
8797 (pass_store_merging::process_store): New method.
8798 (pass_store_merging::execute): Use process_store method. Adjust
8799 terminate_all_aliasing_chains caller.
8801 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8803 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8804 Return true for more constants, symbols and label references.
8805 (aarch64_valid_floating_const): Remove unused function.
8807 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8810 * config/aarch64/aarch64.c (aarch64_layout_frame):
8811 Undo forcing of LR at bottom of frame.
8813 2017-11-03 Jeff Law <law@redhat.com>
8816 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8817 for int_registers_saved.
8819 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8820 extracted from tree-ssa-dom.c.
8821 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8822 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8823 (record_equivalences_from_incoming_edge): Add additional argument
8824 to single_pred_edge_ignoring_loop_edges call.
8825 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8826 (uncprop_dom_walker::before_dom_children): Add additional argument
8827 to single_pred_edge_ignoring_loop_edges call.
8828 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8829 single_pred_edge_ignoring_loop_edges rather than open coding.
8830 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8832 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8834 * match.pd (-(-A)): Rewrite.
8836 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8838 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8839 (rs6000_emit_int_cmove): New declaration.
8840 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8841 (rs6000_emit_sISEL): Delete.
8842 (rs6000_emit_int_cmove): Make non-static.
8843 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8844 instead of rs6000_emit_sISEL.
8846 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8848 * asan.c (create_cond_insert_point): Maintain profile.
8849 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8851 * basic-block.h (struct basic_block_def): Remove frequency.
8852 (EDGE_FREQUENCY): Use to_frequency
8853 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8855 (find_traces): Update to use to_frequency.
8856 (find_traces_1_round): Likewise; use only IPA counts.
8857 (bb_to_key): Likewise.
8858 (connect_traces): Use IPA counts only.
8859 (copy_bb_p): Update to use to_frequency.
8860 (fix_up_crossing_landing_pad): Likewise.
8861 (sanitize_hot_paths): Likewise.
8862 * bt-load.c (basic_block_freq): Likewise.
8863 * cfg.c (init_flow): Set count_max to uninitialized.
8864 (check_bb_profile): Remove frequencies; check counts.
8865 (dump_bb_info): Do not dump frequencies.
8866 (update_bb_profile_for_threading): Update counts only.
8867 (scale_bbs_frequencies_int): Likewise.
8868 (MAX_SAFE_MULTIPLIER): Remove.
8869 (scale_bbs_frequencies_gcov_type): Update counts only.
8870 (scale_bbs_frequencies_profile_count): Update counts only.
8871 (scale_bbs_frequencies): Update counts only.
8872 * cfg.h (struct control_flow_graph): Add count-max.
8873 (update_bb_profile_for_threading): Update prototype.
8874 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8875 (find_many_sub_basic_blocks): Likewise.
8876 * cfgcleanup.c (try_forward_edges): Likewise.
8877 (try_crossjump_to_edge): Likewise.
8878 * cfgexpand.c (expand_gimple_cond): Likewise.
8879 (expand_gimple_tailcall): Likewise.
8880 (construct_init_block): Likewise.
8881 (construct_exit_block): Likewise.
8882 * cfghooks.c (verify_flow_info): Check consistency of counts.
8883 (dump_bb_for_graph): Do not dump frequencies.
8884 (split_block_1): Do not update frequencies.
8885 (split_edge): Do not update frequencies.
8886 (make_forwarder_block): Do not update frequencies.
8887 (duplicate_block): Do not update frequencies.
8888 (account_profile_record): Do not update frequencies.
8889 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8890 for global heuristics.
8891 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8892 (expected_loop_iterations_unbounded): Use counts only.
8893 * cfgloopmanip.c (scale_loop_profile): Simplify.
8894 (create_empty_loop_on_edge): Simplify
8896 (duplicate_loop_to_header_edge): Simplify
8897 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8898 (update_br_prob_note): Take care of removing note when profile
8900 (relink_block_chain): Do not dump frequency.
8901 (rtl_account_profile_record): Use to_frequency.
8902 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8903 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8904 (cgraph_update_edges_for_call_stmt_node): Likewise.
8905 (cgraph_edge::verify_count_and_frequency): Update.
8906 (cgraph_node::verify_node): Temporarily disable frequency verification.
8907 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8908 to_cgraph_frequency.
8909 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8910 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8912 (cgraph_node::expand_thunk): Update profile.
8913 * except.c (dw2_build_landing_pads): Do not update frequency.
8914 * final.c (compute_alignments): Use to_frequency.
8915 (dump_basic_block_info): Do not dump frequency.
8916 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8917 (dump_gimple_bb_header): Do not dump frequency.
8918 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8920 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8921 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8922 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8923 (init_before_recovery): Do not update frequency.
8924 (sched_create_recovery_edges): Do not update frequency.
8925 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8926 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8927 (ipa_cp_c_finalize): Set max_count to uninitialized.
8928 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8929 (param_change_prob): Use counts.
8930 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8932 * ipa-split.c (consider_split): Use to_frequency.
8933 (split_function): Use to_frequency.
8934 * ira-build.c (loop_compare_func): Likewise.
8935 (mark_loops_for_removal): Likewise.
8936 (mark_all_loops_for_removal): Likewise.
8937 * loop-doloop.c (doloop_modify): Do not update frequency.
8938 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8940 * lto-streamer-in.c (input_function): Update count_max.
8941 * omp-expand.c (expand_omp_taskreg): Update count_max.
8942 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8943 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8944 (maybe_hot_count_p): Use ipa counts only.
8945 (maybe_hot_bb_p): Simplify.
8946 (maybe_hot_edge_p): Simplify.
8947 (probably_never_executed): Do not take frequency argument.
8948 (probably_never_executed_bb_p): Do not pass frequency.
8949 (probably_never_executed_edge_p): Likewise.
8950 (combine_predictions_for_bb): Check that profile is nonzero.
8951 (propagate_freq): Do not set frequency.
8952 (drop_profile): Simplify.
8953 (counts_to_freqs): Simplify.
8954 (expensive_function_p): Use to_frequency.
8955 (propagate_unlikely_bbs_forward): Simplify.
8956 (determine_unlikely_bbs): Simplify.
8957 (estimate_bb_frequencies): Add hack to silence graphite issues.
8958 (compute_function_frequency): Use ipa counts.
8959 (pass_profile::execute): Update.
8960 (rebuild_frequencies): Use counts only.
8961 (force_edge_cold): Use counts only.
8962 * profile-count.c (profile_count::dump): Dump new count types.
8963 (profile_count::differs_from_p): Check compatiblity.
8964 (profile_count::to_frequency): New function.
8965 (profile_count::to_cgraph_frequency): New function.
8966 * profile-count.h (struct function): Declare.
8967 (enum profile_quality): Add profile_guessed_local and
8968 profile_guessed_global0.
8969 (class profile_proability): Decrease number of bits to 29;
8970 update from_reg_br_prob_note and to_reg_br_prob_note.
8971 (class profile_count: Update comment; decrease number of bits
8972 to 61. Check compatibility.
8973 (profile_count::compatible_p): New private member function.
8974 (profile_count::ipa_p): New member function.
8975 (profile_count::operator<): Handle global zero correctly.
8976 (profile_count::operator>): Handle global zero correctly.
8977 (profile_count::operator<=): Handle global zero correctly.
8978 (profile_count::operator>=): Handle global zero correctly.
8979 (profile_count::nonzero_p): New member function.
8980 (profile_count::force_nonzero): New member function.
8981 (profile_count::max): New member function.
8982 (profile_count::apply_scale): Handle IPA scalling.
8983 (profile_count::guessed_local): New member function.
8984 (profile_count::global0): New member function.
8985 (profile_count::ipa): New member function.
8986 (profile_count::to_frequency): Declare.
8987 (profile_count::to_cgraph_frequency): Declare.
8988 * profile.c (OVERLAP_BASE): Delete.
8989 (compute_frequency_overlap): Delete.
8990 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8991 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8992 * sched-ebb.c (rank): Use counts only.
8993 * shrink-wrap.c (handle_simple_exit): Use counts only.
8994 (try_shrink_wrapping): Use counts only.
8995 (place_prologue_for_one_component): Use counts only.
8996 * tracer.c (find_best_predecessor): Use to_frequency.
8997 (find_trace): Use to_frequency.
8998 (tail_duplicate): Use to_frequency.
8999 * trans-mem.c (expand_transaction): Do not update frequency.
9000 * tree-call-cdce.c: Do not update frequency.
9001 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9002 (gimple_merge_blocks): Likewise.
9003 (gimple_split_edge): Likewise.
9004 (gimple_duplicate_sese_region): Likewise.
9005 (gimple_duplicate_sese_tail): Likewise.
9006 (move_sese_region_to_fn): Likewise.
9007 (gimple_account_profile_record): Likewise.
9008 (insert_cond_bb): Likewise.
9009 * tree-complex.c (expand_complex_div_wide): Likewise.
9010 * tree-eh.c (lower_resx): Update profile.
9011 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9013 (initialize_cfun): Do not initialize frequencies
9014 (freqs_to_counts): Delete.
9015 (copy_cfg_body): Ignore count parameter.
9016 (copy_body): Update.
9017 (expand_call_inline): Update count_max.
9018 (optimize_inline_calls): Update count_max.
9019 (tree_function_versioning): Update count_max.
9020 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9021 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9023 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9024 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9025 (try_peel_loop): Likewise.
9026 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9028 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9029 (tree_transform_and_unroll_loop): Do not use frequencies
9030 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9031 Use reliable prediction only.
9032 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9033 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9034 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9035 probability scaling.
9036 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9037 not update frequency
9038 (any_remaining_duplicated_blocks): Likewise.
9039 (update_profile): Likewise.
9040 (estimated_freqs_path): Delete.
9041 (freqs_to_counts_path): Delete.
9042 (clear_counts_path): Delete.
9043 (ssa_fix_duplicate_block_edges): Likewise.
9044 (duplicate_thread_path): Likewise.
9045 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9046 * tree-tailcall.c (decrease_profile): Do not update frequency.
9047 (eliminate_tail_call): Likewise.
9048 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9049 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9050 (optimize_mask_stores): Likewise.
9051 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9052 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9053 (ubsan_expand_ptr_ifn): Update profile.
9054 * value-prof.c (gimple_ic): Simplify.
9055 * value-prof.h (gimple_ic): Update prototype.
9056 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9057 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9059 (want_inline_self_recursive_call_p): Likewise.
9060 (resolve_noninline_speculation): Only cummulate defined counts.
9061 (inline_small_functions): Use nonzero_p.
9062 (ipa_inline): Do not access freed node.
9064 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9066 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9067 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9069 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9071 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9072 non-legitimate address.
9074 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9076 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9077 (*lt0_<mode>di, *lt0_<mode>si): New.
9079 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9081 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9082 TARGET_PAIRED_FLOAT.
9084 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9085 Jim Wilson <jim.wilson@linaro.org>
9087 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9088 * config/aarch64/aarch64-tune.md: Regenerated.
9089 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9090 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9092 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9094 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9095 (arc_expand_prologue): Restore blink for millicode.
9096 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9098 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9101 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9102 gen_vec_duplicate after forcing the scalar into a register.
9104 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9106 * combine (try_combine): Print the insns input to try_combine to the
9109 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9112 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9113 Remove second argument from aarch64_process_target_attr call.
9114 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9116 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9118 (aarch64_handle_attr_arch): Remove second argument.
9119 (aarch64_handle_attr_cpu): Ditto.
9120 (aarch64_handle_attr_tune): Ditto.
9121 (aarch64_handle_attr_isa_flags): Ditto.
9122 (aarch64_process_one_target_attr): Ditto.
9123 (aarch64_process_target_attr): Ditto.
9124 (aarch64_option_valid_attribute_p): Remove second argument.
9125 on aarch64_process_target_attr call.
9127 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9129 * diagnostic.c: Include "selftest-diagnostic.h".
9130 (selftest::assert_location_text): New function.
9131 (selftest::test_diagnostic_get_location_text): New function.
9132 (selftest::diagnostic_c_tests): Call it.
9134 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9136 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9137 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9138 (class selftest::test_diagnostic_context): Move to...
9139 * selftest-diagnostic.c: New file.
9140 * selftest-diagnostic.h: New file.
9142 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9144 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9145 offset range for FT32B.
9146 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9147 * config/ft32/ft32.md: Add TARGET_NOPM.
9148 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9149 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9151 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9153 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9155 2017-11-02 Jeff Law <law@redhat.com>
9157 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9158 virtual keyword on FINAL OVERRIDE members.
9160 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9161 virtuals together. Add virtual destructor.
9162 (substitute_and_fold_engine): Similarly.
9164 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9166 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9168 2017-11-02 Richard Biener <rguenther@suse.de>
9170 PR tree-optimization/82795
9171 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9173 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9175 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9177 (gcc_GAS_CHECK_FEATURE): Remove.
9178 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9179 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9180 * configure: Regenerate.
9182 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9184 * config/arc/arc.c (hwloop_optimize): Account for empty
9187 2017-11-02 Richard Biener <rguenther@suse.de>
9190 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9191 Truncate ARRAY_REF index and element size.
9193 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9195 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9197 2017-11-01 Jeff Law <law@redhat.com>
9199 * tree-ssa-ccp.c (ccp_folder): New class derived from
9200 substitute_and_fold_engine.
9201 (ccp_folder::get_value): New member function.
9202 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9203 (ccp_fold_stmt): Remove prototype.
9204 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9205 * tree-ssa-copy.c (copy_folder): New class derived from
9206 substitute_and_fold_engine.
9207 (copy_folder::get_value): Renamed from get_value.
9208 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9209 * tree-vrp.c (vrp_folder): New class derived from
9210 substitute_and_fold_engine.
9211 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9212 (vrp_folder::get_value): New member function.
9213 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9214 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9215 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9216 provide a class interface to folder/substitute routines.
9217 (ssa_prop_fold_stmt_fn): Remove typedef.
9218 (ssa_prop_get_value_fn): Likewise.
9219 (subsitute_and_fold): Remove prototype.
9220 (replace_uses_in): Likewise.
9221 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9222 Renamed from replace_uses_in. Call the virtual member function
9223 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9224 (substitute_and_fold_dom_walker): Remove initialization of
9225 data member entries for calbacks. Add substitute_and_fold_engine
9226 member and initialize it.
9227 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9228 member functions for get_value, replace_phi_args_in c
9229 replace_uses_in, and fold_stmt calls.
9230 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9231 substitute_and_fold. Remove assert. Update ctor call.
9233 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9234 (ssa_prop_visit_phi_fn): Likewise.
9235 (class ssa_propagation_engine): New class to provide an interface
9237 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9239 (ssa_prop_visit_phi): Likewise.
9240 (ssa_propagation_engine::simulate_stmt): Moved into class.
9241 Call visit_phi/visit_stmt from the class rather than via
9242 file scoped static variables.
9243 (ssa_propagation_engine::simulate_block): Moved into class.
9244 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9245 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9246 set file scoped statics for the visit_stmt/visit_phi callbacks.
9247 * tree-complex.c (complex_propagate): New class derived from
9248 ssa_propagation_engine.
9249 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9250 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9251 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9253 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9254 ssa_propagation_engine.
9255 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9256 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9257 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9258 * tree-ssa-copy.c (copy_prop): New class derived from
9259 ssa_propagation_engine.
9260 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9261 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9262 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9263 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9264 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9265 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9266 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9268 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9270 PR rtl-optimization/82778
9271 PR rtl-optimization/82597
9272 * compare-elim.c (struct comparison): Add in_a_setter field.
9273 (find_comparison_dom_walker::before_dom_children): Remove killed
9274 bitmap and df_simulate_find_defs call, instead walk the defs.
9275 Compute last_setter and initialize in_a_setter. Merge definitions
9276 with first initialization for a few variables.
9277 (try_validate_parallel): Use insn_invalid_p instead of
9278 recog_memoized. Return insn rather than just the pattern.
9279 (try_merge_compare): Fix up comment. Don't uselessly test if
9280 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9282 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9285 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9286 Alan Hayward <alan.hayward@arm.com>
9287 David Sherwood <david.sherwood@arm.com>
9289 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9290 aarch64_hard_regno_nregs to get the number of registers
9293 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9294 Alan Hayward <alan.hayward@arm.com>
9295 David Sherwood <david.sherwood@arm.com>
9297 * config/aarch64/constraints.md (Upl): Rename to...
9299 * config/aarch64/aarch64.md
9300 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9303 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9304 Alan Hayward <alan.hayward@arm.com>
9305 David Sherwood <david.sherwood@arm.com>
9307 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9308 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9311 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9312 Alan Hayward <alan.hayward@arm.com>
9313 David Sherwood <david.sherwood@arm.com>
9315 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9316 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9317 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9319 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9320 of a single element.
9321 * config/aarch64/iterators.md: Add a comment above the permute
9322 unspecs to say that they are generated directly by
9323 aarch64_expand_vec_perm_const.
9324 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9326 2017-11-01 Nathan Sidwell <nathan@acm.org>
9328 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9330 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9332 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9333 and medany code models, and describe what they do.
9335 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9337 Revert accidental duplicate:
9339 * combine.c (can_change_dest_mode): Reject changes in
9340 REGMODE_NATURAL_SIZE.
9342 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9344 PR rtl-optimization/64682
9345 PR rtl-optimization/69567
9346 PR rtl-optimization/69737
9347 PR rtl-optimization/82683
9348 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9349 register mentioned in the note, drop the note, unless it came from I3,
9350 in which case it should go to I3 again.
9352 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9354 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9355 and return false if not.
9356 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9358 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9360 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9361 range is known to be empty.
9363 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9364 Alan Hayward <alan.hayward@arm.com>
9365 David Sherwood <david.sherwood@arm.com>
9367 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9368 and CONST_VECTOR_NUNITS instead of computing the number of units from
9369 the byte sizes of the vector and element.
9370 (simplify_binary_operation_1): Likewise.
9371 (simplify_const_binary_operation): Likewise.
9372 (simplify_ternary_operation): Likewise.
9374 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9375 Alan Hayward <alan.hayward@arm.com>
9376 David Sherwood <david.sherwood@arm.com>
9378 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9379 (int_mem_offset): ...this new function.
9380 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9381 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9382 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9385 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9386 Alan Hayward <alan.hayward@arm.com>
9387 David Sherwood <david.sherwood@arm.com>
9389 * lower-subreg.c (interesting_mode_p): New function.
9390 (compute_costs, find_decomposable_subregs, decompose_register)
9391 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9392 (resolve_clobber, dump_choices): Use it.
9394 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9395 Alan Hayward <alan.hayward@arm.com>
9396 David Sherwood <david.sherwood@arm.com>
9398 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9400 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9401 Alan Hayward <alan.hayward@arm.com>
9402 David Sherwood <david.sherwood@arm.com>
9404 * alias.c (find_base_value, find_base_term): Only process integer
9405 truncations. Check the precision rather than the size.
9407 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9408 Alan Hayward <alan.hayward@arm.com>
9409 David Sherwood <david.sherwood@arm.com>
9411 * machmode.h (is_narrower_int_mode): New function
9412 * optabs.c (expand_float, expand_fix): Use it.
9413 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9415 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9416 Alan Hayward <alan.hayward@arm.com>
9417 David Sherwood <david.sherwood@arm.com>
9419 * rtl.h (narrower_subreg_mode): New function.
9420 * ira-color.c (update_costs_from_allocno): Use it.
9422 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9423 Alan Hayward <alan.hayward@arm.com>
9424 David Sherwood <david.sherwood@arm.com>
9426 * optabs-query.h (convert_optab_p): New function, split out from...
9427 (convert_optab_handler): ...here.
9428 (widening_optab_handler): Delete.
9429 (find_widening_optab_handler): Remove permit_non_widening parameter.
9430 (find_widening_optab_handler_and_mode): Likewise. Provide an
9431 override that operates on mode class wrappers.
9432 * optabs-query.c (widening_optab_handler): Delete.
9433 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9434 parameter. Assert that the two modes are the same class and that
9435 the "from" mode is narrower than the "to" mode. Use
9436 convert_optab_handler instead of widening_optab_handler.
9437 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9438 instead of widening_optab_handler.
9439 * expr.c (expand_expr_real_2): Update calls to
9440 find_widening_optab_handler.
9441 * optabs.c (expand_widen_pattern_expr): Likewise.
9442 (expand_binop_directly): Take the insn_code as a parameter.
9443 (expand_binop): Only call find_widening_optab_handler for
9444 conversion optabs; use optab_handler otherwise. Update calls
9445 to find_widening_optab_handler and expand_binop_directly.
9446 Use convert_optab_handler instead of widening_optab_handler.
9447 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9448 find_widening_optab_handler and use scalar_mode rather than
9450 (convert_plusminus_to_widen): Likewise.
9452 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9453 Alan Hayward <alan.hayward@arm.com>
9454 David Sherwood <david.sherwood@arm.com>
9456 * machmode.h (fixed_size_mode): New class.
9457 * rtl.h (get_pool_mode): Return fixed_size_mode.
9458 * gengtype.c (main): Add fixed_size_mode.
9459 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9460 (get_raw_arg_mode): Likewise.
9461 * doc/tm.texi: Regenerate.
9462 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9463 * targhooks.c (default_get_reg_raw_mode): Likewise.
9464 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9465 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9466 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9467 (msp430_get_raw_result_mode): Likewise.
9468 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9469 * dbxout.c (dbxout_parms): Require fixed-size modes.
9470 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9471 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9472 * omp-low.c (lower_oacc_reductions): Likewise.
9473 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9474 (simplify_subreg): Update accordingly.
9475 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9476 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9477 that aren't fixed-size.
9478 (get_pool_mode): Return a fixed_size_mode.
9479 (output_constant_pool_2): Take a fixed_size_mode.
9481 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9482 Alan Hayward <alan.hayward@arm.com>
9483 David Sherwood <david.sherwood@arm.com>
9485 * doc/rtl.texi (vec_series): Document.
9486 (const): Say that the operand can be a vec_series.
9487 * rtl.def (VEC_SERIES): New rtx code.
9488 * rtl.h (const_vec_series_p_1): Declare.
9489 (const_vec_series_p): New function.
9490 * emit-rtl.h (gen_const_vec_series): Declare.
9491 (gen_vec_series): Likewise.
9492 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9493 (gen_vec_series): Likewise.
9494 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9495 * simplify-rtx.c (simplify_unary_operation): Handle negations
9497 (simplify_binary_operation_series): New function.
9498 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9499 (test_vector_ops_series): New function.
9500 (test_vector_ops): Call it.
9501 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9502 gen_const_vec_series.
9503 (altivec_lvsr): Likewise.
9504 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9506 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9507 Alan Hayward <alan.hayward@arm.com>
9508 David Sherwood <david.sherwood@arm.com>
9510 * doc/rtl.texi (const): Update description of address constants.
9511 Say that vector constants are allowed too.
9512 * common.md (E, F): Use CONSTANT_P instead of checking for
9514 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9515 checking for CONST_VECTOR.
9516 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9518 * expr.c (expand_expr_real_2): Check for vector modes instead
9519 of checking for CONST_VECTOR.
9520 * rtl.h (const_vec_p): New function.
9521 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9522 (unwrap_const_vec_duplicate): Handle them here too.
9524 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9525 David Malcolm <dmalcolm@redhat.com>
9526 Alan Hayward <alan.hayward@arm.com>
9527 David Sherwood <david.sherwood@arm.com>
9529 * rtl.h (vec_duplicate_p): New function.
9530 * selftest-rtl.c (assert_rtx_eq_at): New function.
9531 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9532 (assert_rtx_eq_at): Declare.
9533 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9534 * selftest-run-tests.c (selftest::run_tests): Call it.
9535 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9536 (simplify_unary_operation_1): Recursively handle vector duplicates.
9537 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9539 (simplify_subreg): Handle subregs of vector duplicates.
9540 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9541 (selftest::simplify_rtx_c_tests): New functions.
9543 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9544 Alan Hayward <alan.hayward@arm.com>
9545 David Sherwood <david.sherwood@arm.com>
9547 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9548 (gen_vec_duplicate): Likewise.
9549 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9551 (gen_const_vector): ...here.
9552 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9553 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9554 whose elements are all equal.
9555 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9556 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9557 (simplify_relational_operation): Likewise.
9558 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9560 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9561 (aarch64_expand_vector_init): Likewise.
9562 * config/arm/arm.c (neon_vdup_constant): Likewise.
9563 (neon_expand_vector_init): Likewise.
9564 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9565 (arm_block_set_unaligned_vect): Likewise.
9566 (arm_block_set_aligned_vect): Likewise.
9567 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9568 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9569 (expand_vec_perm_even_odd_pack): Likewise.
9570 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9571 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9572 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9573 gen_const_vec_duplicate.
9574 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9575 * config/mips/mips.c (mips_gen_const_int_vector): Use
9576 gen_const_vec_duplicate.
9577 (mips_expand_vector_init): Use CONST0_RTX.
9578 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9579 (define_split): Use gen_const_vec_duplicate.
9580 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9581 (define_split): Use gen_const_vec_duplicate.
9582 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9583 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9584 * config/spu/spu.c (spu_const): Likewise.
9586 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9587 Alan Hayward <alan.hayward@arm.com>
9588 David Sherwood <david.sherwood@arm.com>
9590 * combine.c (can_change_dest_mode): Reject changes in
9591 REGMODE_NATURAL_SIZE.
9593 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9595 * configure.ac (--enable-libssp): New.
9596 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9597 trying to determine target-specific default. Adjust indentation.
9598 * configure: Regenerated.
9599 * doc/install.texi (Configuration): Expand --disable-libssp
9602 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9604 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9607 2017-10-31 Martin Jambor <mjambor@suse.cz>
9610 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9612 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9614 * auto-profile.c (autofdo_source_profile::read): Use
9615 UNKNOWN_LOCATION rather than 0.
9616 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9617 (warning_at): ...this overload.
9618 (warning_at_rich_loc_n): Rename to...
9619 (warning_n): ...this overload.
9620 (error_at_rich_loc): Rename to...
9621 (error_at): ...this overload.
9622 (pedwarn_at_rich_loc): Rename to...
9623 (pedwarn): ...this overload.
9624 (permerror_at_rich_loc): Rename to...
9625 (permerror): ...this overload.
9626 (inform_at_rich_loc): Rename to...
9627 (inform): ...this overload.
9628 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9629 (diagnostic_n_impl_richloc): Rename to...
9630 (diagnostic_n_impl): ...this rich_location *-based decl.
9631 (inform_at_rich_loc): Rename to...
9632 (inform): ...this, and add an assertion.
9633 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9634 (warning_at_rich_loc): Rename to...
9635 (warning_at): ...this, and add an assertion.
9636 (warning_at_rich_loc_n): Rename to...
9637 (warning_n): ...this, and add an assertion.
9638 (warning_n): Update location_t-based implementation for removal of
9639 location_t-based diagnostic_n_impl.
9640 (pedwarn_at_rich_loc): Rename to...
9641 (pedwarn): ...this, and add an assertion.
9642 (permerror_at_rich_loc): Rename to...
9643 (permerror): ...this, and add an assertion.
9644 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9645 (error_at_rich_loc): Rename to...
9646 (error_at): ...this, and add an assertion.
9647 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9648 (driver::do_spec_on_infiles): Likewise.
9649 * substring-locations.c (format_warning_va): Update for renaming
9650 of inform_at_rich_loc.
9652 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9654 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9655 _Float<N>X built-in functions so that the variant without the
9656 "__builtin_" prefix is only enabled for the GNU C and Objective C
9657 languages when they are in non-strict ANSI/ISO mode.
9658 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9659 * target.def (floatn_builtin_p): Add a target hook to control
9660 whether _Float<N> and _Float<N>X built-in functions without the
9661 "__builtin_" prefix are enabled, and return true for C and
9662 Objective C in the default hook. Include langhooks.h in
9664 * targhooks.h (default_floatn_builtin_p): Likewise.
9665 * targhooks.c (default_floatn_builtin_p): Likewise.
9666 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9667 floatn_builtin_p target hook.
9668 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9670 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9671 Eric Botcazou <ebotcazou@adacore.com>
9673 PR rtl-optimization/81803
9674 * lra-constraints.c (curr_insn_transform): Also reload the whole
9675 register for a strict subreg no wider than a word if this is for
9676 a WORD_REGISTER_OPERATIONS target.
9678 2017-10-31 Jason Merrill <jason@redhat.com>
9680 * gdbinit.in: Skip over inlines from timevar.h.
9682 2017-10-31 Martin Liska <mliska@suse.cz>
9684 * doc/gcov.texi: Document new option.
9685 * gcov.c (print_usage): Likewise print it.
9686 (process_args): Support the argument.
9687 (format_count): New function.
9688 (format_gcov): Use the function.
9690 2017-10-31 Martin Liska <mliska@suse.cz>
9692 * gcov.c (struct name_map): do not use typedef.
9693 Define operator== and operator<.
9694 (name_search): Remove.
9695 (name_sort): Remove.
9696 (main): Do not allocate names.
9697 (process_file): Add vertical space.
9698 (generate_results): Use std::find.
9699 (release_structures): Do not release memory.
9700 (find_source): Use std::find.
9702 2017-10-31 Martin Liska <mliska@suse.cz>
9704 * gcov.c (struct line_info): Remove it's typedef.
9705 (line_info::line_info): Add proper ctor.
9706 (line_info::has_block): Do not use a typedef.
9707 (struct source_info): Do not use typedef.
9708 (circuit): Likewise.
9709 (get_cycles_count): Likewise.
9710 (output_intermediate_file): Iterate via vector iterator.
9711 (add_line_counts): Use std::vector methods.
9712 (accumulate_line_counts): Likewise.
9713 (output_lines): Likewise.
9715 2017-10-31 Martin Liska <mliska@suse.cz>
9717 * gcov.c (struct source_info): Remove typedef.
9718 (source_info::source_info): Add proper ctor.
9719 (accumulate_line_counts): Use struct, not it's typedef.
9720 (output_gcov_file): Likewise.
9721 (output_lines): Likewise.
9722 (main): Do not allocate an array.
9723 (output_intermediate_file): Use size of vector container.
9724 (process_file): Resize the vector.
9725 (generate_results): Do not preallocate, use newly added vector
9727 (release_structures): Do not release sources.
9728 (find_source): Use vector methods.
9729 (add_line_counts): Do not use typedef.
9731 2017-10-31 Martin Liska <mliska@suse.cz>
9733 * doc/gcov.texi: Document that.
9734 * gcov.c (add_line_counts): Mark lines with a non-executed
9736 (output_line_beginning): Handle such lines.
9737 (output_lines): Pass new argument.
9738 (output_intermediate_file): Print it in intermediate format.
9740 2017-10-31 Martin Liska <mliska@suse.cz>
9742 * color-macros.h: New file.
9743 * diagnostic-color.c: Factor out color related to macros to
9745 * doc/gcov.texi: Document -k option.
9746 * gcov.c (INCLUDE_STRING): Include string.h.
9747 (print_usage): Add -k option.
9748 (process_args): Parse it.
9749 (pad_count_string): New function.
9750 (output_line_beginning): Likewise.
9751 (DEFAULT_LINE_START): New macro.
9752 (output_lines): Support color output.
9754 2017-10-31 Martin Liska <mliska@suse.cz>
9756 PR gcov-profile/82633
9757 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9758 their interaction with GCOV infrastructure.
9759 * configure.ac: Add -fkeep-{inline,static}-functions to
9761 * configure: Regenerate.
9763 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
9766 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
9768 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
9771 * config/rs6000/rs6000.md (allocate_stack): Force update interval
9772 into a register if it does not fit into an immediate offset field.
9774 2017-10-31 Olivier Hainque <hainque@adacore.com>
9776 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9778 2017-10-31 Julia Koval <julia.koval@intel.com>
9780 * config.gcc: Add gfniintrin.h.
9781 * config/i386/gfniintrin.h: New.
9782 * config/i386/i386-builtin-types.def
9783 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
9784 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
9785 __builtin_ia32_vgf2p8affineinvqb_v32qi,
9786 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
9787 __builtin_ia32_vgf2p8affineinvqb_v16qi,
9788 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
9789 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
9790 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
9791 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
9792 V64QI_FTYPE_V64QI_V64QI_INT): New types.
9793 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
9794 * config/i386/immintrin.h: Include gfniintrin.h.
9795 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
9797 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9799 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9801 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9803 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9804 (arm_ashldi3_1bit): Remove pattern.
9805 (ashrdi3): Remove shift by 1 expansion.
9806 (arm_ashrdi3_1bit): Remove pattern.
9807 (lshrdi3): Remove shift by 1 expansion.
9808 (arm_lshrdi3_1bit): Remove pattern.
9809 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9810 cost of ashldi3 by 1.
9811 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9812 (<shift>di3_neon): Likewise.
9814 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9816 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9817 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9818 and (*aarch64_simd_mov<VQ:mode>).
9819 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9820 pattern alternative.
9821 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9822 attributes to match pattern alternative.
9824 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9826 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9827 * config/rs6000/emmintrin.h: New file.
9828 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9830 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9832 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9833 * (movdi_vfp_cortexa8): Remove pattern.
9835 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9837 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9839 (Specific, alpha*-dec-osf5.1): Remove.
9840 (Specific, mips-sgi-irix5): Remove.
9841 (Specific, mips-sgi-irix6): Remove.
9843 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9846 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9847 arguments to clear_bit_region_be.
9849 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9851 * gimplify.c: Include memmodel.h.
9853 2017-10-30 Martin Jambor <mjambor@suse.cz>
9855 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9856 condition whether loop should be executed at all.
9858 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9860 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9861 gimple folding of vec_madd() intrinsics.
9862 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9863 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9864 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9866 2017-10-30 Richard Biener <rguenther@suse.de>
9868 PR tree-optimization/82762
9870 2017-10-23 Richard Biener <rguenther@suse.de>
9872 PR tree-optimization/82129
9874 2017-08-01 Richard Biener <rguenther@suse.de>
9876 PR tree-optimization/81181
9877 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9878 (compute_antic): ... end of iteration here.
9880 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9882 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9883 -std=iso9899:2017 and -std=gnu17.
9884 * doc/standards.texi (C Language): Document C17 support.
9885 * doc/cpp.texi (Overview): Mention -std=c17.
9886 (Standard Predefined Macros): Document C11 and C17 values of
9887 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9888 * doc/extend.texi (Inline): Do not list individual options for
9889 standards newer than C99.
9890 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9892 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9895 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9897 * asan.c (asan_finish_file): Align asan globals array by shadow
9900 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9903 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9904 (struct store_immediate_info): Add bitregion_start and bitregion_end
9906 (store_immediate_info::store_immediate_info): Add brs and bre
9907 arguments and initialize bitregion_{start,end} from those.
9908 (struct merged_store_group): Add bitregion_start, bitregion_end,
9909 align_base and mask fields. Drop unnecessary struct keyword from
9910 struct store_immediate_info. Add do_merge method.
9911 (clear_bit_region_be): Use memset instead of loop storing zeros.
9912 (merged_store_group::do_merge): New method.
9913 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9914 stores as long as the surrounding bitregions have no gaps.
9915 (merged_store_group::merge_overlapping): Use do_merge.
9916 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9917 is byte aligned, rather than requiring that start and width are
9918 byte aligned. Drop unnecessary struct keyword from
9919 struct store_immediate_info. Allocate and populate also mask array.
9920 Make start of the arrays relative to bitregion_start rather than
9921 start and size them according to bitregion_{end,start} difference.
9922 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9923 struct store_immediate_info.
9924 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9925 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9926 struct keyword from struct store_immediate_info.
9927 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9928 between stores as long as the surrounding bitregions have no gaps.
9930 (struct split_store): Add orig non-static data member.
9931 (split_store::split_store): Initialize orig to false.
9932 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9933 if there is exactly a single original stmt. Change stmts argument
9934 to pointer from reference, if NULL, don't push anything to it. Add
9935 first argument, use it to optimize skipping over orig stmts that
9936 are known to be before bitpos already. Simplify.
9937 (split_group): Return unsigned int count how many stores are or
9938 would be needed rather than a bool. Add allow_unaligned argument.
9939 Change split_stores argument from reference to pointer, if NULL,
9940 only do a dry run computing how many stores would be produced.
9941 Rewritten algorithm to use both alignment and misalign if
9942 !allow_unaligned and handle bitfield stores with gaps.
9943 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9944 from bitregion_start instead of start. Compute allow_unaligned
9945 here, if true, do 2 split_group dry runs to compute which one
9946 produces fewer stores and prefer aligned if equal. Punt if
9947 new count is bigger or equal than original before emitting any
9948 statements, rather than during that. Remove no longer needed
9949 new_ssa_names tracking. Replace num_stmts with
9950 split_stores.length (). Use 32-bit stack allocated entries
9951 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9952 if possible. Handle bitfields with gaps.
9953 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9954 Compute bitregion_{start,end} for the stores and construct
9955 store_immediate_info with that. Formatting fixes.
9957 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9960 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9961 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9963 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9965 * gimplify.c: Include tm_p.h.
9967 * common.opt (gcoff): Re-add as ignored option.
9968 (gcoff1, gcoff2, gcoff3): Likewise.
9970 * Makefile.in (OBJS): Delete sdbout.o.
9971 (GTFILES): Delete $(srcdir)/sdbout.c.
9972 * debug.h: Delete sdb_debug_hooks.
9973 * final.c: Delete sdbout.h include.
9974 (final_scan_insn): Delete SDB_DEBUG check.
9975 (rest_of_clean_state): Likewise.
9976 * output.h: Delete sdb_begin_function_line.
9979 * toplev.c: Delete sdbout.h include.
9980 (process_options): Delete SDB_DEBUG check.
9981 * tree-core.h (tree_type_common): Delete pointer field of
9983 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9984 TYPE_SYMTAB_POINTER.
9985 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9986 (TYPE_SYMTAB_IS_POINTER): Delete.
9987 (TYPE_SYMTAB_IS_DIE): Renumber.
9988 * xcoffout.c: Refer to former sdbout.c file.
9989 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9991 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9992 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9993 (Debugging Options): Delete -gcoff.
9994 (-gstabs): Delete SDB reference.
9996 (-gcoff@var{level}): Delete.
9997 * doc/passes.texi (Debugging information output): Delete SDB and
9998 sdbout.c references.
9999 * doc/tm.texi: Regenerate.
10000 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10001 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10003 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10004 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10005 (SDB_DEBUGGING_INFO): Delete.
10006 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10007 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10008 * target.def (output_source_filename): Delete COFF reference.
10010 * common.opt (gcoff): Delete.
10011 (gxcoff+): Update Negative chain.
10012 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10014 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10015 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10016 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10017 and SDB references.
10018 (expand_function_start): Change sdb reference to past tense.
10019 (expand_function_end): Change sdb reference to past tense.
10020 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10021 * opts.c (debug_type_names): Delete coff entry.
10022 (common_handle_option): Delete OPT_gcoff case.
10023 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10025 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10026 * config/cris/cris.h: Delete SDB reference in comment.
10027 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10028 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10029 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10030 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10032 (ix86_expand_prologue): Likewise.
10033 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10034 * config/ia64/ia64.h: Likewise.
10035 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10036 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10038 * config/mmix/mmix.h: Likewise.
10039 * config/nds32/nds32.c: Likewise.
10040 * config/stormy/storym16.h: Likewise.
10041 * config/visium/visium.h: Likewise.
10042 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10044 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10046 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10047 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10048 FRAME_POINTER_REGNUM point at high end of local var area.
10050 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10052 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10053 Move comment around. Do not reset best_edge for a copiable
10054 destination if the copy would cause a partition change.
10055 (better_edge_p): Remove redundant check.
10057 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10059 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10061 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10063 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10064 for math functions that have _Float<N> and _Float<N>X variants.
10065 (mathfn_built_in_2): Add support for math functions that have
10066 _Float<N> and _Float<N>X variants.
10067 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10068 (expand_builtin_mathfn_ternary): Add support for fma with
10069 _Float<N> and _Float<N>X variants.
10070 (expand_builtin): Likewise.
10071 (fold_builtin_3): Likewise.
10072 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10073 create math function _Float<N> and _Float<N>X variants as external
10075 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10076 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10077 the __builtin_ prefix and if not strict ansi, without the prefix.
10078 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10079 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10080 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10081 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10082 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10083 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10084 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10085 function signatures for fma _Float<N> and _Float<N>X variants.
10086 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10087 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10088 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10089 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10090 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10091 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10092 * gencfn-macros.c (print_case_cfn): Add support for math functions
10093 that have _Float<N> and _Float<N>X variants.
10094 (print_define_operator_list): Likewise.
10095 (fltfn_suffixes): Likewise.
10097 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10098 for math functions that have _Float<N> and _Float<N>X variants.
10099 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10100 and _Float<N>X variants.
10101 (COPYSIGN): Likewise.
10104 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10105 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10107 (integer_valued_read_call_p): Likewise.
10108 * fold-const-call.c (fold_const_call_ss): Likewise.
10109 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10110 _Float<N> and _Float<N>X variants.
10111 (fold_const_call_ssss): Add support for fma _Float<N> and
10112 _Float<N>X variants.
10113 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10114 support for copysign and fma _Float<N> and _Float<N>X variants.
10115 (backprop::process_builtin_call_use): Likewise.
10116 * tree-call-cdce.c (can_test_argument_range); Add support for
10117 sqrt _Float<N> and _Float<N>X variants.
10118 (edom_only_function): Likewise.
10119 (get_no_error_domain): Likewise.
10120 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10121 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10122 copysign _Float<N> and _Float<N>X variants.
10123 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10124 handled by machine independent code.
10125 (FMAF128): Likewise.
10126 * doc/cpp.texi (Common Predefined Macros): Document defining
10127 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10128 fma _Float<N> and _Float<N>X variants.
10130 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10133 * config/i386/i386-modes.def (CCFPU): Remove definition.
10134 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10135 (ix86_cc_modes_compatible): Ditto.
10136 (ix86_expand_carry_flag_compare): Ditto.
10137 (ix86_expand_int_movcc): Ditto.
10138 (ix86_expand_int_addcc): Ditto.
10139 (ix86_reverse_condition): Ditto.
10140 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10141 Return true/false for unordered/ordered fp comparisons.
10142 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10143 (ix86_prepare_fp_compare_args): Update for rename.
10144 (ix86_expand_fp_compare): Update for rename. Generate unordered
10145 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10146 (ix86_expand_sse_compare_and_jump): Ditto.
10147 * config/i386/predicates.md (fcmov_comparison_operator):
10148 Remove CCFPU mode handling.
10149 (ix86_comparison_operator): Ditto.
10150 (ix86_carry_flag_operator): Ditto.
10151 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10152 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10153 (*cmpu<mode>_cc_i387): Ditto.
10154 (FPCMP): Remove mode iterator.
10155 (unord): Remove mode attribute.
10156 (unord_subst): New define_subst transformation
10157 (unord): New define_subst attribute.
10158 (unordered): Ditto.
10159 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10160 (*cmpi<unord>xf_i387): Ditto.
10161 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10162 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10163 using unord_subst transformation.
10164 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10165 (round_saeonly): Also handle CCFP mode.
10166 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10167 Remove UNSPEC_SAHF unspec handling.
10169 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10171 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10173 2017-10-27 Jeff Law <law@redhat.com>
10175 * gimple-ssa-sprintf.c: Include domwalk.h.
10176 (class sprintf_dom_walker): New class, derived from dom_walker.
10177 (sprintf_dom_walker::before_dom_children): New function.
10178 (struct call_info): Moved into sprintf_dom_walker class
10179 (compute_formath_length, handle_gimple_call): Likewise.
10180 (sprintf_length::execute): Call the dominator walker rather
10181 than walking the statements.
10183 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10184 gimple statement locations.
10185 (check_array_bounds): Corresponding changes. Get the statement's
10186 location directly from wi->stmt.
10188 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10191 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10193 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10195 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10196 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10198 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10201 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10202 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10203 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10204 maybe_get_pool_constant.
10205 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10208 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10210 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10211 2.26 caveat. Update gas and gld versions.
10212 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10215 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10217 * cgraph.h (set_malloc_flag): Declare.
10218 * cgraph.c (set_malloc_flag_1): New function.
10219 (set_malloc_flag): Likewise.
10220 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10221 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10223 (read_ipa_call_summary): Add support for reading is_return_callee.
10224 (write_ipa_call_summary): Stream is_return_callee.
10225 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10226 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10227 ipa-prop.h, ipa-fnsummary.h.
10228 (pure_const_names): Change to static.
10229 (malloc_state_e): Define.
10230 (malloc_state_names): Define.
10231 (funct_state_d): Add field malloc_state.
10232 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10233 (check_retval_uses): New function.
10234 (malloc_candidate_p): Likewise.
10235 (analyze_function): Add support for malloc attribute.
10236 (pure_const_write_summary): Stream malloc_state.
10237 (pure_const_read_summary): Add support for reading malloc_state.
10238 (dump_malloc_lattice): New function.
10239 (propagate_malloc): New function.
10240 (warn_function_malloc): New function.
10241 (ipa_pure_const::execute): Call propagate_malloc and
10242 ipa_free_fn_summary.
10243 (pass_local_pure_const::execute): Add support for malloc attribute.
10244 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10245 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10247 2017-10-27 Martin Liska <mliska@suse.cz>
10249 PR gcov-profile/82457
10250 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10251 for fork-like functions to be properly instrumented.
10253 2017-10-27 Richard Biener <rguenther@suse.de>
10255 PR middle-end/81659
10256 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10257 info when we redirected EH.
10259 2017-10-26 Michael Collison <michael.collison@arm.com>
10261 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10263 (<optab>_trunchf<GPI:mode>2: New pattern.
10264 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10265 * config/aarch64/iterators.md (wv): New mode attribute.
10266 (vf, VF): New mode attributes.
10267 (vgp, VGP): New mode attributes.
10268 (s): Update attribute with SImode and DImode prefixes.
10270 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10272 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10273 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10274 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10275 (nios2_option_overide): Initialize it. Don't allow R0-relative
10276 addressing with PIC.
10277 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10278 (nios2_symbolic_constant_p): Likewise.
10279 (nios2_legitimate_address_p): Likewise.
10280 (nios2_r0rel_section_name_p): New.
10281 (nios2_symbol_ref_in_r0rel_data_p): New.
10282 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10283 (r0rel_constant_p): New.
10284 (nios2_print_operand_address): Handle r0rel_constant_p.
10285 (nios2_cdx_narrow_form_p): Likewise.
10286 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10287 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10288 (Nios II Options): Document -mr0rel-sec.
10290 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10292 * config/nios2/nios2.c: Include xregex.h.
10293 (nios2_gprel_sec_regex): New.
10294 (nios2_option_overide): Initialize it. Don't allow GP-relative
10295 addressing with PIC.
10296 (nios2_small_section_name_p): Check for regex match.
10297 * config/nios2/nios2.opt (mgprel-sec=): New option.
10298 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10299 (Nios II Options): Document -mgprel-sec.
10301 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10303 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10305 2017-10-26 Tom de Vries <tom@codesourcery.com>
10307 PR tree-optimization/82707
10308 * gimple.c (gimple_copy): Fix unsharing of
10309 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10311 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10313 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10314 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10315 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10316 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10317 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10318 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10319 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10320 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10321 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10322 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10323 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10324 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10325 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10326 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10327 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10328 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10329 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10331 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10333 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10335 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10336 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10337 warning to rs6000.c. Remove the Undocumented flag, since it has
10339 (-mabi=ibmlongdouble): Likewise.
10340 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10341 already set, set the default format for long double.
10342 (rs6000_debug_reg_global): Print whether long double is IBM or
10344 (rs6000_option_override_internal): Rework setting long double
10345 format. Only warn if the user is changing the long double default
10346 and they did not use -Wno-psabi.
10347 * doc/invoke.texi (PowerPC options): Update the documentation for
10348 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10350 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10351 Alan Hayward <alan.hayward@arm.com>
10352 David Sherwood <david.sherwood@arm.com>
10354 * rtl.h (wider_subreg_mode): New function.
10355 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10356 rather than an unsigned int *.
10357 * ira-color.c (regno_max_ref_width): Replace with...
10358 (regno_max_ref_mode): ...this new variable.
10359 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10360 Use wider_subreg_mode.
10361 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10362 rather than an unsigned int *.
10363 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10364 (process_alt_operands): Likewise.
10365 (invariant_p): Likewise.
10366 * lra-spills.c (assign_mem_slot): Likewise.
10367 (add_pseudo_to_slot): Likewise.
10368 * lra.c (collect_non_operand_hard_regs): Likewise.
10369 (add_regs_to_insn_regno_info): Likewise.
10370 * reload1.c (regno_max_ref_width): Replace with...
10371 (regno_max_ref_mode): ...this new variable.
10372 (reload): Update accordingly. Update call to
10373 ira_sort_regnos_for_alter_reg.
10374 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10375 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10376 (scan_paradoxical_subregs): Likewise.
10378 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10380 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10381 (aarch64_frame): Add emit_frame_chain boolean.
10382 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10383 Move eh_return case to aarch64_layout_frame.
10384 (aarch64_layout_frame): Initialize emit_frame_chain.
10385 (aarch64_expand_prologue): Use emit_frame_chain.
10387 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10389 * config/aarch64/aarch64.c (aarch64_layout_frame):
10390 Ensure LR is always stored at the bottom of the callee-saves.
10391 Remove rarely used frame layout which saves callee-saves at top of
10392 frame, so the store of LR can be used as a valid probe in all cases.
10394 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10396 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10397 Improve unaligned TImode/TFmode base/offset split.
10399 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10400 Alan Hayward <alan.hayward@arm.com>
10401 David Sherwood <david.sherwood@arm.com>
10403 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10404 * combine.c (find_single_use_1): Likewise.
10405 (expand_field_assignment): Likewise.
10406 (move_deaths): Likewise.
10407 * lra-constraints.c (simplify_operand_subreg): Likewise.
10408 (curr_insn_transform): Likewise.
10409 * lra.c (collect_non_operand_hard_regs): Likewise.
10410 (add_regs_to_insn_regno_info): Likewise.
10411 * rtlanal.c (reg_referenced_p): Likewise.
10412 (covers_regno_no_parallel_p): Likewise.
10414 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10416 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10417 Don't print any bits outside the precision of the value.
10418 * wide-int.cc (test_printing): Add some new tests.
10420 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10422 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10423 supports -xbrace_comment option.
10424 * configure: Regenerate.
10425 * config.in: Regenerate.
10426 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10427 (ASM_CPU_SPEC): Use it.
10429 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10431 * target.def (static_rtx_alignment): New hook.
10432 * targhooks.h (default_static_rtx_alignment): Declare.
10433 * targhooks.c (default_static_rtx_alignment): New function.
10434 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10435 * doc/tm.texi: Regenerate.
10436 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10437 instead of targetm.constant_alignment. Remove call to
10438 set_mem_attributes.
10439 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10440 (cris_preferred_mininum_alignment): New function, split out from...
10441 (cris_constant_alignment): ...here.
10442 (cris_static_rtx_alignment): New function.
10443 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10445 (ix86_constant_alignment): ...here.
10446 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10447 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10448 (mmix_static_rtx_alignment): New function.
10449 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10450 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10452 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10455 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10456 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10458 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10460 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10461 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10462 mark as a sign-extending load.
10463 (local_pic_load_u): Define.
10465 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10467 PR middle-end/82062
10468 * fold-const.c (operand_equal_for_comparison_p): Also return true
10469 if ARG0 is a simple variant of ARG1 with narrower precision.
10470 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10472 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10474 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10476 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10477 scatter_static, scatter_per_elt.
10478 * x86-tune-costs.h: Add new cost entries.
10480 2017-10-25 Richard Biener <rguenther@suse.de>
10482 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10483 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10484 class pass_fre): Move to ...
10485 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10486 class pass_fre): ... here and adjust for statistics.
10488 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10491 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10492 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10493 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10494 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10497 2017-10-25 Richard Biener <rguenther@suse.de>
10499 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10500 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10501 eliminate_push_avail, eliminate_insert): Move inside...
10502 (class eliminate_dom_walker): ... this class in preparation
10504 (fini_eliminate): Remove by merging with ...
10505 (eliminate): ... this function. Adjust for class changes.
10506 (pass_pre::execute): Remove fini_eliminate call.
10507 (pass_fre::execute): Likewise.
10509 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10512 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10513 (VPERMI2, VPERMI2I): New mode iterators.
10514 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10515 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10517 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10518 mode iterator. Remove 3 old define_insn patterns.
10519 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10520 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10521 VPERMI2 mode iterator, remove the other two expanders.
10522 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10523 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10524 instructions, remove the other two patterns.
10525 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10526 mode iterator, remove the other two patterns.
10527 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10528 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10529 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10530 and adjust argument order accordingly.
10531 (ix86_expand_vec_perm): Adjust caller.
10532 (expand_vec_perm_1): Likewise.
10533 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10534 (expand_vec_perm_vpermt2_vpshub2): ... this.
10535 (ix86_expand_vec_perm_const_1): Adjust caller.
10536 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10539 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10540 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10541 (vec_shl_<mode>): Remove unused expander.
10542 (avx512bw_<shift_insn><mode>3): New define_insn.
10543 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10544 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10546 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10549 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10552 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10554 PR rtl-optimization/82396
10555 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10556 (autopref_multipass_init): Simplify initialization.
10557 (autopref_rank_data): Simplify sort order.
10558 * gcc/sched-int.h (autopref_multipass_data_): Remove
10559 multi_mem_insn_p, min_offset and max_offset.
10561 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10563 PR middle-end/60580
10564 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10565 Check special value of flag_omit_frame_pointer.
10566 (aarch64_can_eliminate): Likewise.
10567 (aarch64_override_options_after_change_1): Simplify handling of
10568 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10570 2017-10-24 Richard Biener <rguenther@suse.de>
10572 PR tree-optimization/82697
10573 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10574 zero for conditional load and unconditional store.
10576 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10578 * doc/install.texi: Document bootstrap-cet.
10580 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10583 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10584 ENDBR instruction at function entrance if function is only
10587 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10590 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10591 patterns to better describe from which operation the CF is computed.
10592 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10593 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10594 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10595 is 0, use _0 suffixed expanders instead of emitting a comparison
10598 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10600 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10601 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10603 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10605 PR middle-end/82569
10606 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10607 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10608 * loop-iv.c (iv_get_reaching_def): Likewise.
10609 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10610 variable is promoted and the partition contains undefined values.
10612 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10614 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10616 (nios2_address_cost): Define.
10617 (nios2_legitimize_address): Recognize (exp + constant) directly.
10618 (TARGET_ADDRESS_COST): Define.
10620 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10622 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10623 (nios2_symbolic_memory_operand_p): Declare.
10624 (nios2_split_large_constant): Declare.
10625 (nios2_split_symbolic_memory_operand): Declare.
10626 * config/nios2/nios2.c: Adjust includes.
10627 (nios2_symbolic_constant_allowed): New.
10628 (nios2_symbolic_constant_p): New.
10629 (nios2_plus_symbolic_constant_p): New.
10630 (nios2_valid_addr_expr_p): Recognize addresses involving
10631 symbolic constants.
10632 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10633 (nios2_symbolic_memory_operand_p): New.
10634 (nios2_large_constant_p): New.
10635 (nios2_split_large_constant): New.
10636 (nios2_split_plus_large_constant): New.
10637 (nios2_split_symbolic_memory_operand): New.
10638 (nios2_legitimize_address): Code refactoring. Handle addresses
10639 involving symbolic constants.
10640 (nios2_emit_move_sequence): Likewise.
10641 (nios2_print_operand): Improve error output.
10642 (nios2_print_operand_address): Handle LO_SUM.
10643 (nios2_cdx_narrow_form_p): Likewise.
10644 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10645 operands involving symbolic constants.
10646 (movhi_internal, movsi_internal): Likewise.
10647 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10648 (extendhisi2, extendqi<mode>2): Likewise.
10650 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10652 * tree-pass.h (PROP_rtl_split_insns): Define.
10653 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10655 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10657 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10659 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10662 * target.def (const_not_ok_for_debug_p): Default to
10663 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10664 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10665 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10666 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10667 which targetm.const_not_ok_for_debug_p returned true.
10668 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10670 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10672 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10673 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10674 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10676 (ix86_const_not_ok_for_debug_p): New function.
10677 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10678 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10680 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10683 * system.h: Conditionally include "unique-ptr.h" if
10684 INCLUDE_UNIQUE_PTR is defined.
10685 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10686 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10688 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10690 * config/rl78/rl78.md: New define_expand "subdi3".
10692 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10695 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10696 DF_REF_INSN if DF_REF_INSN_INFO is false.
10698 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10700 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10701 xmm_move instead of sse_move.
10702 (sse_store_index): New function.
10703 (ix86_register_move_cost): Be more sensible about mismatch stall;
10704 model AVX moves correctly; make difference between sse->integer and
10706 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10707 moves; make difference between SSE and AVX.
10708 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10709 and zmm_move. Increase size of sse load and store tables;
10710 add unaligned load and store tables; add ssemmx_to_integer.
10711 * x86-tune-costs.h: Update all entries according to real
10712 move latencies from Agner Fog's manual and chip documentation.
10714 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10717 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
10718 * config/i386/constraints.md (Wf): New constraint.
10719 * config/i386/i386.md (UNSPEC_SBB): New unspec.
10720 (cmp<dwi>_doubleword): Removed.
10721 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
10722 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
10723 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
10724 expand with cmp<dwi>_doubleword. For LTU and GEU use
10725 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
10727 * common.opt (gcolumn-info): Enable by default.
10728 * doc/invoke.texi (gcolumn-info): Document new default.
10730 2017-10-23 Richard Biener <rguenther@suse.de>
10732 PR tree-optimization/82672
10733 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10734 Fold the stmt if we propagated into it.
10736 2017-10-23 Richard Biener <rguenther@suse.de>
10738 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
10739 (bitmap_remove_expr_from_set): ... this. All callers call this
10740 for non-constant values.
10741 (bitmap_set_subtract): Rename to...
10742 (bitmap_set_subtract_expressions): ... this. Adjust and
10744 (bitmap_set_contains_value): Remove superfluous check.
10745 (bitmap_set_replace_value): Inline into single caller ...
10746 (bitmap_value_replace_in_set): ... here and simplify.
10747 (dependent_clean): Merge into ...
10748 (clean): ... this using an overload. Adjust.
10749 (prune_clobbered_mems): Adjust.
10750 (compute_antic_aux): Likewise.
10751 (compute_partial_antic_aux): Likewise.
10753 2017-10-23 Richard Biener <rguenther@suse.de>
10755 PR tree-optimization/82129
10757 2017-08-01 Richard Biener <rguenther@suse.de>
10759 PR tree-optimization/81181
10760 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10761 (compute_antic): ... end of iteration here.
10763 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10765 * target.def (starting_frame_offset): New hook.
10766 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
10767 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
10768 * doc/tm.texi.in: Regenerate.
10769 * hooks.h (hook_hwi_void_0): Declare.
10770 * hooks.c (hook_hwi_void_0): New function.
10771 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
10772 STARTING_FRAME_OFFSET.
10773 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10774 * reload1.c (reload): Likewise.
10775 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10776 instead of STARTING_FRAME_OFFSET.
10777 * function.c (try_fit_stack_local): Likewise.
10778 (assign_stack_local_1): Likewise
10779 (instantiate_virtual_regs): Likewise.
10780 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
10781 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
10782 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
10783 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
10784 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
10785 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
10786 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
10787 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
10788 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
10789 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
10790 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
10791 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
10792 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
10793 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
10794 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
10795 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
10796 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
10797 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10798 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10799 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10800 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10801 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10802 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10803 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10804 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10805 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10806 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10807 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10808 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10809 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10810 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10811 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10812 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10813 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10814 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10815 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10816 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10817 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10818 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10819 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10820 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10821 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10822 return a HOST_WIDE_INT.
10823 (avr_builtin_setjmp_frame_value): Use it instead of
10824 STARTING_FRAME_OFFSET.
10825 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10826 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10827 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10829 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10830 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10831 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10832 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10833 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10834 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10835 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10836 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10837 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10838 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10839 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10840 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10842 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10843 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10844 * config/mips/mips.c (mips_compute_frame_info): Refer to
10845 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10846 (mips_starting_frame_offset): New function.
10847 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10848 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10849 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10850 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10851 and return a HOST_WIDE_INT.
10852 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10853 (mmix_initial_elimination_offset): Refer to
10854 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10855 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10856 * config/pa/pa.c (pa_starting_frame_offset): New function.
10857 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10858 (pa_expand_prologue): Likewise.
10859 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10860 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10861 !FRAME_GROWS_DOWNWARD handling to...
10862 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10863 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10864 !FRAME_GROWS_DOWNWARD handling to...
10865 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10866 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10867 !FRAME_GROWS_DOWNWARD handling to...
10868 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10869 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10871 (rs6000_starting_frame_offset): New function.
10872 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10873 !FRAME_GROWS_DOWNWARD handling to...
10874 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10875 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10876 !FRAME_GROWS_DOWNWARD handling to...
10877 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10878 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10879 !FRAME_GROWS_DOWNWARD handling to...
10880 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10881 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10882 (rs6000_starting_frame_offset): New function.
10883 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10884 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10885 * config/vax/vax.c (vax_starting_frame_offset): New function.
10886 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10887 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10888 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10889 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10890 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10891 * system.h (STARTING_FRAME_OFFSET): Poison.
10893 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10895 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10896 SCALAR_TYPE_MODE instead of TYPE_MODE.
10898 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10899 Alan Hayward <alan.hayward@arm.com>
10900 David Sherwood <david.sherwood@arm.com>
10902 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10904 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10905 Alan Hayward <alan.hayward@arm.com>
10906 David Sherwood <david.sherwood@arm.com>
10908 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10910 2017-10-23 Richard Biener <rguenther@suse.de>
10912 PR tree-optimization/82129
10913 * tree-ssa-pre.c (bitmap_set_and): Remove.
10914 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10915 canonicalizing expressions in the set to those with lowest
10916 ID rather than taking that from the first edge.
10918 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10920 * combine.c (rtx_equal_for_field_assignment_p): Use
10921 byte_lowpart_offset.
10923 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10924 Alan Hayward <alan.hayward@arm.com>
10925 David Sherwood <david.sherwood@arm.com>
10927 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10928 to a SUBREG_PROMOTED_VAR.
10930 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10931 Alan Hayward <alan.hayward@arm.com>
10932 David Sherwood <david.sherwood@arm.com>
10934 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10935 (expand_debug_source_expr): Likewise.
10936 * combine.c (combine_simplify_rtx): Likewise.
10937 * cse.c (fold_rtx): Likewise.
10938 * optabs.c (expand_float): Likewise.
10939 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10940 (simplify_binary_operation_1): Likewise.
10942 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10943 Alan Hayward <alan.hayward@arm.com>
10944 David Sherwood <david.sherwood@arm.com>
10946 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10947 (record_promoted_value): Likewise.
10948 * expr.c (expand_expr_real_2): Likewise.
10949 * ree.c (update_reg_equal_equiv_notes): Likewise.
10950 (combine_set_extension): Likewise.
10951 * rtlanal.c (low_bitmask_len): Likewise.
10952 * simplify-rtx.c (neg_const_int): Likewise.
10953 (simplify_binary_operation_1): Likewise.
10955 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10956 Alan Hayward <alan.hayward@arm.com>
10957 David Sherwood <david.sherwood@arm.com>
10959 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10960 * regcprop.c (maybe_mode_change): Likewise.
10961 * reload1.c (alter_reg): Likewise.
10963 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10965 * inchash.h (inchash::hash::add_wide_int): New function.
10966 * lto-streamer-out.c (hash_tree): Use it.
10968 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10970 * inchash.h (inchash::hash::add_wide_int): Rename to...
10971 (inchash::hash::add_hwi): ...this.
10972 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10973 (polymorphic_call_target_hasher::hash): Likewise.
10974 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10975 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10976 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10977 * lto-streamer-out.c (hash_tree): Likewise.
10978 * optc-save-gen.awk: Likewise.
10979 * tree.c (add_expr): Likewise.
10981 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10984 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10985 for ordered inequality comparisons even with TARGET_IEEE_FP.
10987 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10990 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10991 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10992 Expand with cmp<dwi>_doubleword.
10994 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10996 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10997 List CET intrinsics.
10998 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10999 specific to -fcf-protection option.
11001 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11003 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11004 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11005 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11006 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11007 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11008 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11009 (extra_objs): Add cet.o for Linux/x86 targets.
11010 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11011 * config/i386/cet.c: New file.
11012 * config/i386/cetintrin.h: Likewise.
11013 * config/i386/t-cet: Likewise.
11014 * config/i386/cpuid.h (bit_SHSTK): New.
11015 (bit_IBT): Likewise.
11016 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11017 pass IBT and SHSTK bits.
11018 * config/i386/i386-builtin-types.def
11019 (VOID_FTYPE_UNSIGNED_PVOID): New.
11020 (VOID_FTYPE_UINT64_PVOID): Likewise.
11021 * config/i386/i386-builtin.def: Add CET intrinsics.
11022 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11023 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11024 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11025 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11027 * config/i386/i386.c (rest_of_insert_endbranch): New.
11028 (pass_data_insert_endbranch): Likewise.
11029 (pass_insert_endbranch): Likewise.
11030 (make_pass_insert_endbranch): Likewise.
11031 (ix86_notrack_prefixed_insn_p): Likewise.
11032 (ix86_target_string): Add -mibt, -mshstk flags.
11033 (ix86_option_override_internal): Add flag_cf_protection
11035 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11036 (ix86_print_operand): Add 'notrack' prefix output.
11037 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11038 (ix86_expand_builtin): Expand CET intrinsics.
11039 (x86_output_mi_thunk): Add 'endbranch' instruction.
11040 * config/i386/i386.h (TARGET_IBT): New.
11041 (TARGET_IBT_P): Likewise.
11042 (TARGET_SHSTK): Likewise.
11043 (TARGET_SHSTK_P): Likewise.
11044 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11045 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11046 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11047 (builtin_setjmp_setup): New pattern.
11048 (builtin_longjmp): Likewise.
11049 (rdssp<mode>): Likewise.
11050 (incssp<mode>): Likewise.
11051 (saveprevssp): Likewise.
11052 (rstorssp): Likewise.
11053 (wrss<mode>): Likewise.
11054 (wruss<mode>): Likewise.
11055 (setssbsy): Likewise.
11056 (clrssbsy): Likewise.
11057 (nop_endbr): Likewise.
11058 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11060 * config/i386/immintrin.h: Include <cetintrin.h>.
11061 * config/i386/linux-common.h
11062 (file_end_indicate_exec_stack_and_cet): New prototype.
11063 (TARGET_ASM_FILE_END): New.
11065 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11067 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11068 latencies instead of having separate table; make difference between
11069 integer and float costs.
11070 * i386.h (processor_costs): Remove scalar_stmt_cost,
11071 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11072 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11074 * x86-tune-costs.h: Remove entries which has been removed in
11075 procesor_costs from all tables; make cond_taken_branch_cost
11076 and cond_not_taken_branch_cost COST_N_INSNS based.
11078 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11080 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11082 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11084 * config/i386/i386.md (isa): Remove fma_avx512f.
11085 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11086 <avx512>_fmadd_<mode>_mask3<round_name>,
11087 <avx512>_fmsub_<mode>_mask<round_name>,
11088 <avx512>_fmsub_<mode>_mask3<round_name>,
11089 <avx512>_fnmadd_<mode>_mask<round_name>,
11090 <avx512>_fnmadd_<mode>_mask3<round_name>,
11091 <avx512>_fnmsub_<mode>_mask<round_name>,
11092 <avx512>_fnmsub_<mode>_mask3<round_name>,
11093 <avx512>_fmaddsub_<mode>_mask<round_name>,
11094 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11095 <avx512>_fmsubadd_<mode>_mask<round_name>,
11096 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11097 (*vec_widen_umult_even_v16si<mask_name>,
11098 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11099 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11101 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11103 * extend.texi: Add 'nocf_check' documentation.
11104 * gimple.texi: Add second parameter to
11105 gimple_build_call_from_tree.
11106 * invoke.texi: Add -fcf-protection documentation.
11107 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11109 2017-10-20 Richard Biener <rguenther@suse.de>
11111 PR tree-optimization/82473
11112 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11113 the largest input type.
11115 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11117 * c-attribs.c (handle_nocf_check_attribute): New function.
11118 (c_common_attribute_table): Add 'nocf_check' handling.
11119 * gimple-parser.c: Add second argument NULL to
11120 gimple_build_call_from_tree.
11121 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11122 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11124 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11125 * common.opt: Add fcf-protection flag.
11126 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11127 * flag-types.h: Add enum cf_protection_level.
11128 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11129 Add 'nocf_check' attribute propagation to gimple call.
11130 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11131 (gimple_build_call_from_tree): Update prototype.
11132 (gimple_call_nocf_check_p): New function.
11133 (gimple_call_set_nocf_check): Likewise.
11134 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11135 * ipa-icf.c: Add nocf_check attribute in statement hash.
11136 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11137 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11138 * toplev.c (process_options): Add flag_cf_protection handling.
11140 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11142 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11144 2017-10-20 Richard Biener <rguenther@suse.de>
11146 PR tree-optimization/82603
11147 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11148 remove false predicated stores.
11150 2017-10-20 Richard Biener <rguenther@suse.de>
11152 * graphite-isl-ast-to-gimple.c
11153 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11154 Remove return value and simplify, dump copied stmt after lhs
11156 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11157 Reduce dump verbosity.
11158 (gsi_insert_earliest): Likewise.
11159 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11160 * graphite.c (print_global_statistics): Adjust dumping.
11161 (print_graphite_scop_statistics): Likewise.
11162 (print_graphite_statistics): Do not dump loops here.
11163 (graphite_transform_loops): But here.
11165 2017-10-20 Nicolas Roche <roche@adacore.com>
11167 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11168 * configure: Regenerate.
11170 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11173 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11174 functions when optimizing replace GIMPLE_RETURN stmts with
11175 calls to __builtin_unreachable ().
11178 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11179 for -fsanitize=thread link of executables.
11180 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11181 link of executables.
11184 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11185 New mode iterators.
11186 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11187 define_insns for logical vector shifts to use VI248_AVX512BW
11188 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11189 condition, useless isa and prefix attributes. Change the first
11190 2 of these define_insns to ...
11191 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11192 define_insn for avx512vl.
11193 (<shift_insn><mode>3): ... and this, new define_insn without
11194 masking for non-avx512vl.
11197 * config/i386/sse.md (*andnot<mode>3,
11198 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11199 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11200 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11201 not applied use empty suffix even for TARGET_AVX512VL.
11202 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11203 is applied, supply evex,evex or evex,evex,evex instead of just
11206 2017-10-20 Julia Koval <julia.koval@intel.com>
11208 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11209 (OPTION_MASK_ISA_GFNI_UNSET): New.
11210 (ix86_handle_option): Handle OPT_mgfni.
11211 * config/i386/cpuid.h (bit_GFNI): New.
11212 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11213 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11214 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11215 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11216 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11217 * config/i386/i386.opt: Add mgfni.
11219 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11221 * config/msp430/msp430.c (msp430_option_override): Disable
11222 -fdelete-null-pointer-checks.
11223 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11225 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11227 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11228 of x87 and SSE instructions.
11230 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11232 * asan.c (create_cond_insert_point): Do not update edge count.
11233 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11234 (afdo_propagate_circuit): Likewise.
11235 (afdo_calculate_branch_prob): Likewise.
11236 (afdo_annotate_cfg): Likewise.
11237 * basic-block.h (struct edge_def): Remove count.
11238 (edge_def::count): New accessor.
11239 * bb-reorder.c (rotate_loop): Update.
11240 (find_traces_1_round): Update.
11241 (connect_traces): Update.
11242 (sanitize_hot_paths): Update.
11243 * cfg.c (unchecked_make_edge): Update.
11244 (make_single_succ_edge): Update.
11245 (check_bb_profile): Update.
11246 (dump_edge_info): Update.
11247 (update_bb_profile_for_threading): Update.
11248 (scale_bbs_frequencies_int): Update.
11249 (scale_bbs_frequencies_gcov_type): Update.
11250 (scale_bbs_frequencies_profile_count): Update.
11251 (scale_bbs_frequencies): Update.
11252 * cfganal.c (connect_infinite_loops_to_exit): Update.
11253 * cfgbuild.c (compute_outgoing_frequencies): Update.
11254 (find_many_sub_basic_blocks): Update.
11255 * cfgcleanup.c (try_forward_edges): Update.
11256 (try_crossjump_to_edge): Update
11257 * cfgexpand.c (expand_gimple_cond): Update
11258 (expand_gimple_tailcall): Update
11259 (construct_exit_block): Update
11260 * cfghooks.c (verify_flow_info): Update
11261 (redirect_edge_succ_nodup): Update
11262 (split_edge): Update
11263 (make_forwarder_block): Update
11264 (duplicate_block): Update
11265 (account_profile_record): Update
11266 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11267 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11268 * cfgloopmanip.c (scale_loop_profile): Update.
11270 (lv_adjust_loop_entry_edge): Update.
11271 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11272 (force_nonfallthru_and_redirect): Update.
11273 (purge_dead_edges): Update.
11274 (rtl_flow_call_edges_add): Update.
11275 * cgraphunit.c (init_lowered_empty_function): Update.
11276 (cgraph_node::expand_thunk): Update.
11277 * gimple-pretty-print.c (dump_probability): Update.
11278 (dump_edge_probability): Update.
11279 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11280 * haifa-sched.c (sched_create_recovery_edges): Update.
11281 * hsa-gen.c (convert_switch_statements): Update.
11282 * ifcvt.c (dead_or_predicable): Update.
11283 * ipa-inline-transform.c (inline_transform): Update.
11284 * ipa-split.c (split_function): Update.
11285 * ipa-utils.c (ipa_merge_profiles): Update.
11286 * loop-doloop.c (add_test): Update.
11287 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11288 * lto-streamer-in.c (input_cfg): Update.
11289 (input_function): Update.
11290 * lto-streamer-out.c (output_cfg): Update.
11291 * modulo-sched.c (sms_schedule): Update.
11292 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11293 * predict.c (maybe_hot_edge_p): Update.
11294 (unlikely_executed_edge_p): Update.
11295 (probably_never_executed_edge_p): Update.
11296 (dump_prediction): Update.
11297 (drop_profile): Update.
11298 (propagate_unlikely_bbs_forward): Update.
11299 (determine_unlikely_bbs): Update.
11300 (force_edge_cold): Update.
11301 * profile.c (compute_branch_probabilities): Update.
11302 * reg-stack.c (better_edge): Update.
11303 * shrink-wrap.c (handle_simple_exit): Update.
11304 * tracer.c (better_p): Update.
11305 * trans-mem.c (expand_transaction): Update.
11306 (split_bb_make_tm_edge): Update.
11307 * tree-call-cdce.c: Update.
11308 * tree-cfg.c (gimple_find_sub_bbs): Update.
11309 (gimple_split_edge): Update.
11310 (gimple_duplicate_sese_region): Update.
11311 (gimple_duplicate_sese_tail): Update.
11312 (gimple_flow_call_edges_add): Update.
11313 (insert_cond_bb): Update.
11314 (execute_fixup_cfg): Update.
11315 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11316 * tree-complex.c (expand_complex_div_wide): Update.
11317 * tree-eh.c (lower_resx): Update.
11318 (unsplit_eh): Update.
11319 (cleanup_empty_eh_move_lp): Update.
11320 * tree-inline.c (copy_edges_for_bb): Update.
11321 (freqs_to_counts): Update.
11322 (copy_cfg_body): Update.
11323 * tree-ssa-dce.c (remove_dead_stmt): Update.
11324 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11325 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11326 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11327 (unloop_loops): Update.
11328 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11329 * tree-ssa-loop-split.c (connect_loops): Update.
11330 (split_loop): Update.
11331 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11332 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11333 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11334 * tree-ssa-reassoc.c (branch_fixup): Update.
11335 * tree-ssa-tail-merge.c (replace_block_by): Update.
11336 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11337 (compute_path_counts): Update.
11338 (update_profile): Update.
11339 (recompute_probabilities): Update.
11340 (update_joiner_offpath_counts): Update.
11341 (estimated_freqs_path): Update.
11342 (freqs_to_counts_path): Update.
11343 (clear_counts_path): Update.
11344 (ssa_fix_duplicate_block_edges): Update.
11345 (duplicate_thread_path): Update.
11346 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11347 (case_bit_test_cmp): Update.
11348 (collect_switch_conv_info): Update.
11349 (gen_inbound_check): Update.
11350 (do_jump_if_equal): Update.
11351 (emit_cmp_and_jump_insns): Update.
11352 * tree-tailcall.c (decrease_profile): Update.
11353 (eliminate_tail_call): Update.
11354 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11355 (vect_do_peeling): Update.
11356 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11357 * ubsan.c (ubsan_expand_null_ifn): Update.
11358 (ubsan_expand_ptr_ifn): Update.
11359 * value-prof.c (gimple_divmod_fixed_value): Update.
11360 (gimple_mod_pow2): Update.
11361 (gimple_mod_subtract): Update.
11362 (gimple_ic): Update.
11363 (gimple_stringop_fixed_value): Update.
11365 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11368 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11370 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11372 PR rtl-optimization/82395
11373 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11374 based on non_spilled_static_chain_regno_p.
11376 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11378 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11379 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11380 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11382 2017-10-19 Martin Sebor <msebor@redhat.com>
11384 PR tree-optimization/82596
11385 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11387 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11389 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11390 (get_mem_refs_of_builtin_call): Likewise.
11391 * builtins.c (expand_builtin_apply): Adjust call to
11392 allocate_dynamic_stack_space.
11393 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11394 the third argument to allocate_dynamic_stack_space, otherwise -1.
11395 (expand_builtin): Deal with all alloca variants.
11396 (is_inexpensive_builtin): Likewise.
11397 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11398 * calls.c (special_function_p): Deal with all alloca variants.
11399 (initialize_argument_information): Adjust call to
11400 allocate_dynamic_stack_space.
11401 (expand_call): Likewise.
11402 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11403 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11404 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11405 use it for the stack usage computation.
11406 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11407 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11408 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11409 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11410 (in_loop_p): Remove first argument and useless check.
11411 (pass_walloca::execute): Remove useless test and adjust call to above.
11412 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11413 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11414 (gimplify_call_expr): Deal with all alloca variants.
11415 * hsa-gen.c (gen_hsa_alloca): Likewise.
11416 (gen_hsa_insns_for_call): Likewise.
11417 * ipa-pure-const.c (special_builtin_state): Likewise.
11418 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11419 * tree-object-size.c (alloc_object_size): Likewise.
11420 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11421 (call_may_clobber_ref_p_1): Likewise.
11422 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11423 (ccp_fold_stmt): Likewise.
11424 (optimize_stack_restore): Likewise.
11425 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11426 (mark_all_reaching_defs_necessary_1): Likewise.
11427 (propagate_necessity): Likewise.
11428 (eliminate_unnecessary_stmts): Likewise.
11429 * tree.c (build_common_builtin_nodes): Build
11430 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11431 (build_alloca_call_expr): New function.
11432 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11433 (CASE_BUILT_IN_ALLOCA): Likewise.
11434 (build_alloca_call_expr): Declare.
11435 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11437 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11440 * dwarf2out.c (new_die_raw): New static inline function.
11441 (new_die): Use it to create the DIE.
11442 (add_AT_external_die_ref): Likewise.
11443 (clone_die): Likewise.
11444 (clone_as_declaration): Likewise.
11445 (dwarf2out_vms_debug_main_pointer): Likewise.
11446 (base_type_die): Likewise. Remove early return for corner cases.
11447 Do not call add_pubtype on the DIE here.
11448 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11449 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11450 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11451 native order exists for base types, attach the DIE manually and call
11452 add_pubtype on it. Do not equate a reverse order DIE to the type.
11454 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11456 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11457 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11459 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11461 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11463 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11464 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11465 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11466 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11467 * opts.c (sanitizer_opts): Add builtin.
11468 * ubsan.c (instrument_builtin): New function.
11469 (pass_ubsan::execute): Call it.
11470 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11471 * doc/invoke.texi: Document -fsanitize=builtin.
11473 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11474 builtins, store max (log2 (align), 0) into uchar field instead of
11475 align into uptr field.
11476 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11477 store uchar 0 field instead of uptr 0 field.
11478 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11479 instead of passing one address of struct with 2 locations pass
11480 two addresses of structs with 1 location each.
11481 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11482 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11483 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11484 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11485 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11486 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11487 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11488 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11490 2017-10-19 Martin Liska <mliska@suse.cz>
11493 * file-find.c (remove_prefix): Remove.
11494 * file-find.h (remove_prefix): Likewise.
11495 * gcc-ar.c: Remove smartness of lookup.
11497 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11499 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11500 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11501 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11503 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11506 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11507 (setcc + and to xor + setcc): New peephole2.
11509 2017-10-19 Tom de Vries <tom@codesourcery.com>
11511 * doc/sourcebuild.texi (Test Directives, Variants of
11512 dg-require-support): Add dg-require-stack-size.
11514 2017-10-19 Martin Liska <mliska@suse.cz>
11517 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11518 that have a large alignment.
11519 (gimplify_target_expr): Likewise.
11521 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11523 PR rtl-optimization/82602
11524 * ira.c (rtx_moveable_p): Return false for volatile asm.
11526 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11529 * config/i386/i386-modes.def (CCGZ): New CC mode.
11530 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11531 * config/i386/predicates.md (ix86_comparison_operator):
11533 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11534 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11535 with double-word subtraction.
11536 (put_condition_code): Handle CCGZmode.
11538 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11540 * wide-int.cc (debug (const wide_int &)): New.
11541 (debug (const wide_int *)): New.
11542 (debug (const widest_int &)): New.
11543 (debug (const widest_int *)): New.
11545 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11547 PR middle-end/82556
11548 * lra-constraints.c (curr_insn_transform): Use non-input operand
11549 instead of output one for matched reload.
11551 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11553 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11554 (tree-ssa-loop-ivopts.h): New header file.
11555 (struct builtin_info): New fields.
11556 (classify_builtin_1): Compute and record base and offset parts for
11557 memset builtin partition by calling strip_offset.
11558 (offset_cmp, fuse_memset_builtins): New functions.
11559 (finalize_partitions): Fuse adjacent memset partitions by calling
11561 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11562 Expose the interface.
11563 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11565 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11567 PR tree-optimization/82574
11568 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11569 that data reference must be executed exactly once per iteration
11570 against the outermost loop in nest.
11571 (classify_partition): Update call to above function.
11573 2017-10-18 Richard Biener <rguenther@suse.de>
11575 PR tree-optimization/82591
11576 * graphite.c (graphite_transform_loops): Move code gen message
11578 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11579 Here. Handle scop_to_isl_ast failing.
11580 (scop_to_isl_ast): Limit the number of ISL operations.
11582 2017-10-18 Richard Biener <rguenther@suse.de>
11584 * graphite-isl-ast-to-gimple.c
11585 (translate_isl_ast_to_gimple::set_rename): Simplify.
11586 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11587 (graphite_copy_stmts_from_block): ... here.
11588 (copy_bb_and_scalar_dependences): Simplify.
11589 (add_parameters_to_ivs_params): Canonicalize.
11590 (generate_entry_out_of_ssa_copies): Simplify.
11591 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11592 by passing in ISL dimension.
11593 (parameter_index_in_region_1): Rename to ...
11594 (parameter_index_in_region): ... this.
11595 (extract_affine): Adjust assert, pass down parameter index.
11596 (add_param_constraints): Use range-info when available.
11597 (build_scop_context): Adjust.
11598 * sese.c (new_sese_info): Adjust.
11599 (free_sese_info): Likewise.
11600 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11601 Remove unused typedefs.
11602 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11604 2017-10-18 Martin Liska <mliska@suse.cz>
11606 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11608 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11610 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11611 (s390_sched_init): Do not reset s390_sched_state if we entered the
11612 current basic block via a fallthru edge and all others are unlikely.
11614 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11616 * config/s390/s390.c (NUM_SIDES): New variable.
11617 (LONGRUNNING_THRESHOLD): New variable.
11618 (LATENCY_FACTOR): New variable.
11619 (s390_sched_score): Decrease score for long-running instructions on
11621 (s390_sched_variable_issue): Perform bookkeeping for long-running
11624 2017-10-18 Richard Biener <rguenther@suse.de>
11626 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11627 Simplify with removal of the parameter rename map.
11628 (set_rename): Likewise.
11629 (should_copy_to_new_region): Likewise.
11630 (graphite_copy_stmts_from_block): Likewise.
11631 (copy_bb_and_scalar_dependences): Remove initialization of
11632 unused copied_bb_map.
11633 (copy_def): Remove.
11634 (copy_internal_parameters): Likewise.
11635 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11636 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11637 Use INTEGRAL_TYPE_P.
11638 (parameter_index_in_region_1): Rename to ...
11639 (assign_parameter_index_in_region): ... this. Assert we have
11640 a parameter we handle.
11641 (scan_tree_for_params): Adjust.
11642 * sese.h (parameter_rename_map_t): Remove.
11643 (struct sese_info_t): Remove unused parameter_rename_map and
11644 copied_bb_map members.
11645 * sese.c (new_sese_info): Adjust.
11646 (free_sese_info): Likewise.
11648 2017-10-18 Martin Liska <mliska@suse.cz>
11651 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11652 on an abnormal edge.
11654 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11656 * doc/invoke.texi (ffunction-sections and fdata-sections):
11659 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11661 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11662 the use statement can throw internally.
11664 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11666 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11667 any RTX present on the RHS of a SET.
11668 * compare-elim.c (try_eliminate_compare): Restore comment.
11670 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11672 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11673 may be also called on tcc_type nodes.
11674 * langhooks.c (lhd_tree_size): Likewise.
11676 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11678 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11679 format_warning_at_substring.
11680 (maybe_warn): Convert source_range * param to a location_t. Pass
11681 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11682 (format_directive): Remove code to extract source_ranges and
11683 source_range * in favor of just a location_t.
11684 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11686 * substring-locations.c (format_warning_va): Convert
11687 source_range * param to a location_t.
11688 (format_warning_at_substring): Likewise.
11689 * substring-locations.h (format_warning_va): Likewise.
11690 (format_warning_at_substring): Likewise.
11692 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11694 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11696 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11697 and scatter/gather ops.
11699 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11700 vec_gather_load and vec_scatter_store.
11701 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11702 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11703 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11704 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11705 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11706 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11708 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11710 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11711 Detect FTST instruction and handle its register pops. Only pop
11712 second operand if can_pop_second_op is true.
11713 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
11714 set can_pop_second_op to false in the compare_for_stack_reg call.
11716 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
11717 output_fp_compare for stack register operands.
11718 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
11719 instructions here. Do not emit stack register pops here. Assert
11720 that FCOMPP pops next to top stack register. Rewrite function.
11722 2017-10-17 Nathan Sidwell <nathan@acm.org>
11724 PR middle-end/82577
11725 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
11726 use DECL_ASSEMBLER_NAME_RAW.
11728 PR middle-end/82546
11729 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
11732 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
11733 Wilco Dijkstra <wilco.dijkstra@arm.com>
11735 * builtins.c (expand_builtin_update_setjmp_buf): Add a
11736 converstion to Pmode from the buf_addr.
11738 2017-10-17 Richard Biener <rguenther@suse.de>
11740 * graphite-dependences.c (scop_get_reads_and_writes): Change
11741 output parameters to references.
11743 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11745 PR 71026/tree-optimization
11746 * fold-const.c (distribute_real_division): Removed.
11747 (fold_binary_loc): Remove calls to distribute_real_divison.
11749 2017-10-17 Richard Biener <rguenther@suse.de>
11751 * graphite-scop-detection.c
11752 (scop_detection::stmt_has_simple_data_refs_p): Always use
11753 the full nest as region.
11754 (try_generate_gimple_bb): Likewise.
11755 * sese.c (scalar_evolution_in_region): Simplify now that
11756 SCEV can handle instantiation in regions.
11757 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11758 in the non-loop part of a function if requested.
11760 2017-10-17 Richard Biener <rguenther@suse.de>
11762 PR tree-optimization/82563
11763 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11765 (graphite_regenerate_ast_isl): Call it.
11766 * graphite-scop-detection.c (build_scops): Remove entry edge split.
11768 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11770 PR tree-optimization/82549
11771 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
11772 Formatting fixes. Instead of calling make_bit_field_ref with negative
11775 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11777 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11778 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
11779 _mm_maskz_reduce_ss): New.
11780 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
11781 __builtin_ia32_reducess_mask): Ditto..
11782 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
11783 * config/i386/sse.md (reduces<mode>): Renamed to ...
11784 (reduces<mode><mask_scalar_name>): ... this.
11785 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
11787 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
11788 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
11790 2017-10-16 David Malcolm <dmalcolm@redhat.com>
11792 * Makefile.in (OBJS): Add unique-ptr-tests.o.
11793 * selftest-run-tests.c (selftest::run_tests): Call
11794 selftest::unique_ptr_tests_cc_tests.
11795 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
11796 * unique-ptr-tests.cc: New file.
11798 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11801 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11803 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11804 (make_hard_regno_born, make_hard_regno_dead): Update
11805 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11807 2017-10-16 Jeff Law <law@redhat.com>
11809 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11811 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11813 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11815 2017-10-16 Olivier Hainque <hainque@adacore.com>
11817 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11818 with_cpu if we were configured for an e500v2 target cpu name.
11820 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11822 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11823 * doc/invoke.texi: Document +nodsp as a valid extension for
11826 2017-10-16 Martin Liska <mliska@suse.cz>
11828 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11829 (test_set_range): Likewise.
11830 (test_range_functions): Rename to ...
11831 (test_bit_in_range): ... this.
11832 (sbitmap_c_tests): Add new test.
11834 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11836 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11838 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11839 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11841 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11843 * config/aarch64/aarch64-builtins.c
11844 (aarch64_types_quadopu_lane_qualifiers): New.
11845 (TYPES_QUADOPU_LANE): New.
11846 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11847 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11848 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11849 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11850 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11851 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11852 (Vdottype, DOTPROD): New.
11853 (sur): Add SDOT and UDOT.
11855 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11857 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11858 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11859 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11860 Add TARGET_DOTPROD.
11861 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11862 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11863 Enable TARGET_DOTPROD.
11864 (cortex-a75.cortex-a55): Likewise.
11865 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11867 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11869 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11870 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11872 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11874 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11875 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11876 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11877 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11878 * config/arm/types.md (neon_dot, neon_dot_q): New.
11879 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11881 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11883 * config/arm/arm.h (TARGET_DOTPROD): New.
11884 * config/arm/arm.c (arm_arch_dotprod): New.
11885 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11886 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11887 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11888 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11889 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11890 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11891 * doc/invoke.texi (armv8.2-a): Document dotprod
11893 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11895 * i386.c (ix86_vec_cost): New function.
11896 (ix86_rtx_costs): Handle vector operations better.
11897 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11898 * x86-tune-costs.h: Add new costs to all tables.
11900 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11902 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11904 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11905 divsd, sqrtss and sqrtsd
11906 * x86-tune-costs.h: Add new entries to all costs.
11907 (znver1_cost): Fix to match real instruction latencies.
11909 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11910 Michael Collison <michael.collison@arm.com>
11912 * compare-elim.c: Include emit-rtl.h.
11913 (can_merge_compare_into_arith): New function.
11914 (try_validate_parallel): Likewise.
11915 (try_merge_compare): Likewise.
11916 (try_eliminate_compare): Call the above when no previous clobber
11918 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11921 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11923 PR middle-end/62263
11924 PR middle-end/82498
11925 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11926 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11928 PR middle-end/62263
11929 PR middle-end/82498
11930 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11931 to be any operand_equal_p operands. For & (B - 1) require
11932 B to be power of 2. Recognize
11933 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11935 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11938 * optabs.c (expand_memory_blockage): Fix call of
11939 targetm.have_memory_blockage.
11941 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11944 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11945 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11946 objects to extra_objs instead of overwriting it.
11948 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11950 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11951 Use any_fp_register_operand as operand[3] predicate. Simplify
11952 equality test for operands[2] and operands[4] memory location.
11953 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11954 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11955 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11956 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11957 any_fp_register_operand as operand[1] predicate. Simplify
11958 equality test for operands[0] and operands[3] memory location.
11959 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11960 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11961 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11963 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11965 * target-insns.def: Add memory_blockage.
11966 * optabs.c (expand_memory_blockage): New function.
11967 (expand_asm_memory_barrier): Rename ...
11968 (expand_asm_memory_blockage): ... to this.
11969 (expand_mem_thread_fence): Call expand_memory_blockage
11970 instead of expand_asm_memory_barrier.
11971 (expand_mem_singnal_fence): Ditto.
11972 (expand_atomic_load): Ditto.
11973 (expand_atomic_store): Ditto.
11974 * doc/md.texi (Standard Pattern Names For Generation):
11975 Document memory_blockage instruction pattern.
11977 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11979 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11980 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11981 * config/rl78/rl78.md: New define_expand "adddi3".
11983 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11985 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11988 2017-10-13 Jeff Law <law@redhat.com>
11990 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11992 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11995 * internal-fn.c (expand_mul_overflow): If both operands have
11996 the same highpart of -1 or 0 and the topmost bit of lowpart
11997 is different, overflow is if res <= 0 rather than res < 0.
11999 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12001 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12002 TARGET_P9_VECTOR code for unaligned_load case.
12004 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12006 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12008 2017-10-13 Nathan Sidwell <nathan@acm.org>
12010 * tree-core.h (tree_contains_struct): Make bool.
12011 * tree.c (tree_contains_struct): Likewise.
12012 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12013 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12014 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12015 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12016 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12018 2017-10-13 Richard Biener <rguenther@suse.de>
12020 * graphite-isl-ast-to-gimple.c
12021 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12022 parameters and dominance check.
12023 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12024 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12025 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12026 Do not update SSA form here or do intermediate IL verification.
12027 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12028 (graphite_initialize): Remove check on the number of loops in
12029 the function and inline into graphite_transform_loops.
12030 (graphite_finalize): Inline into graphite_transform_loops.
12031 (graphite_transform_loops): Perform SSA update and IL verification
12033 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12035 2017-10-13 Richard Biener <rguenther@suse.de>
12037 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12038 graphite_expression_type_precision): Avoid global constructor
12040 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12041 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12042 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12043 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12044 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12045 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12047 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12050 * config/i386/i386.h (ix86_red_zone_size): New.
12051 * config/i386/i386.md (push peephole2s): Replace
12052 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12054 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12055 Alan Hayward <alan.hayward@arm.com>
12056 David Sherwood <david.sherwood@arm.com>
12058 * combine.c (can_change_dest_mode): Reject changes in
12059 REGMODE_NATURAL_SIZE.
12061 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12062 Alan Hayward <alan.hayward@arm.com>
12063 David Sherwood <david.sherwood@arm.com>
12065 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12066 (expand_debug_source_expr): Likewise.
12067 * combine.c (combine_simplify_rtx): Likewise.
12068 * cse.c (fold_rtx): Likewise.
12069 * fwprop.c (canonicalize_address): Likewise.
12070 * targhooks.c (default_shift_truncation_mask): Likewise.
12072 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12073 Alan Hayward <alan.hayward@arm.com>
12074 David Sherwood <david.sherwood@arm.com>
12076 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12077 (widened_mode): Likewise.
12078 (expand_unop): Likewise.
12079 * ree.c (transform_ifelse): Likewise.
12080 (merge_def_and_ext): Likewise.
12081 (combine_reaching_defs): Likewise.
12082 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12084 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12085 Alan Hayward <alan.hayward@arm.com>
12086 David Sherwood <david.sherwood@arm.com>
12088 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12089 * combine.c (gen_lowpart_for_combine): Likewise.
12090 * dwarf2out.c (rtl_for_decl_location): Likewise.
12091 * final.c (alter_subreg): Likewise.
12092 * rtlhooks.c (gen_lowpart_general): Likewise.
12093 (gen_lowpart_if_possible): Likewise.
12095 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12096 Alan Hayward <alan.hayward@arm.com>
12097 David Sherwood <david.sherwood@arm.com>
12099 * calls.c (expand_call): Use subreg_lowpart_offset.
12100 * cse.c (cse_insn): Likewise.
12101 * regcprop.c (copy_value): Likewise.
12102 (copyprop_hardreg_forward_1): Likewise.
12104 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12107 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12108 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12109 =Q constraints to +Q and into insn condition add check
12110 that operands[0] and operands[1] are equal.
12111 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12112 =Q constraints to +Q and into insn condition add check
12113 that operands[0] is equal to either operands[1] or operands[2].
12116 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12117 instead of handling MINUS_EXPR twice (once for each argument),
12118 canonicalize operand order and handle just once, use rtype where
12119 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12122 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12123 any values of __C while still being pattern recognizable as a simple
12124 rotate instruction.
12126 2017-10-13 Richard Biener <rguenther@suse.de>
12128 PR tree-optimization/82451
12130 2017-10-02 Richard Biener <rguenther@suse.de>
12132 PR tree-optimization/82355
12133 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12134 a mapping for the enclosing loop but avoid generating one for
12135 the loop tree root.
12136 (copy_bb_and_scalar_dependences): Remove premature codegen
12137 error on PHIs in blocks duplicated into multiple places.
12138 * graphite-scop-detection.c
12139 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12140 in the region use it as loop and nest to analyze the DR in.
12141 (try_generate_gimple_bb): Likewise.
12142 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12143 (add_loop_constraints): For blocks in a loop not in the region
12144 create a dimension with a single iteration.
12145 * sese.h (gbb_loop_at_index): Remove assert.
12147 * cfgloop.c (loop_preheader_edge): For the loop tree root
12148 return the single successor of the entry block.
12149 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12150 Reset the SCEV hashtable and niters.
12151 * graphite-scop-detection.c
12152 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12153 assert that we only have POLYNOMIAL_CHREC that vary in loops
12154 contained in the region.
12155 (scop_detection::graphite_can_represent_expr): Adjust.
12156 (scop_detection::stmt_has_simple_data_refs_p): For loops
12157 not in the region set loop to NULL. The nest is now the
12158 entry edge to the region.
12159 (try_generate_gimple_bb): Likewise.
12160 * sese.c (scalar_evolution_in_region): Adjust for
12161 instantiate_scev change.
12162 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12163 Make nest parameter the edge into the region.
12164 (create_data_ref): Likewise.
12165 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12166 entry edge into a region and adjust instantiate_scev calls.
12167 (create_data_ref): Likewise.
12168 (graphite_find_data_references_in_stmt): Likewise.
12169 (find_data_references_in_stmt): Pass the loop preheader edge
12170 from the nest argument.
12171 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12172 parameter the edge into the region.
12173 (instantiate_parameters): Use the loop preheader edge as entry.
12174 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12176 (get_instantiated_value_entry): Make instantiate_below parameter
12177 the edge into the region.
12178 (instantiate_scev_name): Likewise. Adjust dominance checks,
12179 when we cannot use loop-based instantiation instantiate by
12180 walking use-def chains.
12181 (instantiate_scev_poly): Adjust.
12182 (instantiate_scev_binary): Likewise.
12183 (instantiate_scev_convert): Likewise.
12184 (instantiate_scev_not): Likewise.
12185 (instantiate_array_ref): Remove.
12186 (instantiate_scev_3): Likewise.
12187 (instantiate_scev_2): Likewise.
12188 (instantiate_scev_1): Likewise.
12189 (instantiate_scev_r): Do not blindly handle N-operand trees.
12190 Do not instantiate array-refs. Handle all constants and invariants.
12191 (instantiate_scev): Make instantiate_below parameter
12192 the edge into the region.
12193 (resolve_mixers): Use the loop preheader edge for the region
12194 parameter to instantiate_scev_r.
12195 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12197 2017-10-13 Richard Biener <rguenther@suse.de>
12199 PR tree-optimization/82525
12200 * graphite-isl-ast-to-gimple.c
12201 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12203 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12204 Fail code generation when we cannot represent the isl integer.
12205 (binary_op_to_tree): Elide modulo operations that are no-ops
12206 in the type we code generate. Remove now superfluous code
12209 2017-10-13 Richard Biener <rguenther@suse.de>
12211 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12212 (scop_detection::harmful_loop_in_region): Remove premature
12213 IV type restriction.
12214 (scop_detection::graphite_can_represent_scev): We can handle
12215 pointer IVs just fine.
12217 2017-10-13 Alan Modra <amodra@gmail.com>
12219 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12220 "Clobbers and Scratch Registers". Add paragraph on
12221 alternative to clobbers for scratch registers and OpenBLAS
12224 2017-10-13 Alan Modra <amodra@gmail.com>
12226 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12227 example of a memory input for a string of known length. Move
12228 commentary out of table. Add a number of new examples
12229 covering array memory inputs.
12231 2017-10-12 Martin Liska <mliska@suse.cz>
12233 PR tree-optimization/82493
12234 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12235 (test_range_functions): New function.
12236 (sbitmap_c_tests): Likewise.
12237 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12238 * selftest.h (sbitmap_c_tests): New function.
12240 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12242 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12244 * config/rs6000/amo.h: Fix spacing issue.
12246 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12249 * config/i386/i386.md (*ashl<mode>3_mask_1,
12250 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12251 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12254 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12256 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12257 (profile_probability): Set max_probability
12258 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12260 * profile-count.c (profile_probability::differs_from_p): Do not
12261 rely on max_probaiblity == 10000
12263 2017-10-12 Jeff Law <law@redhat.com>
12265 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12268 2017-10-12 Martin Sebor <msebor@redhat.com>
12272 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12273 (handle_alias_pairs): Call it.
12274 * common.opt (-Wattribute-alias): New option.
12275 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12276 * doc/invoke.texi (-Wattribute-alias): Document.
12278 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12281 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12283 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12285 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12286 (make_hard_regno_born, make_hard_regno_dead): Update
12287 bb_killed_pseudos and bb_gen_pseudos.
12289 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12291 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12293 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12295 * config/alpha/alpha.c (alpha_split_conditional_move):
12296 Use std::swap instead of manually swapping.
12297 (alpha_stdarg_optimize_hook): Ditto.
12298 (alpha_canonicalize_comparison): Ditto.
12300 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12302 * tree-loop-distribution.c (struct builtin_info): New struct.
12303 (struct partition): Refactor fields into struct builtin_info.
12304 (partition_free): Free struct builtin_info.
12305 (build_size_arg_loc, build_addr_arg_loc): Delete.
12306 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12307 information from struct builtin_info.
12308 (find_single_drs): New function refactored from classify_partition.
12309 Also moved builtin validity checks to this function.
12310 (compute_access_range, alloc_builtin): New functions.
12311 (classify_builtin_st, classify_builtin_ldst): New functions.
12312 (classify_partition): Refactor code into functions find_single_drs,
12313 classify_builtin_st and classify_builtin_ldst.
12314 (distribute_loop): Don't do runtime alias check when distributing
12316 (find_seed_stmts_for_distribution): New function.
12317 (pass_loop_distribution::execute): Refactor code finding seed
12318 stmts into above function. Support distribution for the innermost
12319 two-level loop nest. Adjust dump information.
12321 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12323 * tree-loop-distribution.c: Adjust the general comment.
12324 (NUM_PARTITION_THRESHOLD): New macro.
12325 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12326 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12327 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12328 in call to build_partition_graph.
12329 (finalize_partitions): New parameter. Make loop distribution more
12330 conservative by fusing more partitions.
12331 (distribute_loop): Don't do runtime alias check in case of loop nest
12333 (find_seed_stmts_for_distribution): New function.
12334 (prepare_perfect_loop_nest): New function.
12335 (pass_loop_distribution::execute): Refactor code finding seed stmts
12336 and loop nest into above functions. Support loop nest distribution.
12337 Adjust dump information accordingly.
12339 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12341 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12342 and set PTYPE_SEQUENTIAL for merged partition.
12344 2017-10-12 Richard Biener <rguenther@suse.de>
12346 PR tree-optimization/69728
12348 2017-09-19 Richard Biener <rguenther@suse.de>
12350 PR tree-optimization/69728
12351 * graphite-sese-to-poly.c (schedule_error): New global.
12352 (add_loop_schedule): Handle empty domain by failing the
12354 (build_original_schedule): Handle schedule_error.
12356 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12357 domain by returning an unchanged schedule.
12359 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12361 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12362 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12364 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12366 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12369 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12372 * expr.c (store_field): Don't optimize away bitsize == 0 store
12373 from CALL_EXPR with addressable return type.
12375 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12377 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12378 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12379 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12380 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12381 TARGET_ISEL instead of TARGET_ISEL<sel>.
12383 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12385 * config/rs6000/rs6000.c
12386 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12388 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12390 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12392 (reg_or_cint_operand): Fix comment.
12393 (reg_or_zero_operand): New predicate.
12394 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12395 * config/rs6000/rs6000.c (output_isel): Delete.
12396 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12397 instead of reg_or_cint_operand. Output instruction directly (not via
12399 (isel_unsigned_<mode>): Ditto.
12400 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12401 gpc_reg_operand. Add an instruction alternative for this. Output
12402 instruction directly.
12403 (*isel_reversed_unsigned_<mode>): Ditto.
12405 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12407 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12408 (TARGET_CANONICALIZE_COMPARISON): Define.
12410 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12413 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12414 Check whether the dest is REG before adding REG_EQUIV note.
12416 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12419 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12421 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12422 (make_hard_regno_born, make_hard_regno_dead): Update
12423 bb_killed_pseudos and bb_gen_pseudos.
12425 2017-10-11 Nathan Sidwell <nathan@acm.org>
12427 * incpath.h (enum incpath_kind): Name enum, prefix values.
12428 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12429 * incpath.c (heads, tails): Use INC_MAX.
12430 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12431 (merge_include_chains, split_quote_chain,
12432 register_include_chains): Update incpath_kind names.
12433 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12434 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12436 (add_framework_path, darwin_register_objc_includes): Likewise.
12437 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12439 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12441 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12442 Do not use float_operator operator predicate.
12443 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12444 * config/i386/predicates.md (float_operator): Remove predicate.
12446 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12448 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12449 (*jccxf_i387): Ditto.
12450 (*jcc<mode>_i387): Ditto.
12451 (*jccu<mode>_i387): Ditto.
12452 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12453 (*jcc_*_i387 splitters): Remove.
12454 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12455 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12456 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12459 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12461 * profile-count.h (slow_safe_scale_64bit): New function.
12462 (safe_scale_64bit): New inline.
12463 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12464 * profile-count.c: Include wide-int.h
12465 (slow_safe_scale_64bit): New.
12467 2017-10-11 Nathan Sidwell <nathan@acm.org>
12469 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12470 HAS_DECL_ASSEMBLER_NAME_P.
12471 * gimple-expr.c (gimple_decl_printable_name: Check
12472 HAS_DECL_ASSEMBLER_NAME_P too.
12473 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12474 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12475 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12476 * passes.c (rest_of_decl_compilation): Check
12477 HAS_DECL_ASSEMBLER_NAME_P too.
12478 * recog.c (verify_changes): Likewise.
12479 * tree-pretty-print.c (dump_decl_name): Likewise.
12480 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12482 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12483 (SET_DECL_ASSEMBLER_NAME): Use it.
12484 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12485 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12486 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12488 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12490 * config.gcc (i386, x86_64): Add extra objects.
12491 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12492 (ix86_min_insn_size): Declare.
12493 (ix86_issue_rate): Declare.
12494 (ix86_adjust_cost): Declare.
12495 (ia32_multipass_dfa_lookahead): Declare.
12496 (ix86_macro_fusion_p): Declare.
12497 (ix86_macro_fusion_pair_p): Declare.
12498 (ix86_bd_has_dispatch): Declare.
12499 (ix86_bd_do_dispatch): Declare.
12500 (ix86_core2i7_init_hooks): Declare.
12501 (ix86_atom_sched_reorder): Declare.
12502 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12503 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12504 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12505 (rip_relative_addr_p): Rename to ...
12506 (ix86_rip_relative_addr_p): ... this one; export.
12507 (memory_address_length): Update.
12508 (ix86_issue_rate): Move to x86-tune-sched.c.
12509 (ix86_flags_dependent): Move to x86-tune-sched.c.
12510 (ix86_agi_dependent): Move to x86-tune-sched.c.
12511 (exact_dependency_1): Move to x86-tune-sched.c.
12512 (exact_store_load_dependency): Move to x86-tune-sched.c.
12513 (ix86_adjust_cost): Move to x86-tune-sched.c.
12514 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12515 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12516 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12517 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12518 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12519 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12520 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12521 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12522 (min_insn_size): Rename to ...
12523 (ix86_min_insn_size): ... this one; export.
12524 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12525 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12526 (core2i7_first_cycle_multipass_backtrack): Move to
12527 x86-tune-sched-core.c.
12528 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12529 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12530 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12531 (ix86_avoid_jump_mispredicts): Update.
12532 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12533 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12534 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12535 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12536 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12537 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12538 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12539 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12540 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12541 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12542 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12543 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12544 (BIG): Move to ix86-tune-sched-bd.c.
12545 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12546 (enum insn_path): Move to ix86-tune-sched-bd.c.
12547 (get_mem_group): Move to ix86-tune-sched-bd.c.
12548 (is_cmp): Move to ix86-tune-sched-bd.c.
12549 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12550 (is_branch): Move to ix86-tune-sched-bd.c.
12551 (is_prefetch): Move to ix86-tune-sched-bd.c.
12552 (init_window): Move to ix86-tune-sched-bd.c.
12553 (allocate_window): Move to ix86-tune-sched-bd.c.
12554 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12555 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12556 (process_end_window): Move to ix86-tune-sched-bd.c.
12557 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12558 (find_constant): Move to ix86-tune-sched-bd.c.
12559 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12560 (has_immediate): Move to ix86-tune-sched-bd.c.
12561 (get_insn_path): Move to ix86-tune-sched-bd.c.
12562 (get_insn_group): Move to ix86-tune-sched-bd.c.
12563 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12564 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12565 (add_insn_window): Move to ix86-tune-sched-bd.c.
12566 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12567 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12568 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12569 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12570 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12571 (do_dispatch): Move to ix86-tune-sched-bd.c.
12572 (has_dispatch): Move to ix86-tune-sched-bd.c.
12573 * config/i386/t-i386: Add new object files.
12574 * config/i386/x86-tune-costs.h: New file.
12575 * config/i386/x86-tune-sched-atom.c: New file.
12576 * config/i386/x86-tune-sched-bd.c: New file.
12577 * config/i386/x86-tune-sched-core.c: New file.
12578 * config/i386/x86-tune-sched.c: New file.
12580 2017-10-11 Liu Hao <lh_mouse@126.com>
12582 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12583 the generic version below instead.
12584 (should_colorize): Recognize Windows consoles as terminals
12586 * pretty-print.c [__MINGW32__] (write_all): New function.
12587 [__MINGW32__] (find_esc_head): Likewise.
12588 [__MINGW32__] (find_esc_terminator): Likewise.
12589 [__MINGW32__] (eat_esc_sequence): Likewise.
12590 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12592 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12595 2017-10-11 Richard Biener <rguenther@suse.de>
12597 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12598 Properly call analyze_scalar_evolution with the loop of the stmt.
12600 2017-10-11 Richard Biener <rguenther@suse.de>
12602 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12603 * tree-core.h (tree_base): Add chrec_var union member.
12604 * tree.h (CHREC_VAR): Remove.
12605 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12606 * tree-chrec.h (build_polynomial_chrec): Adjust.
12607 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12608 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12610 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12612 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12613 * match.pd: ... here.
12614 ((T) X == (T) Y): Relax condition.
12616 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12618 PR tree-optimization/82472
12619 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12621 (break_alias_scc_partitions): Update postorder number.
12623 2017-10-11 Martin Liska <mliska@suse.cz>
12626 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12628 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12630 2017-10-11 Martin Liska <mliska@suse.cz>
12632 * print-rtl.c (print_insn): Move declaration of idbuf
12633 to same scope as name.
12635 2017-10-11 Martin Liska <mliska@suse.cz>
12640 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12641 volatile arguments.
12643 2017-10-11 Martin Liska <mliska@suse.cz>
12646 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12647 volatile arguments.
12649 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12651 * config.gcc (default_gnu_indirect_function): Default to yes for
12652 arm*-*-linux* with glibc.
12654 2017-10-11 Richard Biener <rguenther@suse.de>
12656 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12657 default-defs and types we do not want to analyze.
12658 (interpret_loop_phi): Replace unreachable code with an assert.
12659 (compute_scalar_evolution_in_loop): Remove and inline ...
12660 (analyze_scalar_evolution_1): ... here, replacing condition with
12661 what makes the intent clearer. Remove handling of cases
12662 get_scalar_evolution now handles.
12664 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12666 PR rtl-optimization/81434
12667 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12668 comment for main loop. In sched_group_found if, also add checks for
12669 pass and min_cost_group.
12671 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12673 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12674 (rs6000_insn_cost): New function.
12675 * config/rs6000/rs6000.md (cost): New attribute.
12677 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12678 H.J. Lu <hongjiu.lu@intel.com>
12682 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12683 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12684 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12685 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12686 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12687 OPTION_MASK_ISA_MMX and some other bit set require both
12688 mmx and the ISAs without the mmx bit.
12689 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12690 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12691 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12692 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12693 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12694 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12695 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12696 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12697 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12698 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12699 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12700 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12701 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12702 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12703 Add OPTION_MASK_ISA_MMX.
12705 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12707 * config.gcc (armv7*-*-freebsd*): New target.
12708 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12710 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12712 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
12713 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
12716 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
12718 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
12719 a template parameter.
12720 (WIDE_INT_REF_FOR): Update accordingly.
12721 * tree.h (wi::int_traits <const_tree>): Delete.
12722 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
12723 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
12724 (wi::tree_to_wide_ref): New typedef.
12725 (wi::to_wide): New function.
12726 * calls.c (get_size_range): Use wi::to_wide when operating on
12727 trees as wide_ints.
12728 * cgraph.c (cgraph_node::create_thunk): Likewise.
12729 * config/i386/i386.c (ix86_data_alignment): Likewise.
12730 (ix86_local_alignment): Likewise.
12731 * dbxout.c (stabstr_O): Likewise.
12732 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
12733 * expr.c (const_vector_from_tree): Likewise.
12734 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
12735 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
12736 (fold_negate_expr_1, int_const_binop_1, const_binop)
12737 (fold_convert_const_int_from_real, optimize_bit_field_compare)
12738 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
12739 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
12740 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12741 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12742 (fold_not_const, round_up_loc): Likewise.
12743 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12744 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12745 (alloca_call_type): Likewise.
12746 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12747 * godump.c (go_output_typedef): Likewise.
12748 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12749 * internal-fn.c (get_min_precision): Likewise.
12750 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12751 * ipa-polymorphic-call.c
12752 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12753 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12754 (ipa_modify_call_arguments): Likewise.
12755 * match.pd: Likewise.
12756 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12757 * print-tree.c (print_node_brief, print_node): Likewise.
12758 * stmt.c (expand_case): Likewise.
12759 * stor-layout.c (layout_type): Likewise.
12760 * tree-affine.c (tree_to_aff_combination): Likewise.
12761 * tree-cfg.c (group_case_labels_stmt): Likewise.
12762 * tree-data-ref.c (dr_analyze_indices): Likewise.
12763 (prune_runtime_alias_test_list): Likewise.
12764 * tree-dump.c (dequeue_and_dump): Likewise.
12765 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
12766 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
12767 * tree-pretty-print.c (dump_generic_node): Likewise.
12768 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
12769 (simple_iv_with_niters): Likewise.
12770 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
12771 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
12772 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
12773 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12774 (refine_value_range_using_guard, number_of_iterations_ne_max)
12775 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12776 (get_cst_init_from_scev, record_nonwrapping_iv)
12777 (scev_var_range_cant_overflow): Likewise.
12778 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
12779 * tree-ssa-pre.c (compute_avail): Likewise.
12780 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
12781 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
12782 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
12783 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
12784 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
12785 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
12786 (dump_case_nodes, try_switch_expansion): Likewise.
12787 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
12788 (vect_do_peeling): Likewise.
12789 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
12790 * tree-vect-stmts.c (vectorizable_load): Likewise.
12791 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
12792 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
12793 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
12794 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
12795 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
12796 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
12797 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12798 (evrp_dom_walker::before_dom_children): Likewise.
12799 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12800 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12801 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12802 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12803 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12804 (get_range_pos_neg): Likewise.
12805 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12806 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12807 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12808 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12809 * config/avr/avr.c (avr_fold_builtin): Likewise.
12810 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12811 * config/msp430/msp430.c (msp430_attr): Likewise.
12812 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12813 * config/powerpcspe/powerpcspe-c.c
12814 (altivec_resolve_overloaded_builtin): Likewise.
12815 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12816 (rs6000_expand_ternop_builtin): Likewise.
12817 * config/rs6000/rs6000-c.c
12818 (altivec_resolve_overloaded_builtin): Likewise.
12819 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12820 (rs6000_expand_ternop_builtin): Likewise.
12821 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12823 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12825 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12826 when copying loop nest with only one inner loop.
12828 2017-10-10 Richard Biener <rguenther@suse.de>
12830 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12831 blocks if SCEV is active.
12832 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12834 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12835 (scev_initialize): Assert we are not yet initialized.
12837 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12839 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12840 inner loop's exit stmt by making it always exit the loop, otherwise
12841 we would generate an infinite empty loop.
12843 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12845 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12846 renaming variables in new preheader if it's deleted.
12848 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12850 * tree-loop-distribution.c (struct partition): Remove unused field
12851 loops of the structure.
12852 (partition_alloc, partition_free): Ditto.
12853 (build_rdg_partition_for_vertex): Ditto.
12855 2017-10-09 Jeff Law <law@redhat.com>
12857 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12858 return type to match prototype and documentation.
12860 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12862 * config/rs6000/rs6000.c (processor_costs): Move to ...
12863 * config/rs6000/rs6000.h: ... here.
12864 (rs6000_cost): Declare.
12866 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12868 * except.c (setjmp_fn): New global variable.
12869 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12870 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12871 if DONT_USE_BUILTIN_SETJMP is defined.
12873 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12875 * target.def (insn_cost): New hook.
12876 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12877 * doc/tm.texi: Regenerate.
12878 * rtlanal.c (insn_cost): Use the new hook.
12880 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12882 * combine.c (combine_validate_cost): Compute the new insn_cost,
12883 not just pattern_cost.
12884 (try_combine): Adjust comment.
12886 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12888 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12890 * combine.c (uid_insn_cost): Adjust comment.
12891 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12893 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12894 * dse.c (find_shift_sequence): Ditto.
12895 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12896 (bb_valid_for_noce_process_p): Use pattern_cost.
12897 * rtl.h (insn_rtx_cost): Delete.
12898 (pattern_cost): New prototype.
12899 (insn_cost): New prototype.
12900 * rtlanal.c (insn_rtx_cost): Rename to...
12901 (pattern_cost): ... this.
12904 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12906 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12907 (*jcc<mode>_0_r_i387): Ditto.
12908 (*jccxf_r_i387): Ditto.
12909 (*jcc<mode>_r_i387): Ditto.
12910 (*jccu<mode>_r_i387): Ditto.
12911 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12912 (*jcc): Rename from *jcc_1.
12914 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12916 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12917 deferred rescans after the lvx/stvx recombination pre-pass.
12919 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12921 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12922 memory operation instruction support.
12923 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12924 (rs6000-ibm-aix[789]*): Likewise.
12925 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12926 Document new functions.
12928 2017-10-09 Richard Biener <rguenther@suse.de>
12930 PR tree-optimization/82397
12931 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12932 equality only for semantically equal trees.
12934 2017-10-09 Richard Biener <rguenther@suse.de>
12936 PR tree-optimization/82449
12937 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12938 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12939 allow constant addresses.
12940 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12943 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12945 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12948 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12951 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12954 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12957 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12959 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12962 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12963 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12965 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12967 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12968 (WI_BINARY_PREDICATE_RESULT): Likewise.
12969 (wi::binary_traits::operator_result): New type.
12970 (wi::binary_traits::predicate_result): Likewise.
12971 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12972 (generic_wide_int::operator==, generic_wide_int::operator!=)
12973 (generic_wide_int::operator&, generic_wide_int::and_not)
12974 (generic_wide_int::operator|, generic_wide_int::or_not)
12975 (generic_wide_int::operator^, generic_wide_int::operator+
12976 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12978 (operator~, unary operator-, operator==, operator!=, operator&)
12979 (operator|, operator^, operator+, binary operator-, operator*): New
12981 * expr.c (get_inner_reference): Use wi::bit_and_not.
12982 * fold-const.c (fold_binary_loc): Likewise.
12983 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12984 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12985 (bit_value_binop): Likewise.
12986 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12987 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12988 (extract_range_from_binary_expr_1): Likewise.
12989 (masked_increment): Likewise.
12990 (simplify_bit_ops_using_ranges): Likewise.
12992 2017-10-09 Martin Jambor <mjambor@suse.cz>
12995 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12996 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12997 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12999 (hsa_fixup_mov_insn_type): New function.
13000 (hsa_op_with_type::get_in_type): Use it.
13001 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13002 immediates in an assert.
13003 (hsa_op_with_type::extend_int_to_32bit): New method.
13004 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13005 types. Convert to dest type if necessary.
13006 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13007 (reg_for_gimple_ssa): Pass false as min32int to
13008 hsa_type_for_scalar_tree_type.
13009 (gen_hsa_addr): Fixup type when creating addresable temporary.
13010 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13011 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13012 necessary. Call hsa_fixup_mov_insn_type.
13013 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13014 extend operands and convert to dest type if necessary.
13015 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13016 to dest type if necessary.
13017 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13018 if conversion nt necessary and size matches.
13019 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13020 to dest type if necessary.
13021 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13022 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13024 (gen_hsa_clrsb): Likewise.
13025 (gen_hsa_ffs): Likewise.
13026 (gen_hsa_divmod): Extend operands and convert to dest type if
13028 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13030 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13032 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13033 Remove empty default arguments. Use a brace block as output
13035 (conditional return): Ditto.
13037 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13038 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13040 (group_ending_nop): Ditto.
13041 (doloop_end): Ditto.
13042 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13043 (splitters for those): Ditto.
13045 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13047 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13048 a conditional jump (and the compare for it) so that pc_rtx is the
13050 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13051 for the deleted and renamed ctr<mode>_internal[234] patterns.
13052 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13053 Delete second conditional return pattern.
13054 (ctr<mode>_internal2): Delete this second bdnz pattern.
13055 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13056 (ctr<mode>_internal4): Delete this second bdz pattern.
13058 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13060 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13061 (always_initialized_rtx_for_ssa_name_p): New predicate.
13062 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13063 (finish_out_of_ssa): Free new field of SA.
13064 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13065 * tree-ssa-coalesce.c: Include tree-ssa.h.
13066 (get_parm_default_def_partitions): Remove extern keyword.
13067 (get_undefined_value_partitions): New function.
13068 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13069 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13070 uninitialized bits.
13071 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13073 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13075 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13077 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13079 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13080 for targets that preffer 128bit.
13082 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13084 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13086 2017-10-08 Olivier Hainque <hainque@adacore.com>
13088 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13089 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13091 (thumb_set_return_address): Likewise.
13093 2017-10-08 Olivier Hainque <hainque@adacore.com>
13095 * common/config/arm/arm-common.c (arm_except_unwind_info):
13096 Handle DWARF2_UNWIND_INFO.
13098 2017-10-07 Michael Collison <michael.collison@arm.com>
13100 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13103 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13105 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13106 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13107 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13108 defined, force the creation of a new block for a dispatch label.
13110 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13112 * invoke.texi (Wsuggest-attribute=cold): Document.
13113 * common.opt (Wsuggest-attribute=cold): New
13114 * ipa-pure-const.c (warn_function_cold): New function.
13115 * predict.c (compute_function_frequency): Use it.
13116 * predict.h (warn_function_cold): Declare.
13118 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13120 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13123 2017-10-06 Martin Liska <mliska@suse.cz>
13125 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13126 keyword for member functions.
13127 (struct sanopt_tree_couple): New struct.
13128 (struct sanopt_tree_couple_hash): New function.
13129 (struct sanopt_ctx): Add new hash_map.
13130 (has_dominating_ubsan_ptr_check): New function.
13131 (record_ubsan_ptr_check_stmt): Likewise.
13132 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13133 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13134 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13136 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13139 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13140 aarch64_simd_valid_immediate on CONST_VECTORs.
13141 (aarch64_reg_or_bic_imm): Likewise.
13143 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13145 PR rtl-optimization/82396
13146 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13148 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13150 * graphite-dependences.c (scop_get_reads): Move code to...
13151 (scop_get_must_writes): Move code to...
13152 (scop_get_may_writes): Move code to...
13153 (scop_get_reads_and_writes): ... here.
13154 (scop_get_dependences): Call scop_get_reads_and_writes.
13156 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13158 PR tree-optimization/82434
13159 * fold-const.h (can_native_encode_type_p,
13160 can_native_encode_string_p): Remove.
13161 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13162 don't encode anything, just return what would be otherwise returned.
13163 (native_encode_fixed, native_encode_complex, native_encode_vector):
13165 (native_encode_string): Likewise. Inline by hand
13166 can_native_encode_string_p.
13167 (can_native_encode_type_p): Remove.
13168 (can_native_encode_string_p): Remove.
13169 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13170 STRING_CSTs using can_native_encode_string_p, test all
13171 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13172 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13173 argument from native_encode_expr.
13174 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13175 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13178 2017-10-06 Richard Biener <rguenther@suse.de>
13180 PR tree-optimization/82397
13181 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13182 operand_equal_p but rely on data_ref_compare_tree for detecting
13184 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13185 to match up with dr_group_sort_cmp.
13187 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13190 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13192 * config/s390/s390-builtin-types.def: Regenerate.
13194 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13197 * config/s390/s390-builtin-types.def: Regenerate.
13198 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13199 Change flag from B_VXE to B_VX.
13200 (s390_vec_min_dbl): Remove B_VXE flag.
13202 2017-10-06 Richard Biener <rguenther@suse.de>
13204 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13205 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13206 translate_isl_ast_to_gimple::is_valid_rename,
13207 translate_isl_ast_to_gimple::get_rename,
13208 translate_isl_ast_to_gimple::get_def_bb_for_const,
13209 translate_isl_ast_to_gimple::get_new_name,
13210 translate_isl_ast_to_gimple::collect_all_ssa_names,
13211 translate_isl_ast_to_gimple::copy_loop_phi_args,
13212 translate_isl_ast_to_gimple::collect_all_ssa_names,
13213 translate_isl_ast_to_gimple::copy_loop_phi_args,
13214 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13215 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13216 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13217 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13218 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13219 translate_isl_ast_to_gimple::copy_cond_phi_args,
13220 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13221 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13222 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13223 translate_isl_ast_to_gimple::rename_uses,
13224 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13225 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13226 (set_rename_for_each_def): Likewise.
13227 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13228 here. Handle rewriting SCEV analyzable uses here.
13229 (copy_bb_and_scalar_dependences): Generate code for PHI
13231 (graphite_regenerate_ast_isl): Adjust.
13232 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13233 (add_write, add_read): New functions.
13234 (build_cross_bb_scalars_def): Use it and simplify.
13235 (build_cross_bb_scalars_use): Likewise.
13236 (graphite_find_cross_bb_scalar_vars): Inline into...
13237 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13238 simulating out-of-SSA. Compute liveout and add dependencies.
13239 (build_scops): Force an empty entry block.
13240 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13242 (sese_build_liveouts): Declare.
13243 (sese_trivially_empty_bb_p): Likewise.
13244 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13245 compute liveout and debug_liveout.
13246 (sese_bad_liveouts_use): Remove.
13247 (sese_reset_debug_liveouts_bb): Likewise.
13248 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13249 (sese_build_liveouts): Build liveout and debug_liveout and store
13251 (new_sese_info): Adjust.
13252 (free_sese_info): Likewise.
13253 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13254 do not build liveout here.
13255 (move_sese_in_condition): Adjust region entry.
13256 (scev_analyzable_p): Match up with chrec_apply requirements.
13257 (sese_trivially_empty_bb_p): New.
13258 * tree-into-ssa.c (get_reaching_def): Properly support generating
13259 default-defs for incremental rewrite of anonymous names.
13261 2017-10-06 Richard Biener <rguenther@suse.de>
13263 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13264 precision do not perform modulo reduction.
13266 2017-10-06 Richard Biener <rguenther@suse.de>
13268 PR tree-optimization/82436
13269 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13270 conservatively choose the vectorization factor when checking
13271 whether we can perform the required load permutation.
13272 (vect_transform_slp_perm_load): Assert when we may not fail.
13274 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13276 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13277 message for incompatible -msdata=* and -mcall-* options.
13279 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13281 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13282 rate for post-reload scheduling.
13284 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13286 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13288 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13290 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13291 to improve monte carlo in scimark.
13293 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13295 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13296 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13297 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13298 pentium4_cost, nocona_cost): Set reassociation width to 1.
13299 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13300 width to 2 for fp operations and 1 otherwise.
13301 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13303 (atom_cost): Set reassociation width to 2.
13304 (slm_cost, generic_cost): Set fp reassociation width
13305 to 2 and 1 otherwise.
13306 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13307 (core_cost): Set fp reassociation width to 4 and vector to 2.
13308 (ix86_reassociation_width): Rewrite using cost table; special case
13309 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13310 and TARGET_AVX128_OPTIMAL.
13311 * config/i386/i386.h (processor_costs): Add
13312 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13313 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13314 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13315 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13316 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13317 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13319 2017-10-05 Nathan Sidwell <nathan@acm.org>
13321 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13323 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13325 * config/arm/arm.c (arm_test_fpu_data): New.
13326 (arm_run_selftests): Call arm_test_fpu_data.
13328 2017-10-04 Nathan Sidwell <nathan@acm.org>
13330 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13332 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13334 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13336 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13337 check type for aarch64_simd_valid_immediate.
13338 (aarch64_output_simd_mov_immediate): Update prototype.
13339 (aarch64_simd_valid_immediate): Update prototype.
13340 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13341 support for ORR-immediate.
13342 (and<mode>3): modified pattern to add support for BIC-immediate.
13343 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13344 now checks for valid immediate for BIC and ORR based on new enum
13346 (aarch64_output_simd_mov_immediate): Function now used to output
13347 BIC/ORR imm as well based on new enum argument.
13348 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13350 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13351 (aarch64_reg_or_bic_imm): Likewise.
13353 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13355 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13356 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13358 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13362 PR rtl-optimization/82396
13363 * haifa-sched.c (autopref_multipass_init): Simplify
13365 (autopref_rank_data): Simplify sort order.
13366 * sched-int.h (autopref_multipass_data_): Remove
13367 multi_mem_insn_p, min_offset and max_offset.
13369 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13371 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13373 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13375 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13376 vect_doubleint_cvt.
13378 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13380 * doc/sourcebuild.texi: Document vect_long_mult.
13382 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13384 PR tree-optimization/82413
13385 * fold-const.c (build_range_check): Use widest_int when comparing
13386 the maximum ETYPE value with HIGH.
13388 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13390 PR rtl-optimization/82396
13391 * haifa-sched.c (autopref_multipass_init): Simplify
13393 (autopref_rank_data): Simplify sort order.
13394 * sched-int.h (autopref_multipass_data_): Remove
13395 multi_mem_insn_p, min_offset and max_offset.
13397 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13399 PR tree-optimization/82381
13400 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13401 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13404 PR tree-optimization/82374
13405 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13406 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13407 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13408 current_function_decl to the new decl.
13410 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13412 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13413 helper macro for IEEE float128 hardware built-in functions.
13414 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13416 (TRUNCF128_ODD): Likewise.
13417 (ADDF128_ODD): Likewise.
13418 (SUBF128_ODD): Likewise.
13419 (MULF128_ODD): Likewise.
13420 (DIVF128_ODD): Likewise.
13421 (FMAF128_ODD): Likewise.
13422 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13423 UNSPEC_TRUNC_ROUND_TO_ODD.
13424 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13425 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13426 floating point round to odd instructions.
13427 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13428 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13429 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13430 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13431 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13432 (trunc<mode>sf2_hw): Change the truncate with round to odd
13433 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13434 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13435 to odd hardware instructions.
13436 (sub<mode>3_odd): Likewise.
13437 (mul<mode>3_odd): Likewise.
13438 (div<mode>3_odd): Likewise.
13439 (sqrt<mode>2_odd): Likewise.
13440 (fma<mode>4_odd): Likewise.
13441 (fms<mode>4_odd): Likewise.
13442 (nfma<mode>4_odd): Likewise.
13443 (nfms<mode>4_odd): Likewise.
13444 (trunc<mode>df2_odd): Change the truncate with round to odd
13445 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13447 * doc/extend.texi (PowerPC built-in functions): Update documentation
13448 for existing IEEE float128-bit built-in functions. Add built-in
13449 functions that generate the IEEE 128-bit floating point round to
13452 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13454 PR rtl-optimization/77729
13455 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13456 to (X&(C1&~C2))|C2 transformations.
13458 2017-10-03 Martin Jambor <mjambor@suse.cz>
13460 PR tree-optimization/82363
13461 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13462 mismatch, mark lacc written regardless of racc.
13464 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13466 PR tree-optimization/82381
13467 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13468 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13469 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13470 fallthrough into reassoc_stmt_dominates_stmt_p.
13473 * combine.c (combine_instructions): Don't combine in unreachable
13476 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13479 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13480 function to not use the have_cpu variable. Do not set cpu_index,
13481 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13482 or the default cpu.
13483 (rs6000_valid_attribute_p): Remove duplicate initializations of
13484 old_optimize and func_optimize.
13485 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13486 (rs6000_activate_target_options): Make global.
13487 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13490 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13492 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13493 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13494 Set *poffset to 0 and *psize and *pmax_size to -1 if
13495 *poffset + *psize overflows in HOST_WIDE_INT.
13497 PR tree-optimization/82387
13498 PR tree-optimization/82388
13499 PR tree-optimization/82389
13500 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13501 instead of live_bytes non-NULL.
13503 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13506 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13509 2017-10-02 Richard Biener <rguenther@suse.de>
13511 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13512 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13513 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13515 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13517 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13518 requested precision matches the type's.
13519 * calls.c (alloc_max_size): Calculate the new candidate size as
13520 a widest_int and use wi::to_widest when comparing it with the
13521 current candidate size.
13522 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13523 zero rather than integer_zero_node.
13524 * match.pd: Check for a no-op conversion before using wi::add
13525 rather than after. Use tree_to_uhwi when summing small shift
13526 counts into an unsigned int.
13528 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13529 Alan Hayward <alan.hayward@arm.com>
13530 David Sherwood <david.sherwood@arm.com>
13533 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13534 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13535 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13536 POINTER_AND_FP_REGS.
13538 2017-10-02 Richard Biener <rguenther@suse.de>
13540 PR tree-optimization/82355
13541 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13542 a mapping for the enclosing loop but avoid generating one for
13543 the loop tree root.
13544 (copy_bb_and_scalar_dependences): Remove premature codegen
13545 error on PHIs in blocks duplicated into multiple places.
13546 * graphite-scop-detection.c
13547 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13548 in the region use it as loop and nest to analyze the DR in.
13549 (try_generate_gimple_bb): Likewise.
13550 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13551 (add_loop_constraints): For blocks in a loop not in the region
13552 create a dimension with a single iteration.
13553 * sese.h (gbb_loop_at_index): Remove assert.
13555 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13557 * omp-expand.c (adjust_context_scope): New function.
13558 (expand_parallel_call): Call adjust_context_scope.
13560 2017-10-01 Jeff Law <law@redhat.com>
13562 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13563 dom_opt_dom_walker class with direct access to private members.
13564 Add comments. Call test_for_singularity.
13565 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13566 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13567 m_dummy_cond anymore.
13568 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13570 (pass_dominator:execute): Build the dummy_cond here and pass it
13571 to the dom_opt_dom_walker ctor.
13572 (test_for_singularity): New function.
13574 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13575 Maya Rashish <coypu@sdf.org>
13577 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13578 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13579 (alpha*-*-netbsd*) Use nbsd_tm_file.
13580 (arm*-*-netbsdelf*) Likewise.
13581 (i[34567]86-*-netbsdelf*) Likewise.
13582 (x86_64-*-netbsd*) Likewise.
13583 (mips*-*-netbsd*) Likewise.
13584 (powerpc-*-netbsd*) Likewise.
13585 (sh*-*-netbsd*) Likewise.
13586 (sparc-*-netbsdelf*) Likewise.
13587 (sparc64-*-netbsd*) Likewise.
13588 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13590 (vax-*-netbsdelf*) Likewise.
13591 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13592 (UINT_FAST8_TYPE) Likewise.
13593 (INT_FAST16_TYPE) Check CHAR_FAST16.
13594 (UINT_FAST16_TYPE) Likewise.
13596 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13599 * config/i386/i386.md
13600 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13601 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13602 *divmodsi4_zext_2): New define_insn_and_split.
13603 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13604 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13605 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13606 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13607 New define_insn_and_split.
13608 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13609 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13610 operands[1] having DImode when mode is SImode.
13612 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13613 always SImode for DIV and MOD in REG_EQUAL notes.
13615 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13617 PR middle-end/82319
13618 * match.pd: Fix handling of NaNs in pattern.
13620 2017-09-29 Jeff Law <law@redhat.com>
13622 * sbitmap.c (bitmap_bit_in_range_p): New function.
13623 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13624 * tree-ssa-dse.c (live_bytes_read): New function.
13625 (dse_classify_store): Ignore reads of dead bytes.
13627 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13628 typos and whitespace errors.
13629 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13630 * config/s390/s390.c (s390_emit_prologue): Likewise.
13632 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13635 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13636 with a symbol for LRA.
13638 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13640 PR rtl-optimization/82338
13641 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13643 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13645 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13646 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13647 (qsort_chk): Declare.
13648 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13649 (qsort_chk): New function.
13651 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13653 PR tree-optimization/82337
13654 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13655 phi definition if the PHI result appears in an abnormal PHI.
13656 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13657 basis appears in an abnormal PHI.
13659 2017-09-29 Richard Biener <rguenther@suse.de>
13661 * graphite-isl-ast-to-gimple.c
13662 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13663 (binary_op_to_tree): Use it.
13664 (get_rename_from_scev): Likewise.
13665 (copy_loop_phi_nodes): Likewise.
13666 (copy_bb_and_scalar_dependences): Likewise.
13667 (translate_pending_phi_nodes): Likewise.
13669 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13672 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13673 for movabsq $(i32 << shift), r64.
13675 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13677 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13678 index when encoding %esp as %rsp to avoid 0x67 prefix.
13680 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13682 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13683 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13685 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13687 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13688 Extensions with more than 16 double VFP registers.
13689 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13690 to_clear_mask and all code related to it. Replace the remaining
13691 entry by a sbitmap and adapt code accordingly.
13693 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13695 * brig-builtins.def: Change pure attributes to const.
13697 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13699 * config.gcc (default_gnu_indirect_function): Default to yes for
13700 sparc*-*-linux* with glibc.
13702 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13704 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13705 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13706 when creating .init_array and .fini_array sections with priority
13709 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13712 * config/aarch64/aarch64.c
13713 (aarch64_builtin_support_vector_misalignment): Always return false
13714 when misalignment is unknown.
13716 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13718 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
13719 this function to return false if the definition used by the swap
13720 instruction is artificial, or if the memory address from which the
13721 constant value is loaded is not represented by a base address held
13722 in a register or if the base address register is a frame or stack
13723 pointer. Additionally, return false if the base address of the
13724 loaded constant is a SYMBOL_REF but is not considered to be a
13726 (replace_swapped_load_constant): New function.
13727 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
13728 loaded constant vector with a load of a swapped constant vector.
13730 2017-09-27 Carl Love <cel@us.ibm.com>
13732 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
13733 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
13734 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
13737 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
13739 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
13740 first, always call autopref_rank_data otherwise.
13742 2017-09-27 Richard Biener <rguenther@suse.de>
13744 * graphite-scop-detection.c (find_scop_parameters): Move
13745 loop bound handling ...
13746 (gather_bbs::before_dom_children): ... here, avoiding the need
13747 to build scop_info->loop_nest.
13748 (record_loop_in_sese): Remove.
13749 * sese.h (sese_info_t::loop_nest): Remove.
13750 * sese.c (new_sese_info): Do not allocate loop_nest.
13751 (free_sese_info): Do not free loop_nest.
13753 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13756 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13757 lhs from calls if the lhs has addressable type.
13759 2017-09-27 Richard Biener <rguenther@suse.de>
13761 * graphite.h (scop::max_alias_set): New member.
13762 * graphite-scop-detection.c: Remove references to non-existing
13763 --param in comments.
13764 (build_alias_sets): Record the maximum alias set used for drs.
13765 (build_scops): Support zero as unlimited for
13766 --param graphite-max-arrays-per-scop.
13767 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
13768 and inline into ...
13769 (build_poly_sr_1): ... here. Compute alias set based on the
13770 maximum alias set used for drs rather than
13771 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
13773 2017-09-27 Richard Biener <rguenther@suse.de>
13775 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13776 --param loop-block-tile-size=0 to disable tiling.
13778 2017-09-27 Richard Biener <rguenther@suse.de>
13780 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
13781 (graphite-max-nb-scop-params): Document special value zero.
13782 * domwalk.h (dom_walker::STOP): New symbolical constant.
13783 (dom_walker::dom_walker): Add optional parameter for bb to
13785 (dom_walker::~dom_walker): Declare.
13786 (dom_walker::before_dom_children): Document STOP return value.
13787 (dom_walker::m_user_bb_to_rpo): New member.
13788 (dom_walker::m_bb_to_rpo): Likewise.
13789 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
13790 mapping here if not provided by the user.
13791 (dom_walker::~dom_walker): Free bb to RPO mapping if not
13792 provided by the user.
13793 (dom_walker::STOP): Define.
13794 (dom_walker::walk): Do not compute bb to RPO mapping here.
13795 Support STOP return value from before_dom_children to stop
13797 * graphite-optimize-isl.c (optimize_isl): If the schedule
13798 is the same still generate code if -fgraphite-identity
13799 or -floop-parallelize-all are given.
13800 * graphite-scop-detection.c: Include cfganal.h.
13801 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13803 (gather_bbs::before_dom_children): Return STOP for BBs
13805 (build_scops): Compute bb to RPO mapping and pass it to
13806 the domwalk. Treat --param graphite-max-nb-scop-params=0
13807 as not limiting the number of params.
13808 * graphite.c (graphite_initialize): Remove limit on the
13809 number of basic-blocks in a function.
13810 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13811 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13812 default value of 10.
13814 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13816 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13817 Adjust code to eliminate needing to do the shift right 32-bits
13818 operation after XSCVDPSPN.
13820 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13822 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13823 ((X / Y) != 0 -> X >= Y): Likewise.
13825 2017-09-26 Carl Love <cel@us.ibm.com>
13827 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13828 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13829 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13830 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13831 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13832 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13833 definitions and overloading.
13834 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13835 statement for P9V_BUILTIN_XST_LEN_R.
13836 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13837 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13838 define_expand and define_insn for the instructions and builtins.
13839 * doc/extend.texi: Update the built-in documentation file for the new
13840 built-in functions.
13841 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13842 define_insn for the instructions
13844 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13847 * gcc/config/netbsd-protos.h: New file.
13848 * gcc/config/netbsd.c: New file.
13849 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13850 * gcc/config/t-netbsd: New file.
13851 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13852 (tmake_file) Add t-netbsd.
13853 (extra_objs) Add netbsd.o.
13855 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13859 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13861 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13863 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13864 sign extension from a vector register to a GPR by doing a 32-bit
13865 direct move and then an EXTSW.
13866 (extendsi<mode>2 splitter): Likewise.
13867 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13868 right or vector extract after doing XSCVDPSPN. Use
13869 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13871 (movdi_from_sf_zero_ext): Likewise.
13872 (reload_gpr_from_vsxsf): Likewise.
13873 (p8_mfvsrd_4_disf): Delete, no longer used.
13874 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13875 then needing to move the SFmode to a GPR to use the XSCVDPSP
13876 instruction instead of FRSP and XSCVDPSPN.
13877 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13878 it is adjacent to the other XSCVSPDP insns.
13879 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13880 SFmode to be in traditional Altivec registers.
13881 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13882 (vsx_xscvspdpn): Likewise.
13883 (vsx_xscvspdpn_scalar): Likewise.
13885 2017-09-26 Martin Jambor <mjambor@suse.cz>
13887 * tree-sra.c (compare_access_positions): Put integral types first,
13888 stabilize sorting of integral types, remove conditions putting
13889 non-full-precision integers last.
13890 (sort_and_splice_var_accesses): Disable scalarization if a
13891 non-integert would be represented by a non-full-precision integer.
13893 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13895 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13896 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13897 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13898 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13899 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13900 conditionals inside the function instead of around it. Call
13901 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13902 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13904 2017-09-26 Richard Biener <rguenther@suse.de>
13906 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13908 (scop_detection::build_scop_breadth): ... this. Removed.
13909 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13910 (scop_detection::harmful_stmt_in_bb): Likewise.
13911 (scop_detection::graphite_can_represent_stmt): Likewise.
13912 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13913 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13914 (scop_detection::can_represent_loop_1): ... this. Removed.
13915 (scop_detection::harmful_loop_in_region): Simplify after inlining
13916 the above and remove more quadraticness.
13917 (build_scops): Adjust.
13918 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13921 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13924 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13925 REGNO (base) == SP_REG if base is a REG.
13927 PR middle-end/35691
13928 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13929 if it is different SSA_NAME.
13930 (optimize_range_tests_cmp_bitwise): New function.
13931 (optimize_range_tests): Call it.
13933 2017-09-26 Richard Biener <rguenther@suse.de>
13935 PR tree-optimization/82321
13936 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13937 for the def being inside the loop.
13939 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13941 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13943 * config/s390/s390-builtins.def: Fix constraint on op4.
13945 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13947 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13948 independent expanders.
13949 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13950 ("vec_ordered", "vec_unordered"): New expanders.
13952 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13954 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13957 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13959 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13960 vec_unpacks_lo_v16qi.
13961 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13963 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13965 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13966 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13967 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13969 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13971 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13973 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13975 ("*vec_slb<mode>"): New insn pattern.
13976 ("vec_shr_<mode>"): New expander.
13977 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13978 and force the shift count operand to V16QImode.
13979 ("vec_srb<mode>"): Set shift count mode to V16QI.
13981 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13983 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13984 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13985 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13987 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13990 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13992 2017-09-26 Richard Biener <rguenther@suse.de>
13994 PR tree-optimization/82320
13995 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13998 2017-09-25 Jeff Law <law@redhat.com>
14000 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14001 prototype for new argument.
14002 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14003 mostly extracted from rs6000_emit_allocate_stack.
14004 (rs6000_emit_probe_stack_range_stack_clash): New function.
14005 (rs6000_emit_allocate_stack): Call
14006 rs6000_emit_probe_stack_range_stack_clash as needed.
14007 (rs6000_emit_probe_stack_range): Add additional argument
14008 to call to gen_probe_stack_range{si,di}.
14009 (output_probe_stack_range): New.
14010 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14011 (output_probe_stack_range_stack_clash): New.
14012 (rs6000_emit_prologue): Emit notes into dump file as requested.
14013 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14014 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14015 Add additional operand and pass it to output_probe_stack_range.
14017 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14019 PR tree-optimization/82163
14020 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14021 (checking_verify_loop_closed_ssa): New parameter.
14022 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14023 (check_loop_closed_ssa_stmt): Delete.
14024 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14025 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14026 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14029 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14031 * brig-builtins.def: Treat HSAIL barrier builtins as
14032 setjmp/longjump style functions.
14034 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14036 * target.def (constant_alignment): New hook.
14037 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14038 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14039 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14040 * doc/tm.texi: Regenerate.
14041 * targhooks.h (default_constant_alignment): Declare.
14042 (constant_alignment_word_strings): Likewise.
14043 * targhooks.c (default_constant_alignment): New function.
14044 (constant_alignment_word_strings): Likewise.
14045 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14046 instead of CONSTANT_ALIGNMENT.
14047 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14048 (force_const_mem): Likewise.
14049 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14050 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14051 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14052 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14053 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14055 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14056 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14057 constant_alignment_word_strings.
14058 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14059 (CONSTANT_ALIGNMENT): Likewise.
14060 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14061 (arm_constant_alignment): New function.
14062 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14063 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14064 constant_alignment_word_strings.
14065 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14066 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14067 constant_alignment_word_strings.
14068 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14069 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14070 (cris_constant_alignment): New function.
14071 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14072 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14073 (epiphany_constant_alignment): New function.
14074 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14075 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14076 constant_alignment_word_strings.
14077 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14078 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14079 constant_alignment_word_strings.
14080 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14081 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14082 constant_alignment_word_strings.
14083 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14084 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14085 * config/i386/i386.c (ix86_constant_alignment): Make static.
14086 Use the same interface as the target hook.
14087 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14088 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14089 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14090 constant_alignment_word_strings.
14091 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14092 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14093 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14094 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14095 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14096 constant_alignment_word_strings.
14097 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14098 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14099 constant_alignment_word_strings.
14100 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14101 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14102 constant_alignment_word_strings.
14103 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14104 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14106 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14107 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14108 * config/mips/mips.c (mips_constant_alignment): New function.
14109 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14110 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14111 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14112 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14113 (mmix_constant_alignment): Make static. Use the same interface
14114 as the target hook.
14115 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14116 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14117 constant_alignment_word_strings.
14118 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14119 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14120 constant_alignment_word_strings.
14121 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14122 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14123 constant_alignment_word_strings.
14124 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14125 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14126 (rs6000_constant_alignment): New function.
14127 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14128 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14129 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14130 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14131 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14132 (rs6000_constant_alignment): New function.
14133 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14134 * config/s390/s390.c (s390_constant_alignment): New function.
14135 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14136 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14137 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14138 constant_alignment_word_strings.
14139 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14140 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14141 (sparc_constant_alignment): New function.
14142 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14143 * config/spu/spu.c (spu_constant_alignment): New function.
14144 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14145 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14146 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14147 constant_alignment_word_strings.
14148 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14149 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14150 constant_alignment_word_strings.
14151 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14152 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14153 constant_alignment_word_strings.
14154 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14155 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14156 (visium_constant_alignment): New function.
14157 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14158 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14159 (xtensa_constant_alignment): New function.
14160 * system.h (CONSTANT_ALIGNMENT): Poison.
14162 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14164 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14165 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14166 (rs6000_builtin_valid_without_lhs): New helper function.
14167 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14168 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14170 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14172 * target.h (vec_perm_indices): Use unsigned short rather than
14174 (auto_vec_perm_indices): Likewise.
14175 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14176 Use unsigned int rather than unsigned char.
14177 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14179 2017-09-25 Richard Biener <rguenther@suse.de>
14181 * cfgloop.h (sort_sibling_loops): Declare.
14182 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14183 (sort_sibling_loops): New function sorting the sibling loop list
14185 * graphite.c (graphite_transform_loops): Sort sibling loops.
14187 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14189 * target.def (vec_perm_const_ok): Change sel parameter to
14191 * optabs-query.c (can_vec_perm_p): Update accordingly.
14192 * doc/tm.texi: Regenerate.
14193 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14194 auto_vec_perm_indices and remove separate nelt field.
14195 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14196 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14197 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14198 (aarch64_expand_vec_perm_const): Update accordingly.
14199 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14200 to vec_perm_indices.
14201 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14202 auto_vec_perm_indices and remove separate nelt field.
14203 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14204 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14205 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14207 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14208 to vec_perm_indices.
14209 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14210 sel to vec_perm_indices.
14211 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14212 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14213 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14215 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14218 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14221 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14222 on the FUNCTION_DECL function context if it has a DIE that is a
14225 2017-09-25 Richard Biener <rguenther@suse.de>
14227 PR tree-optimization/82285
14228 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14231 2017-09-25 Tom de Vries <tom@codesourcery.com>
14235 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14238 2017-09-25 Richard Biener <rguenther@suse.de>
14240 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14241 ISL errors other than isl_error_quota happen. Dump if the
14242 schedule is the same.
14243 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14244 errors instead of aborting inside ISL.
14246 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14249 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14250 of libgcc_eh for m64.
14251 * config/i386/darwin64.h: Likewise.
14253 2017-09-25 Richard Biener <rguenther@suse.de>
14255 PR middle-end/82144
14256 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14257 attribute for incomplete types nor twice for complete ones.
14259 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14262 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14263 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14266 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14269 * config/i386/i386.opt (mprefer-avx256): Use
14270 ix86_target_flags variable.
14271 * config/i386/i386.c (ix86_target_string): Move
14272 -mprefer-avx256 to flag2_opts.
14274 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14276 PR middle-end/35691
14277 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14278 and x != -1 | y != -1 into (x & y) != -1.
14280 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14282 * config.gcc: Add new case statement to set
14283 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14284 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14285 s390x-*-linux* case statements. Added aarch64 to the list of
14286 supported architectures.
14288 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14290 PR tree-optimization/82289
14291 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14292 STMT_VINFO_RELEVANT_P.
14294 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14295 Alan Hayward <alan.hayward@arm.com>
14296 David Sherwood <david.sherwood@arm.com>
14298 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14299 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14300 (extract_range_from_binary_expr_1): Don't call
14301 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14303 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14304 Alan Hayward <alan.hayward@arm.com>
14305 David Sherwood <david.sherwood@arm.com>
14307 * target.def (preferred_vector_alignment): New hook.
14308 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14310 * doc/tm.texi: Regenerate.
14311 * targhooks.h (default_preferred_vector_alignment): Declare.
14312 * targhooks.c (default_preferred_vector_alignment): New function.
14313 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14315 (DR_TARGET_ALIGNMENT): New macro.
14316 (aligned_access_p): Update commentary.
14317 (vect_known_alignment_in_bytes): New function.
14318 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14320 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14321 Calculate the misalignment based on the target alignment rather than
14323 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14324 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14325 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14326 the target alignment, rather than masking the element misalignment
14327 with the number of elements in a vector. Also use the target
14328 alignment when calculating the maximum number of peels.
14329 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14330 instead of TYPE_ALIGN_UNIT.
14331 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14332 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14333 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14334 (vect_create_data_ref_ptr): Likewise.
14335 (vect_setup_realignment): Realign by ANDing with
14336 -DR_TARGET_MISALIGNMENT.
14337 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14338 the number of peels based on DR_TARGET_ALIGNMENT.
14339 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14340 with the guaranteed alignment boundary when deciding whether
14342 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14343 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14344 (ensure_base_align): Remove stmt_info parameter. Get the
14345 target base alignment from DR_TARGET_ALIGNMENT.
14346 (vectorizable_store): Update call accordingly. Interpret
14347 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14349 (vectorizable_load): Likewise.
14351 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14352 Alan Hayward <alan.hayward@arm.com>
14353 David Sherwood <david.sherwood@arm.com>
14355 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14356 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14357 (vect_enhance_data_refs_alignment): Likewise.
14359 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14361 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14362 error. Only quit immediately if parsing is complete.
14363 (BEGIN): Initialize fatal_err and parse_done.
14364 (begin fpu, end fpu): Check number of arguments.
14365 (begin arch, end arch): Likewise.
14366 (begin cpu, end cpu): Likewise.
14367 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14368 (optalias): Likewise.
14370 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14372 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14373 * config/arm/arm-isa.h: Delete. Move definitions to ...
14374 * arm-cpus.in: ... here. Use new feature and fgroup values.
14375 * config/arm/arm.c (arm_option_override): Use lower case for feature
14377 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14378 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14379 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14381 (print_isa_bits_for): New function.
14382 (gen_isa): New function.
14383 (gen_comm_data): Use print_isa_bits_for.
14384 (define feature): New keyword.
14385 (define fgroup): New keyword.
14386 * config/arm/t-arm (TM_H): Remove.
14387 (GTM_H): Add arm-isa.h.
14388 (arm-isa.h): Add rule to generate file.
14389 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14390 case for feature bit names.
14392 2017-09-22 Richard Biener <rguenther@suse.de>
14394 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14396 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14397 print of no dependency loops ...
14398 * graphite.c (graphite_transform_loops): ... here.
14399 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14401 (same_close_phi_node, remove_duplicate_close_phi,
14402 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14403 (canonicalize_loop_closed_ssa): ... here and simplify.
14404 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14405 (optimize_isl): Use dump_printf_loc to tell when we stopped
14406 optimizing because of an ISL timeout.
14408 2017-09-22 Richard Biener <rguenther@suse.de>
14410 PR tree-optimization/82291
14411 * tree-if-conv.c (predicate_mem_writes): Make sure to
14412 remove writes in blocks predicated with false.
14414 2017-09-22 Richard Biener <rguenther@suse.de>
14416 * sese.c: Include cfganal.h.
14417 (if_region_set_false_region): Remove.
14418 (create_if_region_on_edge): Likewise.
14419 (move_sese_in_condition): Re-implement without destroying
14422 2017-09-22 Richard Biener <rguenther@suse.de>
14424 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14425 Verify both BBs contain loop PHI nodes before dispatching to
14426 copy_loop_phi_args.
14427 (graphite_regenerate_ast_isl): Do not recompute dominators,
14428 do not verify three times. Restructure for clarity.
14429 * graphite-scop-detection.c (same_close_phi_node,
14430 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14431 defined_in_loop_p, canonicalize_loop_closed_ssa,
14432 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14433 checking and SSA rewrite, move to ...
14434 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14435 (graphite_initialize): Do not pass in ctx, do not reset the
14436 SCEV cache, compute only dominators.
14437 (graphite_transform_loops): Allocate ISL ctx after
14438 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14439 Maintain post-dominators only around build_scops.
14440 * sese.c (if_region_set_false_region): Make static. Free
14441 and recompute dominators.
14442 (move_sese_in_condition): Assert we don't get called with
14443 post-dominators computed.
14444 * sese.h (if_region_set_false_region): Remove.
14446 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14448 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14449 mode attribute for TARGET_AVX512VL.
14451 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14453 * config/i386/i386.opt (mprefer-avx256): New option.
14454 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14456 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14457 for TARGET_PREFER_AVX256.
14458 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14460 2017-09-21 Jeff Law <law@redhat.com>
14462 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14463 Fix dump output if the only stack space is for pushed registers.
14465 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14467 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14470 2017-09-21 Martin Sebor <msebor@redhat.com>
14473 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14474 code (in C++) or code that triggers warnings.
14476 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14478 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14480 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14482 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14483 * sched-rgn.c: Ditto.
14484 * sel-sched-ir.c: Ditto.
14486 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14488 * toplev.h (set_random_seed): Adjust return type.
14489 * toplev.c (init_local_tick): Move eager initialization of random_seed
14490 to get_random_seed. Adjust comment.
14491 (init_random_seed): Inline to get_random_seed, delete.
14492 (get_random_seed): Initialize random_seed lazily.
14493 (set_random_seed): Do not return previous value.
14494 (print_switch_value): Do not call get_random_seed.
14496 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14498 * cgraph.c (delete_function_version): New, broken out from...
14499 (cgraph_node::delete_function_version): ...here. Rename to
14500 cgraph_node::delete_function_version_by_decl. Update all uses.
14501 (cgraph_node::remove): Call delete_function_version.
14503 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14506 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14507 VAR_DECLs to which addressable non-volatile parameters are mapped
14508 and for id->retvar after the return value assignment. Clear
14509 id->retval and id->retbnd after inlining.
14511 2017-09-21 Richard Biener <rguenther@suse.de>
14513 PR tree-optimization/82276
14514 PR tree-optimization/82244
14515 * tree-vrp.c (build_assert_expr_for): Set
14516 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14518 (remove_range_assertions): Revert earlier change.
14520 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14523 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14525 2017-09-21 Richard Biener <rguenther@suse.de>
14527 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14528 Restore valid IL after code generation errors.
14529 * graphite.c (graphite_transform_loops): Diagnose code
14530 generation issues as MSG_MISSED_OPTIMIZATION and continue
14531 with processing SCOPs.
14533 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14534 Alan Hayward <alan.hayward@arm.com>
14535 David Sherwood <david.sherwood@arm.com>
14537 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14538 rather than choosing between plus_constant and gen_rtx_<CODE>.
14539 * expr.c (emit_push_insn): Likewise.
14540 (expand_expr_real_2): Likewise.
14542 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14543 Alan Hayward <alan.hayward@arm.com>
14544 David Sherwood <david.sherwood@arm.com>
14546 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14548 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14549 Alan Hayward <alan.hayward@arm.com>
14550 David Sherwood <david.sherwood@arm.com>
14552 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14553 calling tree_to_uhwi.
14555 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14556 Alan Hayward <alan.hayward@arm.com>
14557 David Sherwood <david.sherwood@arm.com>
14559 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14560 INTEGER_CST rather than a negative test for ADDR_EXPR.
14562 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14563 Alan Hayward <alan.hayward@arm.com>
14564 David Sherwood <david.sherwood@arm.com>
14566 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14567 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14569 2017-09-21 Richard Biener <rguenther@suse.de>
14571 PR tree-optimization/71351
14572 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14573 graphite_create_new_loop_guard): Remove, fold remaining parts
14575 (translate_isl_ast_node_for): ... here and simplify.
14577 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14580 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14581 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14582 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14583 alternative always use QI mode, for -Os imov (=R,R) alternative
14584 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14587 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14588 Jeff Law <law@redhat.com>
14590 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14591 (allocate_stack_space): New function, partially extracted from
14592 s390_emit_prologue.
14593 (s390_emit_prologue): Track offset to most recent stack probe.
14594 Code to allocate space moved into allocate_stack_space.
14595 Dump actions when no stack is allocated.
14596 (s390_prologue_plus_offset): New function.
14597 (s390_emit_stack_probe): Likewise.
14599 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14601 * common.opt (Wa, Wl, Wp, g, gz=): Add
14603 (gno-column-info): Remove.
14604 (gcolumn-info): Drop RejectNegative.
14605 (gno-): New prefix.
14606 (gno-record-gcc-switches): Remove.
14607 (grecord-gcc-switches): Drop RejectNegative.
14608 (gno-split-dwarf): Remove.
14609 (gsplit-dwarf): Drop RejectNegative.
14610 (gno-strict-dwarf): Remove.
14611 (gstrict-dwarf): Drop RejectNegative.
14612 * config/darwin.opt (gfull, gused): Add RejectNegative.
14613 * dwarf2out.c (gen_producer_string): Drop
14614 gno-record-gcc-switches handler.
14615 * optc-gen.awk: Add g to prefixes with negative forms.
14616 * opts-common.c (remapping_prefix_p): New.
14617 (find_opt): Check it.
14618 (generate_canonical_option): Test g prefix.
14619 (option_map): Add -gno- mapping.
14620 (add_misspelling_candidates): Check remapping_prefix_p.
14622 2017-09-20 Jeff Law <law@redhat.com>
14624 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14625 thinko in stack clash protection support.
14627 * explow.c (compute_stack_clash_protection_loop_data): Use
14628 CONST_INT_P instead of explicit test. Verify object is a
14629 CONST_INT_P before looking at INTVAL.
14630 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14631 instead of explicit test.
14633 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14636 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14637 address instead of to r1 and r11.
14639 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14641 * config.gcc: Support "knm".
14642 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14643 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14645 * config/i386/i386.c (m_KNM): Define.
14646 (processor_target_table): Add "knm".
14648 (ix86_option_override_internal): Add "knm".
14649 (ix86_issue_rate): Add PROCESSOR_KNM.
14650 (ix86_adjust_cost): Ditto.
14651 (ia32_multipass_dfa_lookahead): Ditto.
14652 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14653 (fold_builtin_cpu): Add M_INTEL_KNM.
14654 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14655 (processor_type): Add PROCESSOR_KNM.
14656 * config/i386/x86-tune.def: Add m_KNM.
14657 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14659 2017-09-20 Richard Biener <rguenther@suse.de>
14661 PR tree-optimization/80213
14662 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14663 are allowed in empty BBs as well.
14664 (canonicalize_loop_closed_ssa): Also look for other complex
14666 (scop_detection::get_sese): Include the loop-closed PHI block
14668 (scop_detection::merge_sese): Remove code adding extra blocks.
14669 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14670 (build_scops): Assert the final returned scop is invalid.
14672 2017-09-20 Richard Biener <rguenther@suse.de>
14674 PR tree-optimization/82264
14675 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14677 (vn_phi_lookup): Likewise.
14678 (vn_phi_insert): Likewise.
14680 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14682 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14683 that fits into uhwi or shwi, add DW_AT_const_value regardless
14684 of early_dwarf without going through RTL, using add_AT_unsigned
14687 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14688 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14689 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14690 (DEBUG_MACRO_SECTION): Likewise.
14691 (DEBUG_LTO_MACRO_SECTION): Likewise.
14692 (DEBUG_STR_DWO_SECTION): Likewise.
14693 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14694 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14695 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14696 (DEBUG_LTO_LINE_STR_SECTION): Define.
14697 (init_sections_and_labels): Initialize debug_line_str_section
14698 variable. Initialize debug_loc_section for -gdwarf-5 to
14699 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14701 2017-09-20 Richard Biener <rguenther@suse.de>
14703 * graphite-sese-to-poly.c (extract_affine): Properly handle
14704 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14706 2017-09-20 Richard Biener <rguenther@suse.de>
14708 PR tree-optimization/81373
14709 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14710 Force SESE live-out defs to be handled even if they are
14713 2017-09-19 Jeff Law <law@redhat.com>
14715 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
14716 nothing for stack adjustments with REG_STACK_CHECK.
14717 * sched-deps.c (parse_add_or_inc): Reject insns with
14718 REG_STACK_CHECK from dependency breaking.
14719 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
14720 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
14721 * reg-notes.def (STACK_CHECK): New note.
14723 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
14724 (ix86_expand_prologue): Dump stack clash info as needed.
14725 Call ix86_adjust_stack_and_probe_stack_clash as needed.
14727 * function.c (dump_stack_clash_frame_info): New function.
14728 * function.h (dump_stack_clash_frame_info): Prototype.
14729 (enum stack_clash_probes): New enum.
14731 * config/alpha/alpha.c (alpha_expand_prologue): Also check
14732 flag_stack_clash_protection.
14733 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14734 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
14735 (arm_frame_pointer_required): Likewise.
14736 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
14737 (ia64_expand_prologue): Likewise.
14738 * config/mips/mips.c (mips_expand_prologue): Likewise.
14739 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
14740 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14741 (sparc_flat_expand_prologue): Likewise.
14742 * config/spu/spu.c (spu_expand_prologue): Likewise.
14744 * explow.c: Include "params.h".
14745 (anti_adjust_stack_and_probe_stack_clash): New function.
14746 (get_stack_check_protect): Likewise.
14747 (compute_stack_clash_protection_loop_data): Likewise.
14748 (emit_stack_clash_protection_loop_start): Likewise.
14749 (emit_stack_clash_protection_loop_end): Likewise.
14750 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14751 Use anti_adjust_stack_and_probe_stack_clash.
14752 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14753 (emit_stack_clash_protection_loop_start): Likewise.
14754 (emit_stack_clash_protection_loop_end): Likewise.
14755 * rtl.h (get_stack_check_protect): Prototype.
14756 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14757 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14758 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14760 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14762 * doc/tm.texi: Rebuilt.
14763 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14764 get_stack_check_protect.
14765 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
14766 * config/arm/arm.c (arm_expand_prologue): Likewise.
14767 (arm_frame_pointer_required): Likewise.
14768 * config/i386/i386.c (ix86_expand_prologue): Likewise.
14769 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
14770 * config/mips/mips.c (mips_expand_prologue): Likewise.
14771 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
14772 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
14773 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14774 (sparc_flat_expand_prologue): Likewise.
14776 * common.opt (-fstack-clash-protection): New option.
14777 * flag-types.h (enum stack_check_type): Note difference between
14778 -fstack-check= and -fstack-clash-protection.
14779 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
14780 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
14781 * toplev.c (process_options): Issue warnings/errors for cases
14782 not handled with -fstack-clash-protection.
14783 * doc/invoke.texi (-fstack-clash-protection): Document new option.
14784 (-fstack-check): Note additional problem with -fstack-check=generic.
14785 Note that -fstack-check is primarily for Ada and refer users
14786 to -fstack-clash-protection for stack-clash-protection.
14787 Document new params for stack clash protection.
14789 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
14791 * config/i386/i386.c (ix86_split_long_move): Do not handle
14792 address used for LEA in a special way.
14794 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
14796 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14798 2017-09-19 Martin Sebor <msebor@redhat.com>
14801 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14802 of incompatible types.
14804 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14806 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14807 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14808 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14809 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14811 2017-09-19 Richard Biener <rguenther@suse.de>
14813 PR tree-optimization/82244
14814 * tree-vrp.c (remove_range_assertions): Do not propagate
14815 a constant to abnormals but replace the assert with a copy.
14817 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14819 PR rtl-optimization/57878
14820 PR rtl-optimization/68988
14821 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14822 avoidance test involving non_reload_pseudos. Move frequency test
14823 below the general fragmentation avoidance test.
14825 2017-09-19 Richard Biener <rguenther@suse.de>
14827 PR tree-optimization/69728
14828 * graphite-sese-to-poly.c (schedule_error): New global.
14829 (add_loop_schedule): Handle empty domain by failing the
14831 (build_original_schedule): Handle schedule_error.
14833 2017-09-19 Richard Biener <rguenther@suse.de>
14835 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14836 Do not iterate to sibling loops but only to siblings of inner
14839 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14842 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14843 that the registers are different.
14845 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14847 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14848 to processor_model and "amdfam17h" to arch_names_table.
14849 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14851 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14854 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14856 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14857 Alan Hayward <alan.hayward@arm.com>
14858 David Sherwood <david.sherwood@arm.com>
14860 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14862 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14864 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14865 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14866 vect_schedule_slp_instance.
14867 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14868 Update call to vect_slp_analyze_node_operations. Simplify return
14870 (vect_slp_analyze_bb_1): Update call accordingly.
14871 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14872 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14873 (vect_schedule_slp): Update call accordingly.
14875 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14876 Alan Hayward <alan.hayward@arm.com>
14877 David Sherwood <david.sherwood@arm.com>
14879 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14880 with types that aren't in fact scalar.
14882 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14884 * tree-vect-slp.c (vect_record_max_nunits): New function,
14886 (vect_build_slp_tree_1): ...here.
14887 (vect_build_slp_tree_2): Call it for phis too.
14889 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14891 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14892 to vect_get_vec_def_for_operand when getting the mask operand.
14894 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14895 Alan Hayward <alan.hayward@arm.com>
14896 David Sherwood <david.sherwood@arm.com>
14898 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14901 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14902 Alan Hayward <alan.hayward@arm.com>
14903 David Sherwood <david.sherwood@arm.com>
14905 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14906 calculation for vector booleans.
14908 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14909 Alan Hayward <alan.hayward@arm.com>
14910 David Sherwood <david.sherwood@arm.com>
14912 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14914 (vect_transform_stmt): ...here.
14915 (vect_analyze_stmt): Use it instead of calling
14916 vectorizable_live_operation directly.
14918 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14920 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14921 non-SIMT targets in acc vector loops.
14923 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14925 * configure.ac: Add arc and check if assembler supports gdwarf2.
14926 * configure: Regenerate.
14928 2017-09-18 Richard Biener <rguenther@suse.de>
14930 PR tree-optimization/82220
14931 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14932 epilogue niters from the min_profitable_iters compute.
14934 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14937 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14938 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14939 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14941 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14944 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14945 switching to a new text section.
14947 2017-09-18 Richard Biener <rguenther@suse.de>
14949 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14951 (build_alias_set): Reject aliases with no access function.
14953 2017-09-18 Richard Biener <rguenther@suse.de>
14955 PR tree-optimization/79622
14956 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14958 (build_cross_bb_scalars_use): Likewise.
14960 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14962 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14964 2017-09-18 Alan Modra <amodra@gmail.com>
14967 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14968 stack_pointer_rtx for count 0. Update comments. Break up
14969 large rtl expression.
14971 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14973 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14974 Increase to 20 bytes.
14975 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14976 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14977 or avx version of the stub.
14979 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14982 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14983 compute the minimum stack alignment. Also update preferred stack
14984 boundary for leaf functions.
14986 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14988 PR tree-optimization/82228
14989 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14992 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14994 * common/config/nds32/nds32-common.c
14995 (nds32_option_optimization_table): Refine formatting.
14996 (nds32_option_optimization_table): Use -fsched-pressure and
14997 -fomit-frame-pointer for specific optimization level.
14999 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15001 * config/nds32/nds32.c: Refine formatting and comments.
15002 * config/nds32/nds32.h: Likewise.
15003 * config/nds32/nds32.md: Likewise.
15004 * config/nds32/nds32-cost.c: Likewise.
15005 * config/nds32/nds32-isr.c: Likewise.
15006 * config/nds32/nds32-md-auxiliary.c: Likewise.
15007 * config/nds32/nds32-multiple.md: Likewise.
15008 * config/nds32/nds32-predicates.c: Likewise.
15010 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15011 Jakub Jelinek <jakub@redhat.com>
15013 Add support for -std=c++2a.
15014 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15016 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15018 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15021 * doc/extend.texi (Common Function Attributes): Add
15022 references to ARM, AArch64, and S/390 specific attributes.
15023 (Function Specific Option Pragmas): Add AArch64 and S/390
15024 to list of back ends that support the target pragma.
15026 2017-09-15 Nathan Sidwell <nathan@acm.org>
15028 * doc/standards.texi: Fix C++17 description. Update URLs for
15031 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15033 * common.opt (Wcast-align=strict): New warning option.
15034 * doc/invoke.texi: Document -Wcast-align=strict.
15036 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15038 * cgraph.h (cgraph_thunk_info): Add comments.
15039 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15040 assert for VIRTUAL_* arguments stricter.
15042 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15044 PR tree-optimization/71026
15045 * match.pd: Move RDIV patterns from fold-const.c
15046 * fold-const.c (distribute_real_division): Removed.
15047 (fold_binary_loc): Remove calls to distribute_real_divison.
15049 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15051 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15052 c++1z and gnu++1z as deprecated. Change other references to
15053 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15054 Change -Wc++1z-compat to -Wc++17-compat.
15055 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15056 * dwarf2out.c (highest_c_language): Handle C++17.
15057 (gen_compile_unit_die): Likewise.
15059 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15061 PR rtl-optimization/82192
15062 * combine.c (make_extraction): Don't look through non-paradoxical
15063 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15066 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15067 Alan Hayward <alan.hayward@arm.com>
15068 David Sherwood <david.sherwood@arm.com>
15070 * target.def (function_arg_offset): New hook.
15071 * targhooks.h (default_function_arg_offset): Declare.
15072 * targhooks.c (default_function_arg_offset): New function.
15073 * function.c (locate_and_pad_parm): Use
15074 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15075 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15076 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15077 * doc/tm.texi: Regenerate.
15078 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15079 * config/spu/spu.c (spu_function_arg_offset): New function.
15080 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15081 * system.h (FUNCTION_ARG_OFFSET): Poison.
15083 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15084 Alan Hayard <alan.hayward@arm.com>
15085 David Sherwood <david.sherwood@arm.com>
15087 * target.def (truly_noop_truncation): New hook.
15088 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15089 than TRULY_NOOP_TRUNCATION.
15090 * hooks.h (hook_bool_uint_uint_true): Declare.
15091 * hooks.c (hook_bool_uint_uint_true): New function.
15092 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15093 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15094 * doc/tm.texi: Regenerate.
15095 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15096 rather than TRULY_NOOP_TRUNCATION in comments.
15097 (simplify_comparison): Likewise.
15098 (record_truncated_value): Likewise.
15099 * expmed.c (extract_bit_field_1): Likewise.
15100 (extract_split_bit_field): Likewise.
15101 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15102 instead of TRULY_NOOP_TRUNCATION.
15103 * function.c (assign_parm_setup_block): Likewise.
15104 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15105 * rtlhooks.c: Include target.h.
15106 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15107 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15108 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15109 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15110 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15111 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15112 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15113 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15114 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15115 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15116 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15117 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15118 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15119 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15120 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15121 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15122 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15123 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15124 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15125 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15126 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15127 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15128 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15129 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15130 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15131 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15132 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15133 rather than TRULY_NOOP_TRUNCATION in comments.
15134 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15135 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15136 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15137 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15138 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15139 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15140 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15141 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15142 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15143 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15144 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15145 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15146 rather than TRULY_NOOP_TRUNCATION in comments.
15147 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15148 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15149 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15150 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15151 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15152 TRULY_NOOP_TRUNCATION condition.
15153 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15154 (TRULY_NOOP_TRUNCATION): Delete.
15155 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15156 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15157 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15158 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15159 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15160 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15161 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15162 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15163 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15164 rather than TRULY_NOOP_TRUNCATION in comments.
15165 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15166 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15167 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15168 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15169 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15170 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15172 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15175 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15176 (*cmp_ior): Likewise.
15177 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15178 (*ior_scc_scc_cmp): Likewise.
15179 (*and_scc_scc): Likewise.
15180 (*and_scc_scc_cmp): Likewise.
15182 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15183 Alan Hayard <alan.hayward@arm.com>
15184 David Sherwood <david.sherwood@arm.com>
15186 * target.def (can_change_mode_class): New hook.
15187 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15188 (hard_regno_nregs): Likewise.
15189 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15190 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15191 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15192 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15193 (LOAD_EXTEND_OP): Update accordingly.
15194 * doc/tm.texi: Regenerate.
15195 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15196 CANNOT_CHANGE_MODE_CLASS.
15197 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15198 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15199 * combine.c (simplify_set): Update accordingly.
15200 * emit-rtl.c (validate_subreg): Likewise.
15201 * recog.c (general_operand): Likewise.
15202 * regcprop.c (mode_change_ok): Likewise.
15203 * reload1.c (choose_reload_regs): Likewise.
15204 (inherit_piecemeal_p): Likewise.
15205 * rtlanal.c (simplify_subreg_regno): Likewise.
15206 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15207 instead of CANNOT_CHANGE_MODE_CLASS.
15208 (reload_cse_simplify_operands): Likewise.
15209 * reload.c (push_reload): Use targetm.can_change_mode_class
15210 instead of CANNOT_CHANGE_MODE_CLASS.
15211 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15212 REG_CANNOT_CHANGE_MODE_P.
15213 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15214 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15215 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15216 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15217 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15218 (arm_can_change_mode_class): New function.
15219 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15220 than CANNOT_CHANGE_MODE_CLASS in comments.
15221 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15222 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15223 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15224 (ix86_can_change_mode_class): ...this new function, inverting the
15225 sense of the return value.
15226 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15227 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15228 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15229 (ia64_can_change_mode_class): New function.
15230 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15231 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15232 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15233 (m32c_can_change_mode_class): ...this new function, inverting the
15234 sense of the return value.
15235 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15236 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15237 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15238 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15239 (mips_can_change_mode_class): ...this new function, inverting the
15240 sense of the return value.
15241 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15242 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15243 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15244 (msp430_can_change_mode_class): New function.
15245 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15246 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15247 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15248 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15249 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15250 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15251 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15252 (pa_cannot_change_mode_class): Replace with...
15253 (pa_can_change_mode_class): ...this new function, inverting the
15254 sense of the return value.
15255 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15256 than CANNOT_CHANGE_MODE_CLASS in comments.
15257 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15258 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15259 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15260 (pdp11_cannot_change_mode_class): Replace with...
15261 (pdp11_can_change_mode_class): ...this new function, inverting the
15262 sense of the return value.
15263 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15264 * config/powerpcspe/powerpcspe-protos.h
15265 (rs6000_cannot_change_mode_class_ptr): Delete.
15266 * config/powerpcspe/powerpcspe.c
15267 (rs6000_cannot_change_mode_class_ptr): Delete.
15268 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15269 (rs6000_option_override_internal): Assign to
15270 targetm.can_change_mode_class instead of
15271 rs6000_cannot_change_mode_class_ptr.
15272 (rs6000_cannot_change_mode_class): Replace with...
15273 (rs6000_can_change_mode_class): ...this new function, inverting the
15274 sense of the return value.
15275 (rs6000_debug_cannot_change_mode_class): Replace with...
15276 (rs6000_debug_can_change_mode_class): ...this new function.
15277 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15278 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15279 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15280 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15281 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15283 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15284 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15285 (rs6000_option_override_internal): Assign to
15286 targetm.can_change_mode_class instead of
15287 rs6000_cannot_change_mode_class_ptr.
15288 (rs6000_cannot_change_mode_class): Replace with...
15289 (rs6000_can_change_mode_class): ...this new function, inverting the
15290 sense of the return value.
15291 (rs6000_debug_cannot_change_mode_class): Replace with...
15292 (rs6000_debug_can_change_mode_class): ...this new function.
15293 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15294 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15295 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15296 (s390_can_change_mode_class): ...this new function, inverting the
15297 sense of the return value.
15298 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15299 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15300 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15301 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15302 (sh_cannot_change_mode_class): Replace with...
15303 (sh_can_change_mode_class): ...this new function, inverting the
15304 sense of the return value.
15305 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15306 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15307 (sparc_can_change_mode_class): New function.
15308 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15309 * config/spu/spu.c (spu_can_change_mode_class): New function.
15310 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15311 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15312 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15313 (visium_can_change_mode_class): New function.
15314 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15316 2017-09-15 Richard Biener <rguenther@suse.de>
15318 PR tree-optimization/82217
15319 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15320 but not undefined case.
15322 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15325 * postreload.c (reload_cse_simplify_operands): Skip
15326 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15328 2017-09-15 Richard Biener <rguenther@suse.de>
15330 PR tree-optimization/68823
15331 * graphite-scop-detection.c (build_alias_set): If we have a
15332 possible dependence check whether we can handle them by just
15333 looking at the DRs DR_ACCESS_FNs.
15334 (build_scops): If build_alias_set fails, fail the SCOP.
15336 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15338 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15339 to support float128 built-in functions that require the ISA 3.0
15341 (BU_FLOAT128_3_HW): Likewise.
15342 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15343 built-in functions.
15344 (FMAF128): Likewise.
15346 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15347 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15348 floating point instructions.
15349 (rs6000_invalid_builtin): Likewise.
15350 (rs6000_builtin_mask_names): Likewise.
15351 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15352 (RS6000_BTM_FLOAT128_HW): Likewise.
15353 (RS6000_BTM_COMMON): Likewise.
15354 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15356 * doc/extend.texi (RS/6000 built-in functions): Document the
15357 IEEE 128-bit floating point square root and fused multiply-add
15358 built-in functions.
15360 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15362 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15363 reg (r2) isn't in the set of registers defined in the prologue.
15365 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15366 Alan Hayward <alan.hayward@arm.com>
15367 David Sherwood <david.sherwood@arm.com>
15369 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15370 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15371 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15372 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15373 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15375 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15376 max_vectorization_factor.
15377 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15379 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15380 Alan Hayward <alan.hayward@arm.com>
15381 David Sherwood <david.sherwood@arm.com>
15383 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15384 (vect_worthwhile_without_simd_p): Declare.
15385 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15386 (vectorizable_reduction): Use it.
15387 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15388 (vectorizable_operation): Likewise.
15390 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15391 Alan Hayward <alan.hayward@arm.com>
15392 David Sherwood <david.sherwood@arm.com>
15394 * tree-vectorizer.h (vect_get_num_copies): New function.
15395 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15396 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15397 (vectorizable_induction): Likewise.
15398 (vectorizable_live_operation): Likewise.
15399 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15400 (vectorizable_bswap): Likewise.
15401 (vectorizable_call): Likewise.
15402 (vectorizable_conversion): Likewise.
15403 (vectorizable_assignment): Likewise.
15404 (vectorizable_shift): Likewise.
15405 (vectorizable_operation): Likewise.
15406 (vectorizable_store): Likewise.
15407 (vectorizable_load): Likewise.
15408 (vectorizable_condition): Likewise.
15409 (vectorizable_comparison): Likewise.
15410 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15412 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15413 Alan Hayward <alan.hayward@arm.com>
15414 David Sherwood <david.sherwood@arm.com>
15416 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15417 of vect_init_vector.
15419 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15420 Alan Hayward <alan.hayward@arm.com>
15421 David Sherwood <david.sherwood@arm.com>
15423 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15424 an inline wrapper that provides a location.
15425 (gimple_build_vector): Likewise.
15426 * gimple-fold.c (gimple_build_vector_from_val): New function.
15427 (gimple_build_vector): Likewise.
15428 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15429 functions to build the initial value. Always return a gimple value.
15430 (get_initial_defs_for_reduction): Likewise. Only compute
15432 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15433 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15434 (vectorizable_induction): Use gimple_build_vector rather than
15437 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15438 Alan Hayward <alan.hayward@arm.com>
15439 David Sherwood <david.sherwood@arm.com>
15441 * target.h (vec_perm_indices): New typedef.
15442 (auto_vec_perm_indices): Likewise.
15443 * optabs-query.h: Include target.h
15444 (can_vec_perm_p): Take a vec_perm_indices *.
15445 * optabs-query.c (can_vec_perm_p): Likewise.
15446 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15447 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15448 * tree-vect-generic.c (lower_vec_perm): Likewise.
15449 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15450 (vect_grouped_load_supported): Likewise.
15451 (vect_shift_permute_load_chain): Likewise.
15452 (vect_permute_store_chain): Use auto_vec_perm_indices.
15453 (vect_permute_load_chain): Likewise.
15454 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15455 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15456 Update uses of can_vec_perm_p.
15457 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15458 mode with a number of elements. Take a vec_perm_indices *.
15459 (vect_create_epilog_for_reduction): Update accordingly.
15460 Use auto_vec_perm_indices.
15461 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15462 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15463 (vect_transform_slp_perm_load): Likewise.
15464 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15465 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15466 (vect_gen_perm_mask_checked): Likewise.
15467 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15468 (vect_gen_perm_mask_checked): Likewise.
15469 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15470 (vectorizable_store): Likewise.
15471 (vectorizable_load): Likewise.
15472 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15473 (vectorizable_bswap): Likewise.
15475 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15476 Alan Hayward <alan.hayward@arm.com>
15477 David Sherwood <david.sherwood@arm.com>
15479 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15480 * tree.c (build_vector): Likewise.
15481 (build_vector_from_ctor): Update accordingly.
15482 (build_vector_from_val): Likewise.
15483 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15484 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15485 * tree-vect-generic.c (add_rshift): Likewise.
15486 (expand_vector_divmod): Likewise.
15487 (optimize_vector_constructor): Likewise.
15488 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15489 (vect_transform_slp_perm_load): Likewise.
15490 (vect_schedule_slp_instance): Likewise.
15491 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15492 (vectorizable_call): Likewise.
15493 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15494 * expmed.c (make_tree): Likewise.
15495 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15496 a vector passed to build_vector.
15497 (fold_convert_const): Likewise.
15498 (exact_inverse): Likewise.
15499 (fold_ternary_loc): Likewise.
15500 (fold_relational_const): Likewise.
15501 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15502 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15503 (const_unop): Likewise. Store the reduction accumulator in a
15504 variable rather than an array.
15505 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15506 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15507 the new vector, rather than constructing it after the input arrays.
15508 (native_interpret_vector): Use auto_vec<tree> when building
15509 a vector passed to build_vector. Add elements in order.
15510 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15511 auto_vec<tree> when building a vector passed to build_vector.
15512 (vect_create_epilog_for_reduction): Likewise.
15513 (vectorizable_induction): Likewise.
15514 (get_initial_def_for_reduction): Likewise. Fix indentation of
15516 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15518 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15519 passed to build_vector.
15521 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15522 Alan Hayward <alan.hayward@arm.com>
15523 David Sherwood <david.sherwood@arm.com>
15525 * tree-core.h (tree_base::u): Add an "nelts" field.
15526 (tree_vector): Use VECTOR_CST_NELTS as the length.
15527 * tree.c (tree_size): Likewise.
15528 (make_vector): Initialize VECTOR_CST_NELTS.
15529 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15530 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15531 TYPE_VECTOR_SUBPARTS.
15532 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15533 as the number of units, setting it from VECTOR_CST_NELTS.
15534 (const_vector_from_tree): Likewise.
15535 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15536 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15537 (fold_negate_expr_1): Likewise.
15538 (fold_convert_const): Likewise.
15539 (const_binop): Likewise. Differentiate the number of output and
15541 (const_unop): Likewise.
15542 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15543 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15544 in cases that did the opposite.
15546 2017-09-14 Richard Biener <rguenther@suse.de>
15548 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15551 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15553 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15555 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15558 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15559 if and where to split a bb, except for splitting before debug insn
15560 sequences followed by non-label real insn. Delete debug insns
15561 in between basic blocks.
15563 * combine.c (make_compound_operation_int): Formatting fixes.
15565 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15566 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15567 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15568 * config/sol2.h (LINK_EH_SPEC): Likewise.
15569 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15570 * config/s390/linux.h (LINK_SPEC): Likewise.
15571 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15572 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15573 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15574 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15575 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15576 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15577 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15578 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15580 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15582 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15584 (ENDFILE_LINUX_SPEC): Likewise.
15585 (LINK_EH_SPEC): Likewise.
15586 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15587 (LINK_OS_LINUX_SPEC32): Likewise.
15588 (LINK_OS_LINUX_SPEC64): Likewise.
15589 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15590 (LINK_OS_LINUX_SPEC): Likewise.
15592 2017-09-13 Martin Liska <mliska@suse.cz>
15594 PR middle-end/82154
15595 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15596 CASE_HIGH is NULL_TREE.
15598 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15599 Alan Hayward <alan.hayward@arm.com>
15600 David Sherwood <david.sherwood@arm.com>
15602 * target.def (secondary_memory_needed): New hook.
15603 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15604 instead of SECONDARY_MEMORY_NEEDED.
15605 (secondary_memory_needed_mode): Likewise.
15606 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15607 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15608 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15609 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15610 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15611 * doc/tm.texi: Regenerate.
15612 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15613 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15614 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15615 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15616 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15617 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15618 mode argument first and change the reg_class arguments to reg_class_t.
15619 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15620 Make static. Update the call to inline_secondary_memory_needed.
15621 (ix86_register_move_cost): Update the call to
15622 inline_secondary_memory_needed.
15623 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15624 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15626 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15627 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15629 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15630 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15631 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15632 and match hook interface. Add comment from mips.h.
15633 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15634 * config/mmix/mmix.md (truncdfsf2): Refer to
15635 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15637 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15638 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15639 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15640 (pa_secondary_memory_needed): New function.
15641 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15642 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15643 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15644 (pdp11_secondary_memory_needed): Make static and match hook interface.
15645 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15646 * config/powerpcspe/powerpcspe-protos.h
15647 (rs6000_secondary_memory_needed_ptr): Delete.
15648 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15650 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15651 (rs6000_option_override_internal): Assign to
15652 targetm.secondary_memory_needed rather than
15653 rs6000_secondary_memory_needed_ptr.
15654 (rs6000_secondary_memory_needed): Match hook interface.
15655 (rs6000_debug_secondary_memory_needed): Likewise.
15656 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15657 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15658 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15659 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15660 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15661 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15663 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15664 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15665 (rs6000_option_override_internal): Assign to
15666 targetm.secondary_memory_needed rather than
15667 rs6000_secondary_memory_needed_ptr.
15668 (rs6000_secondary_memory_needed): Match hook interface.
15669 (rs6000_debug_secondary_memory_needed): Likewise.
15670 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15671 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15672 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15673 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15674 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15675 (sparc_secondary_memory_needed): New function.
15676 * lra-constraints.c (check_and_process_move): Refer to
15677 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15679 (curr_insn_transform): Likewise.
15680 (process_alt_operands): Use targetm.secondary_memory_needed
15681 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15682 (check_secondary_memory_needed_p): Likewise.
15683 (choose_split_class): Likewise.
15684 * reload.c: Unconditionally include code that was previously
15685 conditional on SECONDARY_MEMORY_NEEDED.
15686 (push_secondary_reload): Use targetm.secondary_memory_needed
15687 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15688 (push_reload): Likewise.
15689 * reload1.c: Unconditionally include code that was previously
15690 conditional on SECONDARY_MEMORY_NEEDED.
15691 (choose_reload_regs): Use targetm.secondary_memory_needed
15692 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15693 (gen_reload): Likewise.
15694 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15696 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15697 Alan Hayward <alan.hayward@arm.com>
15698 David Sherwood <david.sherwood@arm.com>
15700 * target.def (secondary_memory_needed_mode): New hook:
15701 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15702 * targhooks.h (default_secondary_memory_needed_mode): New function.
15703 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15704 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15705 * doc/tm.texi: Regenerate.
15706 * lra-constraints.c (check_and_process_move): Use
15707 targetm.secondary_memory_needed_mode instead of
15708 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15709 (curr_insn_transform): Likewise.
15710 * reload.c (get_secondary_mem): Likewise.
15711 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15712 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
15714 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15715 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15716 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
15717 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15718 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
15720 * config/powerpcspe/powerpcspe-protos.h
15721 (rs6000_secondary_memory_needed_mode): Delete.
15722 * config/powerpcspe/powerpcspe.c
15723 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15724 (rs6000_secondary_memory_needed_mode): Make static.
15725 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15726 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
15728 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15730 (rs6000_secondary_memory_needed_mode): Make static.
15731 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15732 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
15733 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15734 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15735 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15737 (sparc_secondary_memory_needed_mode): New function.
15738 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
15740 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15742 * config/aarch64/constraints.md (Umq): New constraint.
15743 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15745 (mov<mode>): Update condition.
15747 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15749 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15750 when bitposition is the same.
15752 2017-09-13 Richard Biener <rguenther@suse.de>
15754 * dwarf2out.c (output_die_symbol): Remove.
15755 (output_die): Do not output a DIEs symbol.
15757 2017-09-13 Richard Biener <rguenther@suse.de>
15759 PR middle-end/82128
15760 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15761 default-def to avoid breaking iterator update with the weird
15762 interaction with cgraph_update_edges_for_call_stmt_node.
15764 2017-09-13 Richard Biener <rguenther@suse.de>
15766 * tree-cfg.c (verify_gimple_assign_binary): Add verification
15767 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
15768 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
15769 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
15771 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15773 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15774 Disable pc relative literal load irrespective of
15775 TARGET_FIX_ERR_A53_84341 for default.
15777 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
15779 * config/sparc/sparc.c (output_return): Output the source location of
15780 the insn in the delay slot, if any.
15781 (output_sibcall): Likewise.
15783 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
15786 * common.opt (-static-pie): New alias.
15787 (shared): Negate static-pie.
15788 (-no-pie): Update help text.
15790 (static-pie): New option.
15791 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
15792 -static-pie support.
15793 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
15794 (LINK_EH_SPEC): Likewise.
15795 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15796 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15797 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15798 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15799 (init_gcc_specs): Likewise.
15800 (init_spec): Likewise.
15801 (display_help): Update help message for -pie.
15802 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15805 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15807 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15808 (movdi_aarch64): Likewise.
15809 (movti_aarch64): Likewise.
15811 2017-09-12 Simon Wright <simon@pushface.org>
15814 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15815 calculation of the minor version, always output as 0.
15817 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15820 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15821 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15822 on it early, rather than manual conversion late. For
15823 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15824 instead of performing manual conversion.
15826 2017-09-12 Carl Love <cel@us.ibm.com>
15828 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15829 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15831 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15832 VMULOSW): Add definitions.
15833 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15834 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15835 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15836 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15837 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15839 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15841 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15843 (movti_aarch64): Likewise.
15844 (movdf_aarch64): Likewise.
15845 (movtf_aarch64): Likewise.
15846 (load_pairdi): Likewise.
15847 (store_pairdi): Likewise.
15848 (load_pairdf): Likewise.
15849 (store_pairdf): Likewise.
15850 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15851 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15852 (ldr_got_small_<mode>): Likewise.
15853 (ldr_got_small_28k_<mode>): Likewise.
15854 (ldr_got_tiny): Likewise.
15855 * config/aarch64/iterators.md (ldst_sz): New.
15856 (ldpstp_sz): Likewise.
15857 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15859 (thunderx_load): Split load_8 to load_16.
15860 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15862 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15863 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15864 (xgene1_store_pair): Split store_8 to store_16.
15865 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15866 (falkor_st_0_st_sd): Split store_8 to store_16.
15868 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15870 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15871 and store1/2/3/4 to store_4/8/12/16.
15872 * config/aarch64/aarch64.md: Update for rename.
15873 * config/arm/arm.md: Likewise.: Likewise.
15874 * config/arm/arm.c: Likewise.
15875 * config/arm/thumb1.md: Likewise.
15876 * config/arm/thumb2.md: Likewise.
15877 * config/arm/vfp.md: Likewise.
15878 * config/arm/arm-generic.md: Likewise.
15879 * config/arm/arm1020e.md: Likewise.
15880 * config/arm/arm1026ejs.md: Likewise.
15881 * config/arm/arm1136jfs.md: Likewise.
15882 * config/arm/arm926ejs.md: Likewise.
15883 * config/arm/cortex-a15.md: Likewise.
15884 * config/arm/cortex-a17.md: Likewise.
15885 * config/arm/cortex-a5.md: Likewise.
15886 * config/arm/cortex-a53.md: Likewise.
15887 * config/arm/cortex-a57.md: Likewise.
15888 * config/arm/cortex-a7.md: Likewise.
15889 * config/arm/cortex-a8.md: Likewise.
15890 * config/arm/cortex-a9.md: Likewise.
15891 * config/arm/cortex-m4.md: Likewise.
15892 * config/arm/cortex-m7.md: Likewise.
15893 * config/arm/cortex-r4.md: Likewise.
15894 * config/arm/exynos-m1.md: Likewise.
15895 * config/arm/fa526.md: Likewise.
15896 * config/arm/fa606te.md: Likewise.
15897 * config/arm/fa626te.md: Likewise.
15898 * config/arm/fa726te.md: Likewise.
15899 * config/arm/fmp626.md: Likewise.
15900 * config/arm/iwmmxt.md: Likewise.
15901 * config/arm/ldmstm.md: Likewise.
15902 * config/arm/marvell-pj4.md: Likewise.
15903 * config/arm/xgene1.md: Likewise.
15904 * config/aarch64/thunderx.md: Likewise.
15905 * config/aarch64/thunderx2t99.md: Likewise.
15906 * config/aarch64/falkor.md: Likewise.
15908 2017-09-12 Martin Liska <mliska@suse.cz>
15910 * attribs.c (private_lookup_attribute): New function.
15911 * attribs.h (private_lookup_attribute): Declared here.
15912 (lookup_attribute): Called from this place.
15914 2017-09-12 Richard Biener <rguenther@suse.de>
15916 PR tree-optimization/82157
15917 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15918 stmts with side-effects.
15920 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15921 Alan Hayward <alan.hayward@arm.com>
15922 David Sherwood <david.sherwood@arm.com>
15924 * target.def (hard_regno_nregs): New hook.
15925 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15926 * targhooks.h (default_hard_regno_nregs): Declare.
15927 * targhooks.c (default_hard_regno_nregs): New function.
15928 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15929 (TARGET_HARD_REGNO_NREGS): ...this hook.
15930 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15931 (CLASS_MAX_NREGS): Likewise.
15932 * doc/tm.texi: Regenerate.
15933 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15934 instead of HARD_REGNO_NREGS.
15935 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15936 HARD_REGNO_NREGS in the comment.
15937 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15938 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15939 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15940 Return an unsigned int.
15941 (TARGET_HARD_REGNO_NREGS): Redefine.
15942 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15943 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15944 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15945 (arc_hard_regno_nregs): New function.
15946 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15947 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15948 (arm_hard_regno_nregs): New function.
15949 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15950 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15951 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15952 (TARGET_HARD_REGNO_NREGS): Redefine.
15953 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15954 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15955 (HARD_REGNO_NREGS): Delete.
15956 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15957 (cr16_hard_regno_nregs): New function.
15958 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15959 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15960 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15961 (cris_hard_regno_nregs): New function.
15962 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15963 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15964 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15965 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15966 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15967 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15968 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15969 (frv_hard_regno_nregs): Make static. Take and return an
15971 (frv_class_max_nregs): Remove outdated copy of documentation.
15972 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15973 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15974 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15975 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15976 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15977 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15978 (TARGET_HARD_REGNO_NREGS): Redefine.
15979 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15980 (CLASS_MAX_NREGS): Update comment.
15981 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15982 (ia64_hard_regno_nregs): New function.
15983 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15984 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15985 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15986 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15987 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15989 (m32c_hard_regno_nregs): Likewise. Make static.
15990 (TARGET_HARD_REGNO_NREGS): Redefine.
15991 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15992 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15993 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15994 (m68k_hard_regno_nregs): New function.
15995 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15996 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15997 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15998 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15999 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16000 Take and return an unsigned int.
16001 (TARGET_HARD_REGNO_NREGS): Redefine.
16002 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16003 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16004 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16005 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16006 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16007 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16008 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16009 (msp430_hard_regno_nregs): Make static. Take and return an
16011 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16012 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16013 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16014 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16015 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16016 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16017 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16018 (TARGET_HARD_REGNO_NREGS): Redefine.
16019 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16020 (PA_HARD_REGNO_NREGS): ...this.
16021 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16022 (PA_HARD_REGNO_NREGS): ...this.
16023 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16024 (pa_hard_regno_nregs): New function.
16025 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16026 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16027 (pdp11_hard_regno_nregs): New function.
16028 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16029 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16030 (rs6000_hard_regno_nregs_hook): New function.
16031 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16032 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16033 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16034 Take and return an unsigned int. Move earlier in file.
16035 (TARGET_HARD_REGNO_NREGS): Redefine.
16036 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16037 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16038 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16039 (rl78_hard_regno_nregs): Make static. Take and return an
16041 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16042 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16043 (rs6000_hard_regno_nregs_hook): New function.
16044 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16045 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16046 (TARGET_HARD_REGNO_NREGS): Redefine.
16047 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16048 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16049 instead of HARD_REGNO_NREGS.
16050 (s390_hard_regno_nregs): New function.
16051 (s390_hard_regno_mode_ok): Add comment from s390.h.
16052 (TARGET_HARD_REGNO_NREGS): Redefine.
16053 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16054 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16055 (sh_hard_regno_nregs): New function.
16056 (sh_pass_in_reg_p): Use it.
16057 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16058 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16059 (sparc_hard_regno_nregs): New function.
16060 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16061 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16062 (spu_function_arg_advance): Use it, supplying a valid register number.
16063 (TARGET_HARD_REGNO_NREGS): Redefine.
16064 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16065 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16066 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16067 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16068 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16069 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16070 (CLASS_MAX_NREGS): Remove copy of old documentation.
16071 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16072 (visium_hard_regno_nregs): New function.
16073 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16074 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16075 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16076 xtensa_hard_regno_nregs): New function.
16077 * system.h (HARD_REGNO_NREGS): Poison.
16079 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16081 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16082 hard_regno_nregs instead of HARD_REGNO_NREGS.
16083 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16084 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16085 (c6x_expand_epilogue): Likewise.
16086 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16087 (frv_read_iacc_argument): Likewise.
16088 * config/sh/sh.c: Include regs.h.
16089 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16090 (regs_used): Likewise.
16091 (output_stack_adjust): Likewise.
16092 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16093 * expmed.c: Include regs.h.
16094 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16095 * ree.c: Include regs.h.
16096 (combine_reaching_defs): Use hard_regno_nregs instead of
16098 (add_removable_extension): Likewise.
16100 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16102 * regs.h (hard_regno_nregs): Turn into a function.
16103 (end_hard_regno): Update accordingly.
16104 * caller-save.c (setup_save_areas): Likewise.
16105 (save_call_clobbered_regs): Likewise.
16106 (replace_reg_with_saved_mem): Likewise.
16107 (insert_restore): Likewise.
16108 (insert_save): Likewise.
16109 * combine.c (can_change_dest_mode): Likewise.
16110 (move_deaths): Likewise.
16111 (distribute_notes): Likewise.
16112 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16113 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16114 (rs6000_split_multireg_move): Likewise.
16115 (rs6000_register_move_cost): Likewise.
16116 (rs6000_memory_move_cost): Likewise.
16117 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16118 (rs6000_split_multireg_move): Likewise.
16119 (rs6000_register_move_cost): Likewise.
16120 (rs6000_memory_move_cost): Likewise.
16121 * cselib.c (cselib_reset_table): Likewise.
16122 (cselib_lookup_1): Likewise.
16123 * emit-rtl.c (set_mode_and_regno): Likewise.
16124 * function.c (aggregate_value_p): Likewise.
16125 * ira-color.c (setup_profitable_hard_regs): Likewise.
16126 (check_hard_reg_p): Likewise.
16127 (calculate_saved_nregs): Likewise.
16128 (assign_hard_reg): Likewise.
16129 (improve_allocation): Likewise.
16130 (calculate_spill_cost): Likewise.
16131 * ira-emit.c (modify_move_list): Likewise.
16132 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16133 (ira_hard_reg_in_set_p): Likewise.
16134 * ira.c (setup_reg_mode_hard_regset): Likewise.
16135 (clarify_prohibited_class_mode_regs): Likewise.
16136 (check_allocation): Likewise.
16137 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16138 (lra_setup_reg_renumber): Likewise.
16139 (setup_try_hard_regno_pseudos): Likewise.
16140 (spill_for): Likewise.
16141 (assign_hard_regno): Likewise.
16142 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16143 * lra-constraints.c (in_class_p): Likewise.
16144 (lra_constraint_offset): Likewise.
16145 (simplify_operand_subreg): Likewise.
16146 (lra_constraints): Likewise.
16147 (split_reg): Likewise.
16148 (split_if_necessary): Likewise.
16149 (invariant_p): Likewise.
16150 (inherit_in_ebb): Likewise.
16151 * lra-lives.c (process_bb_lives): Likewise.
16152 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16153 (get_hard_regs): Likewise.
16154 (do_remat): Likewise.
16155 * lra-spills.c (assign_spill_hard_regs): Likewise.
16156 * mode-switching.c (create_pre_exit): Likewise.
16157 * postreload.c (reload_combine_recognize_pattern): Likewise.
16158 * recog.c (peep2_find_free_register): Likewise.
16159 * regcprop.c (kill_value_regno): Likewise.
16160 (set_value_regno): Likewise.
16161 (copy_value): Likewise.
16162 (maybe_mode_change): Likewise.
16163 (find_oldest_value_reg): Likewise.
16164 (copyprop_hardreg_forward_1): Likewise.
16165 * regrename.c (check_new_reg_p): Likewise.
16166 (regrename_do_replace): Likewise.
16167 * reload.c (push_reload): Likewise.
16168 (combine_reloads): Likewise.
16169 (find_dummy_reload): Likewise.
16170 (operands_match_p): Likewise.
16171 (find_reloads): Likewise.
16172 (find_equiv_reg): Likewise.
16173 (reload_adjust_reg_for_mode): Likewise.
16174 * reload1.c (count_pseudo): Likewise.
16175 (count_spilled_pseudo): Likewise.
16176 (find_reg): Likewise.
16177 (clear_reload_reg_in_use): Likewise.
16178 (free_for_value_p): Likewise.
16179 (allocate_reload_reg): Likewise.
16180 (choose_reload_regs): Likewise.
16181 (reload_adjust_reg_for_temp): Likewise.
16182 (emit_reload_insns): Likewise.
16183 (delete_output_reload): Likewise.
16184 * rtlanal.c (subreg_get_info): Likewise.
16185 * sched-deps.c (sched_analyze_reg): Likewise.
16186 * sel-sched.c (init_regs_for_mode): Likewise.
16187 (mark_unavailable_hard_regs): Likewise.
16188 (choose_best_reg_1): Likewise.
16189 (verify_target_availability): Likewise.
16190 * valtrack.c (dead_debug_insert_temp): Likewise.
16191 * var-tracking.c (track_loc_p): Likewise.
16192 (emit_note_insn_var_location): Likewise.
16193 * varasm.c (make_decl_rtl): Likewise.
16194 * reginfo.c (choose_hard_reg_mode): Likewise.
16195 (init_reg_modes_target): Refer directly to
16196 this_target_regs->x_hard_regno_nregs.
16198 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16200 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16201 instead of hard_regno_nregs.
16203 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16205 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16206 end_hard_regno instead of hard_regno_nregs.
16207 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16208 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16209 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16210 * ira-color.c (improve_allocation): Likewise.
16211 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16212 * lra-lives.c (mark_regno_live): Likewise.
16213 (mark_regno_dead): Likewise.
16214 * lra-remat.c (operand_to_remat): Likewise.
16215 * lra.c (collect_non_operand_hard_regs): Likewise.
16216 * postreload.c (reload_combine_note_store): Likewise.
16217 (move2add_valid_value_p): Likewise.
16218 * reload.c (regno_clobbered_p): Likewise.
16220 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16222 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16224 * config/v850/v850.c (v850_reorg): Likewise.
16225 * reload.c (refers_to_regno_for_reload_p): Likewise.
16226 (find_equiv_reg): Likewise.
16227 * reload1.c (reload_reg_reaches_end_p): Likewise.
16229 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16231 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16233 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16234 * config/arm/arm.c (output_move_neon): Likewise.
16235 (arm_attr_length_move_neon): Likewise.
16236 (neon_split_vcombine): Likewise.
16237 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16238 (c6x_mark_reg_written): Likewise.
16239 (c6x_dwarf_register_span): Likewise.
16240 * config/i386/i386.c (ix86_save_reg): Likewise.
16241 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16242 (rws_access_reg): Likewise.
16243 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16244 * mode-switching.c (create_pre_exit): Likewise.
16245 * ree.c (combine_reaching_defs): Likewise.
16246 (add_removable_extension): Likewise.
16247 * regcprop.c (find_oldest_value_reg): Likewise.
16248 (copyprop_hardreg_forward_1): Likewise.
16249 * reload.c (reload_inner_reg_of_subreg): Likewise.
16250 (push_reload): Likewise.
16251 (combine_reloads): Likewise.
16252 (find_dummy_reload): Likewise.
16253 (reload_adjust_reg_for_mode): Likewise.
16254 * reload1.c (find_reload_regs): Likewise.
16255 (forget_old_reloads_1): Likewise.
16256 (reload_reg_free_for_value_p): Likewise.
16257 (reload_adjust_reg_for_temp): Likewise.
16258 (emit_reload_insns): Likewise.
16259 (delete_output_reload): Likewise.
16260 * sel-sched.c (choose_best_reg_1): Likewise.
16261 (choose_best_pseudo_reg): Likewise.
16263 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16264 Alan Hayward <alan.hayward@arm.com>
16265 David Sherwood <david.sherwood@arm.com>
16267 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16268 * target.def (slow_unaligned_access): New hook.
16269 * targhooks.h (default_slow_unaligned_access): Declare.
16270 * targhooks.c (default_slow_unaligned_access): New function.
16271 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16272 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16273 * doc/tm.texi: Regenerate.
16274 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16275 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16276 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16278 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16279 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16281 (rs6000_slow_unaligned_access): New function.
16282 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16283 (expand_block_compare): Likewise.
16284 (expand_strn_compare): Likewise.
16285 (rs6000_rtx_costs): Likewise.
16286 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16287 (riscv_slow_unaligned_access): Likewise.
16288 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16289 (riscv_slow_unaligned_access_p): ...this and make static.
16290 (riscv_option_override): Update accordingly.
16291 (riscv_slow_unaligned_access): New function.
16292 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16293 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16294 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16295 (rs6000_slow_unaligned_access): New function.
16296 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16297 (rs6000_rtx_costs): Likewise.
16298 * config/rs6000/rs6000-string.c (expand_block_compare)
16299 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16300 of SLOW_UNALIGNED_ACCESS.
16301 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16302 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16303 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16304 of SLOW_UNALIGNED_ACCESS.
16305 * expmed.c (simple_mem_bitfield_p): Likewise.
16306 * expr.c (alignment_for_piecewise_move): Likewise.
16307 (emit_group_load_1): Likewise.
16308 (emit_group_store): Likewise.
16309 (copy_blkmode_from_reg): Likewise.
16310 (emit_push_insn): Likewise.
16311 (expand_assignment): Likewise.
16312 (store_field): Likewise.
16313 (expand_expr_real_1): Likewise.
16314 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16315 * lra-constraints.c (simplify_operand_subreg): Likewise.
16316 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16317 * gimple-ssa-store-merging.c: Likewise in block comment at start
16319 * tree-ssa-strlen.c: Include target.h.
16320 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16321 of SLOW_UNALIGNED_ACCESS.
16322 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16324 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16326 PR rtl-optimization/82185
16327 * expmed.c (emit_store_flag_int): Only test tem if it has been
16330 2017-09-12 Richard Biener <rguenther@suse.de>
16332 PR middle-end/82149
16333 * match.pd ((FTYPE) N CMP CST): Fix typo.
16335 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16337 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16339 (mips_near_type_p): Add 'short_call' attribute as a synonym
16341 * doc/extend.texi (short_call): Document new function attribute.
16343 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16346 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16347 assertion check that in the condition.
16348 (get_atomic_generic_size): Likewise. Before testing if parameter
16349 has pointer type, if it has array type, call for C++
16350 default_conversion to perform array-to-pointer conversion.
16352 2017-09-12 Richard Biener <rguenther@suse.de>
16354 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16355 for operations we cannot scalarize.
16357 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16359 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16360 vectors heap vectors. Clean up comments.
16361 Make visited_bbs a reference.
16362 (profitable_jump_thread_path): Make GC
16363 vectors heap vectors. Clean up comments.
16365 (convert_and_register_jump_thread_path): Make GC vectors heap
16367 (check_subpath_and_update_thread_path): Same. Clean up comments.
16368 Make visited_bbs a reference.
16369 (handle_phi): Abstract common code to to
16370 register_jump_thread_path_if_profitable.
16371 Rename VAR_BB to DEF_BB.
16373 Make GC vectors heap vectors.
16374 Make visited_bbs a reference.
16375 (handle_assignment): Same.
16376 (register_jump_thread_path_if_profitable): New.
16377 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16379 Make GC vectors heap vectors. Clean up comments.
16380 Make visited_bbs a reference.
16381 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16382 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16385 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16388 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16389 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16391 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16393 Revert r251800 and r251799.
16395 2017-09-11 Martin Jambor <mjambor@suse.cz>
16398 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16399 arguments in advance.
16400 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16401 use it to find predecessor edges.
16402 (naive_outof_ssa): Collect vector of predecessors.
16404 2017-09-08 Jason Merrill <jason@redhat.com>
16406 PR c++/70029 - ICE with ref-qualifier and -flto
16407 * langhooks.h (struct lang_hooks_for_types): Add
16408 copy_lang_qualifiers.
16409 * attribs.c (build_type_attribute_qual_variant): Use it.
16410 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16412 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16413 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16415 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16418 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16419 (*mulsi3_sp64): New instruction.
16420 (mulsi3): New expander.
16422 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16424 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16426 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16428 * sancov.c: Include memmodel.h.
16430 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16433 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16436 2017-09-07 Carl Love <cel@us.ibm.com>
16438 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16439 the sldi instruction.
16441 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16443 * sancov.c: Include tm_p.h.
16445 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16448 * output.h (switch_to_other_text_partition): New declaration.
16449 * varasm.c (switch_to_other_text_partition): New function.
16450 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16451 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16452 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16453 to the other text partition before emitting LCL label and switch back
16454 after emitting the word after it.
16456 2017-09-07 Richard Biener <rguenther@suse.de>
16458 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16459 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16460 critical edges here, after loop init.
16461 (pass_data_pre): Remove PROP_no_crit_edges flags.
16462 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16463 for valueization of call args to avoid leaking VN_TOP.
16464 (visit_use): Assert we do not visit default defs.
16465 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16466 Use error_mark_node to more easily detect leaking VN_TOP.
16467 All default-defs are varying, not VN_TOP. Mark them visited.
16468 (run_scc_vn): Make code match comment.
16470 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16472 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16473 OPTION_MASK_FLOAT128_KEYWORD.
16474 (POWERPC_MASKS): Likewise.
16475 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16476 support for the -mfloat128-type option, and make -mfloat128
16477 default on PowerPC Linux systems. Define or undefine
16478 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16479 Define __float128 to be __ieee128 if IEEE 128-bit support is
16480 enabled, or undefine it.
16481 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16482 Delete defining __FLOAT128_TYPE__.
16483 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16484 -mfloat128-type option and make -mfloat128 default on PowerPC
16486 (TARGET_FLOAT128_TYPE): Likewise.
16487 (-mfloat128-type): Likewise.
16488 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16489 Delete the -mfloat128-type option and make -mfloat128 default on
16490 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16491 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16492 128-bit floating point is enabled. Change tests from using
16493 -mfloat128-type to -mfloat128.
16494 (rs6000_mangle_type): Use the correct mangling for the __float128
16495 type even if normal long double is restricted to 64-bits.
16496 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16498 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16499 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16500 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16502 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16503 documentation for -mfloat128.
16505 2017-09-06 Olivier Hainque <hainque@adacore.com>
16507 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16509 2017-09-06 Wish Wu <wishwu007@gmail.com>
16510 Jakub Jelinek <jakub@redhat.com>
16512 * asan.c (initialize_sanitizer_builtins): Add
16513 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16514 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16515 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16516 BT_FN_VOID_UINT64_PTR variables.
16517 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16518 (BT_FN_VOID_UINT16_UINT16): Likewise.
16519 (BT_FN_VOID_UINT32_UINT32): Likewise.
16520 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16521 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16522 (BT_FN_VOID_UINT64_PTR): Likewise.
16523 * common.opt (flag_sanitize_coverage): New variable.
16524 (fsanitize-coverage=trace-pc): Remove.
16525 (fsanitize-coverage=): Add.
16526 * flag-types.h (enum sanitize_coverage_code): New enum.
16527 * fold-const.c (fold_range_test): Disable non-short-circuit
16528 optimization if flag_sanitize_coverage.
16529 (fold_truth_andor): Likewise.
16530 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16531 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16532 (coverage_sanitizer_opts): New array.
16533 (get_closest_sanitizer_option): Add OPTS argument, handle also
16534 OPT_fsanitize_coverage_.
16535 (parse_sanitizer_options): Adjusted to also handle
16536 OPT_fsanitize_coverage_.
16537 (common_handle_option): Add OPT_fsanitize_coverage_.
16538 * sancov.c (instrument_comparison, instrument_switch): New function.
16539 (sancov_pass): Add trace-cmp support.
16540 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16541 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16542 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16543 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16544 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16545 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16546 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16547 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16548 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16549 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16551 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16553 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16554 error. Only quit immediately if parsing is complete.
16555 (BEGIN): Initialize fatal_err and parse_done.
16556 (begin fpu, end fpu): Check number of arguments.
16557 (begin arch, end arch): Likewise.
16558 (begin cpu, end cpu): Likewise.
16559 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16560 (optalias): Likewise.
16562 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16564 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16565 * config/arm/arm-isa.h: Delete. Move definitions to ...
16566 * arm-cpus.in: ... here. Use new feature and fgroup values.
16567 * config/arm/arm.c (arm_option_override): Use lower case for feature
16569 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16570 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16571 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16573 (print_isa_bits_for): New function.
16574 (gen_isa): New function.
16575 (gen_comm_data): Use print_isa_bits_for.
16576 (define feature): New keyword.
16577 (define fgroup): New keyword.
16578 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16579 (arm-isa.h): Add rule to generate file.
16580 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16581 case for feature bit names.
16583 2017-09-06 Richard Biener <rguenther@suse.de>
16585 * tree-ssa-pre.c (NECESSARY): Remove.
16586 (create_expression_by_pieces): Do not touch pass-local flags.
16587 (insert_into_preds_of_block): Likewise.
16588 (do_pre_regular_insertion): Likewise.
16589 (eliminate_insert): Likewise.
16590 (eliminate_dom_walker::before_dom_children): Likewise.
16591 (fini_eliminate): Do not look at inserted_exprs.
16592 (mark_operand_necessary): Remove.
16593 (remove_dead_inserted_code): Replace with simple work-list
16594 algorithm based on inserted_exprs and SSA uses.
16595 (pass_pre::execute): Re-order fini_eliminate and
16596 remove_dead_inserted_code.
16598 2017-09-06 Olivier Hainque <hainque@adacore.com>
16600 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16601 for VxWorks 7. Adjust surrounding comments.
16603 2017-09-06 Richard Biener <rguenther@suse.de>
16605 * gimple-ssa-strength-reduction.c
16606 (find_candidates_dom_walker::before_dom_children): Also allow
16609 2017-09-06 Richard Biener <rguenther@suse.de>
16611 PR tree-optimization/82108
16612 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16613 for gap in the non-permutation SLP case.
16615 2017-09-06 Martin Jambor <mjambor@suse.cz>
16617 PR tree-optimization/82078
16618 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16619 add_access_to_work_queue...
16620 (build_accesses_from_assign): ...here.
16621 (propagate_all_subaccesses): Make sure racc is the group
16622 representative, if there is one.
16624 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16626 PR middle-end/82095
16627 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16630 2017-09-06 Richard Biener <rguenther@suse.de>
16632 * gimple-ssa-strength-reduction.c
16633 (find_candidates_dom_walker::before_doom_children): Use a
16634 type and not a mode check.
16636 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16639 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16641 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16643 2017-09-05 Jeff Law <law@redhat.com>
16645 PR tree-optimization/64910
16646 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16647 cases where we have 3 or more operands.
16649 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16651 PR middle-end/81768
16652 * omp-low.c (lower_omp_for): Recompute tree invariant if
16653 gimple_omp_for_initial/final is ADDR_EXPR.
16655 PR middle-end/81768
16656 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16657 into gimple val before gimplification fo the COND_EXPR.
16659 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16661 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16662 REGION_COPY argument.
16663 (thread_through_all_blocks): Remove unused argument to
16664 duplicate_thread_path.
16666 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16667 Alan Hayward <alan.hayward@arm.com>
16668 David Sherwood <david.sherwood@arm.com>
16670 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16671 Take a scalar_mode rather than a machine_mode.
16672 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16673 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16674 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16675 (aarch64_gen_adjusted_ldpstp): Likewise.
16676 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16678 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16679 Alan Hayward <alan.hayward@arm.com>
16680 David Sherwood <david.sherwood@arm.com>
16682 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16683 Take a scalar_int_mode instead of a machine_mode.
16684 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16685 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16686 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16687 (aarch64_simd_attr_length_rglist): Delete.
16688 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16689 a scalar_int_mode instead of a machine_mode.
16690 (aarch64_add_offset): Likewise.
16691 (aarch64_internal_mov_immediate): Likewise
16692 (aarch64_add_constant_internal): Likewise.
16693 (aarch64_add_constant): Likewise.
16694 (aarch64_movw_imm): Likewise.
16695 (aarch64_rtx_arith_op_extract_p): Likewise.
16696 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16697 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16698 Remove assert that the mode isn't a vector.
16699 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16700 (aarch64_expand_mov_immediate): Update calls after above changes.
16701 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16702 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16703 (aarch64_move_imm): Likewise.
16704 (aarch64_can_const_movi_rtx_p): Likewise.
16705 (aarch64_strip_extend): Likewise.
16706 (aarch64_extr_rtx_p): Likewise.
16707 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16708 a CONST_INT when the mode parameter is VOIDmode.
16709 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16711 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16713 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
16714 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16715 (bitwise_type_for_mode): Update accordingly.
16717 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16719 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
16720 * stor-layout.c (mode_for_size_tree): Likewise.
16721 (mode_for_array): Update accordingly.
16722 (layout_decl): Likewise.
16723 (compute_record_mode): Likewise. Only set the mode once.
16725 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16727 * target.def (get_mask_mode): Change return type to opt_mode.
16729 * doc/tm.texi: Regenerate.
16730 * targhooks.h (default_get_mask_mode): Return an opt_mode.
16731 * targhooks.c (default_get_mask_mode): Likewise.
16732 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16733 * optabs-query.c (can_vec_mask_load_store_p): Update use of
16734 targetm.get_mask_mode.
16735 * tree.c (build_truth_vector_type): Likewise.
16737 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16739 * machmode.h (mode_for_vector): Return an opt_mode.
16740 * stor-layout.c (mode_for_vector): Likewise.
16741 (mode_for_int_vector): Update accordingly.
16742 (layout_type): Likewise.
16743 * config/i386/i386.c (emit_memmov): Likewise.
16744 (ix86_expand_set_or_movmem): Likewise.
16745 (ix86_expand_vector_init): Likewise.
16746 (ix86_get_mask_mode): Likewise.
16747 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16749 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16750 * expmed.c (extract_bit_field_1): Likewise.
16751 * expr.c (expand_expr_real_2): Likewise.
16752 * optabs-query.c (can_vec_perm_p): Likewise.
16753 (can_vec_mask_load_store_p): Likewise.
16754 * optabs.c (expand_vec_perm): Likewise.
16755 * targhooks.c (default_get_mask_mode): Likewise.
16756 * tree-vect-stmts.c (vectorizable_store): Likewise.
16757 (vectorizable_load): Likewise.
16758 (get_vectype_for_scalar_type_and_size): Likewise.
16760 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16762 * machmode.h (mode_for_int_vector): New function.
16763 * stor-layout.c (mode_for_int_vector): Likewise.
16764 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16765 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
16766 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16767 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16768 (s390_expand_vcond): Likewise.
16770 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16772 * machmode.h (opt_machine_mode): New type.
16773 (opt_mode<T>): Allow construction from anything that can be
16775 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16776 (mode_for_size): Return an opt_machine_mode.
16777 * stor-layout.c (mode_for_size): Likewise.
16778 (mode_for_size_tree): Update call accordingly.
16779 (bitwise_mode_for_mode): Likewise.
16780 (make_fract_type): Likewise.
16781 (make_accum_type): Likewise.
16782 * caller-save.c (replace_reg_with_saved_mem): Update call
16784 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16785 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16786 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16787 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16788 * expmed.c (extract_bit_field_1): Likewise.
16789 * reload.c (get_secondary_mem): Likewise.
16790 * varasm.c (assemble_integer): Likewise.
16791 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
16794 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16796 * machmode.h (decimal_float_mode_for_size): New function.
16797 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16798 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16799 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16800 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16801 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16802 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16804 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16806 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16807 (get_builtin_sync_mode): Likewise.
16808 (expand_ifn_atomic_compare_exchange): Likewise.
16809 (expand_builtin_atomic_clear): Likewise.
16810 (expand_builtin_atomic_test_and_set): Likewise.
16811 (fold_builtin_atomic_always_lock_free): Likewise.
16812 * calls.c (compute_argument_addresses): Likewise.
16813 (emit_library_call_value_1): Likewise.
16814 (store_one_arg): Likewise.
16815 * combine.c (combine_instructions): Likewise.
16816 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16817 * config/arm/arm.c (arm_function_value): Likewise.
16818 (aapcs_allocate_return_reg): Likewise.
16819 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16820 * config/i386/i386.c (construct_container): Likewise.
16821 (ix86_gimplify_va_arg): Likewise.
16822 (ix86_expand_sse_cmp): Likewise.
16823 (emit_memmov): Likewise.
16824 (emit_memset): Likewise.
16825 (expand_small_movmem_or_setmem): Likewise.
16826 (ix86_expand_pextr): Likewise.
16827 (ix86_expand_pinsr): Likewise.
16828 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16829 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16831 * config/mips/mips.c (mips_function_value_1) Likewise.
16832 (mips_block_move_straight): Likewise.
16833 (mips_expand_ins_as_unaligned_store): Likewise.
16834 * config/powerpcspe/powerpcspe.c
16835 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16836 (rs6000_darwin64_record_arg_flush): Likewise.
16837 * config/rs6000/rs6000.c
16838 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16839 (rs6000_darwin64_record_arg_flush): Likewise.
16840 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16841 (sparc_function_value_1): Likewise.
16842 * config/spu/spu.c (adjust_operand): Likewise.
16843 (spu_emit_branch_or_set): Likewise.
16844 (arith_immediate_p): Likewise.
16845 * emit-rtl.c (gen_lowpart_common): Likewise.
16846 * expr.c (expand_expr_real_1): Likewise.
16847 * function.c (assign_parm_setup_block): Likewise.
16848 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16849 * reload1.c (alter_reg): Likewise.
16850 * stor-layout.c (mode_for_vector): Likewise.
16851 (layout_type): Likewise.
16853 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16855 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16856 (spu_convert_move): Likewise.
16857 * lower-subreg.c (resolve_simple_move): Likewise.
16859 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16862 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16863 define_insn to a define_expand.
16864 (altivec_vsum2sws_direct): New define_insn.
16865 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16867 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16869 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16872 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16877 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16878 realign stack if stack alignment needed is less than incoming
16881 2017-09-05 Marek Polacek <polacek@redhat.com>
16884 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16887 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16889 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16891 2017-09-05 Richard Biener <rguenther@suse.de>
16893 PR tree-optimization/82084
16894 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16896 2017-09-05 Richard Biener <rguenther@suse.de>
16898 PR tree-optimization/82102
16899 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16901 2017-09-05 Martin Liska <mliska@suse.cz>
16903 PR tree-optimization/82032
16904 * tree-cfg.c (generate_range_test): New function.
16905 * tree-cfg.h (generate_range_test): Declared here.
16906 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16907 (cleanup_control_expr_graph): Use it.
16908 * tree-switch-conversion.c (try_switch_expansion): Remove
16910 (emit_case_nodes): Use generate_range_test.
16912 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16915 * config/i386/i386.md (*<btsc><mode>_mask): Add
16916 TARGET_USE_BT to insn constraint.
16917 (*btr<mode>_mask): Ditto.
16919 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16921 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16922 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16924 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16927 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16928 TARGET_NEON and TARGET_IWMMXT.
16929 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16930 TARGET_NEON and TARGET_IWMMXT.
16931 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16933 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16935 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16936 (ix86_rewrite_tls_address): Ditto.
16937 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16938 (ix86_rewrite_tls_address_1): Ditto.
16939 (ix86_rewrite_tls_address): Ditto.
16940 * config/i386/predicates.md (tls_address_pattern): New predicate.
16941 * config/i386/i386.md (TLS address splitter): New splitter.
16943 2017-09-04 Richard Biener <rguenther@suse.de>
16945 PR tree-optimization/82084
16946 * fold-const.h (can_native_encode_string_p): Declare.
16947 * fold-const.c (can_native_encode_string_p): Factor out from ...
16948 (native_encode_string): ... here.
16949 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16950 vectorizing stores from constants we later cannot handle.
16952 2017-09-04 Marek Polacek <polacek@redhat.com>
16955 * doc/invoke.texi: Update -Wtautological-compare documentation.
16957 2017-09-04 Jeff Law <law@redhat.com>
16959 PR tree-optimization/64910
16960 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16961 swap the first and last operand if the last is a constant.
16963 2017-09-04 Marek Polacek <polacek@redhat.com>
16966 * convert.c (do_narrow): When sanitizing signed integer overflows,
16967 bail out for signed types.
16968 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16970 2017-09-04 Richard Biener <rguenther@suse.de>
16972 PR tree-optimization/82060
16973 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16974 Move devirtualization after stmt folding and before EH/AB/noreturn
16975 cleanup to get the stmt refs canonicalized. Use a bool instead
16976 of gimple_modified_p since that doesn't work for NOPs. Schedule
16977 NOPs generated by folding for removal.
16979 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16980 Alan Hayward <alan.hayward@arm.com>
16981 David Sherwood <david.sherwood@arm.com>
16983 * coretypes.h (pad_direction): New enum.
16984 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16985 (FUNCTION_ARG_PADDING): Likewise.
16986 * target.def (function_arg_padding): New hook.
16987 * targhooks.h (default_function_arg_padding): Declare.
16988 * targhooks.c (default_function_arg_padding): New function.
16989 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16990 (TARGET_FUNCTION_ARG_PADDING): ...this.
16991 * doc/tm.texi: Regenerate.
16992 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16993 instead of direction.
16994 (compute_argument_addresses): Likewise.
16995 (load_register_parameters): Likewise.
16996 (emit_library_call_value_1): Likewise.
16997 (store_one_arg): Use targetm.calls.function_arg_padding instead
16998 of FUNCTION_ARG_PADDING.
16999 (must_pass_in_stack_var_size_or_pad): Likewise.
17000 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17001 (emit_group_store): Likewise.
17002 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17003 instead of FUNCTION_ARG_PADDING.
17004 (emit_push_insn): Likewise, and propagate enum change throughout
17006 * function.h (direction): Delete.
17007 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17009 * function.c (assign_parm_find_stack_rtl): Likewise.
17010 (assign_parm_setup_block_p): Likewise.
17011 (assign_parm_setup_block): Likewise.
17012 (gimplify_parameters): Likewise.
17013 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17014 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17016 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17017 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17018 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17019 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17020 (aarch64_function_arg_padding): ...this new function.
17021 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17022 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17023 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17024 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17025 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17026 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17027 (arm_pad_arg_upward): Replace with...
17028 (arm_function_arg_padding): ...this new function.
17029 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17031 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17032 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17033 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17034 (ia64_hpux_function_arg_padding): Replace with...
17035 (ia64_function_arg_padding): ...this new function. Use pad_direction
17036 instead of direction. Check for TARGET_HPUX.
17037 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17038 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17039 (iq2000_function_arg_padding): New function.
17040 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17041 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17042 (mips_function_arg_padding): ...this new function.
17043 (mips_pad_reg_upward): Update accordingly.
17044 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17045 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17046 targetm.calls.function_arg_padding.
17047 (FUNCTION_ARG_PADDING): Delete.
17048 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17049 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17050 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17051 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17052 (nios2_block_reg_padding): Return pad_direction instead of direction.
17053 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17054 instead of direction.
17055 (nios2_function_arg_padding): Likewise. Make static.
17056 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17057 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17058 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17059 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17060 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17061 (pa_function_arg_padding): Make static. Return pad_direction instead
17063 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17064 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17065 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17066 instead of direction. Use targetm.calls.function_arg_padding.
17067 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17068 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17069 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17070 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17071 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17073 (function_arg_padding): Rename to...
17074 (rs6000_function_arg_padding): ...this. Make static. Return
17075 pad_direction instead of direction.
17076 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17077 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17078 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17079 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17080 instead of direction. Use targetm.calls.function_arg_padding.
17081 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17082 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17083 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17084 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17085 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17086 (function_arg_padding): Rename to...
17087 (rs6000_function_arg_padding): ...this. Make static. Return
17088 pad_direction instead of direction.
17089 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17090 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17091 * config/s390/s390.c (s390_function_arg_padding): New function.
17092 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17093 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17094 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17095 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17096 (function_arg_padding): Rename to...
17097 (sparc_function_arg_padding): ...this. Make static. Return
17098 pad_direction instead of direction.
17099 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17100 * config/spu/spu.c (spu_function_arg_padding): New function.
17101 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17102 * system.h (FUNCTION_ARG_PADDING): Poison.
17104 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17105 Alan Hayward <alan.hayward@arm.com>
17106 David Sherwood <david.sherwood@arm.com>
17108 * target.def (modes_tieable_p): New hook.
17109 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17110 (TARGET_MODES_TIEABLE_P): ...this.
17111 * doc/tm.texi.in: Regenerate.
17112 * hooks.h (hook_bool_mode_mode_true): Declare.
17113 * hooks.c (hook_bool_mode_mode_true): New function.
17114 * combine.c (subst): Use targetm.modes_tieable_p instead of
17116 * dse.c (find_shift_sequence): Likewise.
17117 * expmed.c (extract_low_bits): Likewise.
17118 * lower-subreg.c: Include target.h.
17119 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17121 * rtlanal.c (rtx_cost): Likewise.
17122 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17123 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17124 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17125 (TARGET_MODES_TIEABLE_P): Redefine.
17126 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17127 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17128 (TARGET_MODES_TIEABLE_P): Redefine.
17129 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17130 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17131 (arc_modes_tieable_p): New function.
17132 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17133 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17134 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17135 (arm_modes_tieable_p): Make static.
17136 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17137 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17138 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17139 (TARGET_MODES_TIEABLE_P): Redefine.
17140 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17141 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17142 (TARGET_MODES_TIEABLE_P): Redefine.
17143 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17144 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17145 (cr16_modes_tieable_p): New function.
17146 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17147 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17148 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17149 (TRULY_NOOP_TRUNCATION): Update comment.
17150 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17151 (TRULY_NOOP_TRUNCATION): Update comment.
17152 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17153 (frv_modes_tieable_p): New function.
17154 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17155 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17156 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17157 (TARGET_MODES_TIEABLE_P): Redefine.
17158 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17159 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17160 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17161 (TARGET_MODES_TIEABLE_P): Redefine.
17162 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17163 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17164 (ia64_modes_tieable_p): New function.
17165 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17166 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17167 (iq2000_modes_tieable_p): New function.
17168 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17169 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17170 (lm32_modes_tieable_p): New function.
17171 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17172 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17173 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17174 (TARGET_MODES_TIEABLE_P): Redefine.
17175 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17176 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17177 (m32r_modes_tieable_p): New function.
17178 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17179 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17180 (m68k_modes_tieable_p): New function.
17181 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17182 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17183 (mcore_modes_tieable_p): New function.
17184 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17185 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17187 (TARGET_MODES_TIEABLE_P): Redefine.
17188 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17189 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17190 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17191 (TARGET_MODES_TIEABLE_P): Redefine.
17192 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17193 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17194 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17195 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17196 (mn10300_modes_tieable_p): ...this and make static.
17197 (TARGET_MODES_TIEABLE_P): Redefine.
17198 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17199 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17200 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17201 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17202 (msp430_modes_tieable_p): Make static.
17203 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17204 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17205 (TARGET_MODES_TIEABLE_P): Redefine.
17206 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17207 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17208 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17209 (TARGET_MODES_TIEABLE_P): Redefine.
17210 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17211 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17212 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17213 (TARGET_MODES_TIEABLE_P): Redefine.
17214 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17215 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17216 (pdp11_modes_tieable_p): New function.
17217 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17218 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17219 (rs6000_modes_tieable_p): New function.
17220 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17221 * config/powerpcspe/powerpcspe.md: Update comment.
17222 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17223 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17224 (TARGET_MODES_TIEABLE_P): Redefine.
17225 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17226 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17227 (rl78_modes_tieable_p): New function.
17228 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17229 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17230 (rs6000_modes_tieable_p): New function.
17231 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17232 * config/rs6000/rs6000.md: Update comment.
17233 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17234 * config/rx/rx.c (rx_modes_tieable_p): New function.
17235 (TARGET_MODES_TIEABLE_P): Redefine.
17236 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17237 * config/s390/s390.c (s390_modes_tieable_p): New function.
17238 (TARGET_MODES_TIEABLE_P): Redefine.
17239 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17240 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17241 (sh_modes_tieable_p): New function.
17242 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17243 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17244 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17245 (sparc_modes_tieable_p): Make static.
17246 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17247 * config/spu/spu.c (spu_modes_tieable_p): New function.
17248 (TARGET_MODES_TIEABLE_P): Redefine.
17249 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17250 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17251 (TARGET_MODES_TIEABLE_P): Redefine.
17252 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17253 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17254 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17255 * config/v850/v850.c (v850_modes_tieable_p): New function.
17256 (TARGET_MODES_TIEABLE_P): Redefine.
17257 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17258 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17259 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17260 (visium_modes_tieable_p): New function.
17261 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17262 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17263 (xtensa_modes_tieable_p): New function.
17264 * system.h (MODES_TIEABLE_P): Poison.
17266 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17267 Alan Hayward <alan.hayward@arm.com>
17268 David Sherwood <david.sherwood@arm.com>
17270 * target.def (hard_regno_mode_ok): New hook.
17271 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17272 (TARGET_HARD_REGNO_MODE_OK): ...this.
17273 * doc/tm.texi.in: Regenerate.
17274 * hooks.h (hook_bool_uint_mode_true): Declare.
17275 * hooks.c (hook_bool_uint_mode_true): New function.
17276 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17277 HARD_REGNO_MODE_OK.
17278 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17279 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17280 instead of HARD_REGNO_MODE_OK.
17281 * caller-save.c: Include target.h.
17282 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17283 HARD_REGNO_MODE_OK.
17284 * combine.c (can_combine_p): Likewise.
17285 (combinable_i3pat): Likewise.
17286 (can_change_dest_mode): Likewise.
17287 * expr.c (init_expr_target): Likewise.
17288 (convert_move): Likewise.
17289 (convert_modes): Likewise.
17290 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17291 (setup_prohibited_mode_move_regs): Likewise.
17292 * ira.h (target_ira): Likewise.
17293 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17294 * lra-constraints.c (process_alt_operands): Likewise.
17295 (split_reg): Likewise.
17296 * recog.c (peep2_find_free_register): Likewise.
17297 * ree.c (combine_reaching_defs): Likewise.
17298 * regcprop.c (maybe_mode_change): Likewise.
17299 * reginfo.c (init_reg_sets_1): Likewise.
17300 (choose_hard_reg_mode): Likewise.
17301 (simplifiable_subregs): Likewise.
17302 * regrename.c (check_new_reg_p): Likewise.
17303 * reload.c (find_valid_class): Likewise.
17304 (find_valid_class_1): Likewise.
17305 (reload_inner_reg_of_subreg): Likewise.
17306 (push_reload): Likewise.
17307 (combine_reloads): Likewise.
17308 (find_dummy_reload): Likewise.
17309 (find_reloads): Likewise.
17310 * reload1.c (find_reg): Likewise.
17311 (set_reload_reg): Likewise.
17312 (allocate_reload_reg): Likewise.
17313 (choose_reload_regs): Likewise.
17314 (reload_adjust_reg_for_temp): Likewise.
17315 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17316 (simplify_subreg_regno): Likewise.
17317 * sel-sched.c (init_regs_for_mode): Likewise.
17318 * varasm.c (make_decl_rtl): Likewise.
17319 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17320 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17321 HARD_REGNO_MODE_OK.
17322 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17323 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17324 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17325 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17326 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17327 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17328 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17329 (arc_mode_class): Delete.
17330 (HARD_REGNO_MODE_OK): Delete.
17331 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17332 (arc_hard_regno_mode_ok): Rename old array to...
17333 (arc_hard_regno_mode_ok_modes): ...this.
17334 (arc_conditional_register_usage): Update accordingly.
17335 (arc_mode_class): Make static.
17336 (arc_hard_regno_mode_ok): New function.
17337 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17338 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17339 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17340 (arm_hard_regno_mode_ok): Make static.
17341 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17342 HARD_REGNO_MODE_OK.
17343 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17344 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17345 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17347 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17348 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17349 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17350 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17351 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17352 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17353 * config/bfin/predicates.md (valid_reg_operand): Use
17354 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17355 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17356 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17357 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17358 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17359 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17360 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17361 (cr16_hard_regno_mode_ok): Make static and return a bool.
17362 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17363 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17364 (cris_hard_regno_mode_ok): New function.
17365 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17366 (epiphany_mode_class): Delete.
17367 (HARD_REGNO_MODE_OK): Delete.
17368 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17369 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17370 (hard_regno_mode_ok): Rename to...
17371 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17372 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17373 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17374 HARD_REGNO_MODE_OK.
17375 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17376 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17377 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17378 (frv_hard_regno_mode_ok): Make static and return a bool.
17379 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17380 HARD_REGNO_MODE_OK.
17381 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17382 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17383 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17384 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17386 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17387 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17388 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17389 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17391 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17392 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17393 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17394 (ia64_hard_regno_mode_ok): New function.
17395 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17396 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17397 (iq2000_hard_regno_mode_ok): New function.
17398 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17399 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17400 (lm32_hard_regno_mode_ok): New function.
17401 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17402 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17403 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17404 instead of HARD_REGNO_MODE_OK.
17405 (m32c_hard_regno_ok): Rename to...
17406 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17407 (m32c_cannot_change_mode_class): Update accordingly.
17408 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17409 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17410 (m32r_mode_class): Delete.
17411 (HARD_REGNO_MODE_OK): Delete.
17412 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17413 (m32r_hard_regno_mode_ok): Rename to...
17414 (m32r_hard_regno_modes): ...this.
17415 (m32r_mode_class): Make static.
17416 (m32r_hard_regno_mode_ok): New function.
17417 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17418 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17419 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17420 (m68k_hard_regno_mode_ok): Make static.
17421 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17422 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17423 (mcore_hard_regno_mode_ok): New function.
17424 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17425 (HARD_REGNO_MODE_OK): Delete.
17426 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17428 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17429 (microblaze_hard_regno_mode_ok): New function.
17430 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17431 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17432 (mips_hard_regno_mode_ok): Delete.
17433 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17434 (mips_hard_regno_mode_ok_p): ...this and make static.
17435 (mips_hard_regno_mode_ok_p): Rename to...
17436 (mips_hard_regno_mode_ok_uncached): ...this.
17437 (mips_hard_regno_mode_ok): New function.
17438 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17439 of HARD_REGNO_MODE_OK.
17440 (mips_option_override): Update after above name changes.
17441 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17442 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17443 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17444 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17445 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17446 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17447 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17448 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17449 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17450 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17451 (msp430_hard_regno_mode_ok): Make static and return a bool.
17452 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17453 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17454 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17456 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17457 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17458 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17459 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17460 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17461 (PA_HARD_REGNO_MODE_OK): ...this
17462 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17463 (PA_HARD_REGNO_MODE_OK): ...this.
17464 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17465 (pa_hard_regno_mode_ok): New function.
17466 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17467 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17468 (pdp11_hard_regno_mode_ok): New function.
17469 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17470 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17472 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17474 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17475 (rs6000_hard_regno_mode_ok): Rename to...
17476 (rs6000_hard_regno_mode_ok_uncached): ...this.
17477 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17478 (rs6000_hard_regno_mode_ok): New function.
17479 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17480 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17481 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17482 (riscv_hard_regno_mode_ok): ...this and make static.
17483 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17484 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17485 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17486 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17487 (rl78_hard_regno_mode_ok): Make static and return bool.
17488 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17489 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17491 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17492 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17493 (rs6000_hard_regno_mode_ok): Rename to...
17494 (rs6000_hard_regno_mode_ok_uncached): ...this.
17495 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17496 (rs6000_hard_regno_mode_ok): New function.
17497 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17498 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17499 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17500 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17501 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17502 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17503 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17504 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17505 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17506 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17507 (sh_hard_regno_mode_ok): Make static.
17508 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17509 instead of HARD_REGNO_MODE_OK.
17510 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17511 (sparc_mode_class): Delete.
17512 (HARD_REGNO_MODE_OK): Delete.
17513 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17514 (hard_regno_mode_classes): Make static.
17515 (sparc_mode_class): Likewise.
17516 (sparc_hard_regno_mode_ok): New function.
17517 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17518 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17519 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17521 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17522 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17523 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17524 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17525 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17526 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17527 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17528 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17529 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17530 (visium_hard_regno_mode_ok): New function.
17531 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17532 instead of HARD_REGNO_MODE_OK.
17533 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17534 (HARD_REGNO_MODE_OK): Delete.
17535 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17536 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17537 (xtensa_option_override): Update accordingly.
17538 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17539 (xtensa_hard_regno_mode_ok): New function.
17540 * system.h (HARD_REGNO_MODE_OK): Poison.
17542 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17543 Alan Hayward <alan.hayward@arm.com>
17544 David Sherwood <david.sherwood@arm.com>
17546 * target.def (hard_regno_call_part_clobbered): New hook.
17547 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17548 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17549 * doc/tm.texi: Regenerate.
17550 * hooks.h (hook_bool_uint_mode_false): Declare.
17551 * hooks.c (hook_bool_uint_mode_false): New function.
17552 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17553 * cselib.c (cselib_process_insn): Use
17554 targetm.hard_regno_call_part_clobbered instead of
17555 HARD_REGNO_CALL_PART_CLOBBERED.
17556 * ira-conflicts.c (ira_build_conflicts): Likewise.
17557 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17558 * lra-constraints.c (need_for_call_save_p): Likewise.
17559 * lra-lives.c: Include target.h.
17560 (check_pseudos_live_through_calls): Use
17561 targetm.hard_regno_call_part_clobbered instead of
17562 HARD_REGNO_CALL_PART_CLOBBERED.
17563 * regcprop.c: Include target.h.
17564 (copyprop_hardreg_forward_1): Use
17565 targetm.hard_regno_call_part_clobbered instead of
17566 HARD_REGNO_CALL_PART_CLOBBERED.
17567 * reginfo.c (choose_hard_reg_mode): Likewise.
17568 * regrename.c (check_new_reg_p): Likewise.
17569 * reload.c (find_equiv_reg): Likewise.
17570 * reload1.c (emit_reload_insns): Likewise.
17571 * sched-deps.c (deps_analyze_insn): Likewise.
17572 * sel-sched.c (init_regs_for_mode): Likewise.
17573 (mark_unavailable_hard_regs): Likewise.
17574 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17575 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17576 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17578 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17579 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17580 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17582 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17584 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17585 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17586 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17588 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17589 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17590 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17592 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17593 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17595 * config/powerpcspe/powerpcspe.c
17596 (rs6000_hard_regno_call_part_clobbered): New function.
17597 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17598 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17599 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17601 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17602 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17603 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17605 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17606 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17607 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17609 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17610 Alan Hayward <alan.hayward@arm.com>
17611 David Sherwood <david.sherwood@arm.com>
17613 * rtl.h (subreg_memory_offset): Declare.
17614 * emit-rtl.c (subreg_memory_offset): New function.
17615 * expmed.c (store_bit_field_1): Use it.
17616 * expr.c (undefined_operand_subword_p): Likewise.
17617 * simplify-rtx.c (simplify_subreg): Likewise.
17619 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17621 PR rtl-optimization/57448
17624 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17625 using atomic_load pattern.
17626 (expand_atomic_store): Likewise.
17628 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17631 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17632 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17633 handling. Use replace_call_with_value with NULL instead of
17634 gsi_replace, unlink_stmt_vdef and release_defs.
17636 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17639 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17641 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17644 * rtl.h (emit_library_call_value_1): Declare.
17645 (emit_library_call): Replace declaration with a series of overloads.
17646 Remove the parameter count argument.
17647 (emit_library_call_value): Likewise.
17648 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17649 with an "rtx_mode_t *".
17650 (emit_library_call_value): Delete.
17651 (emit_library_call): Likewise.
17652 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17653 (asan_emit_allocas_unpoison): Likewise.
17654 * builtins.c (expand_builtin_powi): Likewise.
17655 (expand_asan_emit_allocas_unpoison): Likewise.
17656 * cfgexpand.c (expand_main_function): Likewise.
17657 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17658 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17659 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17660 * config/arm/arm.c (arm_trampoline_init): Likewise.
17661 (arm_call_tls_get_addr): Likewise.
17662 (arm_expand_divmod_libfunc): Likewise.
17663 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17664 (smulsi3_highpart): Likewise.
17665 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17666 (c6x_expand_compare): Likewise.
17667 (c6x_expand_movmem): Likewise.
17668 * config/frv/frv.c (frv_trampoline_init): Likewise.
17669 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17670 (ix86_expand_divmod_libfunc): Likewise.
17671 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17672 (ia64_expand_compare): Likewise.
17673 (ia64_profile_hook): Likewise.
17674 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17675 (nonlocal_goto): Likewise.
17676 (restore_stack_nonlocal): Likewise.
17677 * config/m32r/m32r.c (block_move_call): Likewise.
17678 (m32r_trampoline_init): Likewise.
17679 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17680 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17681 (m68k_call_m68k_read_tp): Likewise.
17682 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17683 (microblaze_expand_divide): Likewise.
17684 * config/mips/mips.h (mips_args): Likewise.
17685 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17686 (MIPS_ICACHE_SYNC): Likewise.
17687 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17688 (nios2_trampoline_init): Likewise.
17689 * config/pa/pa.c (hppa_tls_call): Likewise.
17690 (pa_trampoline_init): Likewise.
17691 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17692 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17693 (expand_strn_compare): Likewise.
17694 (rs6000_generate_compare): Likewise.
17695 (rs6000_expand_float128_convert): Likewise.
17696 (output_profile_hook): Likewise.
17697 (rs6000_trampoline_init): Likewise.
17698 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17699 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17700 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17701 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17702 (rs6000_generate_compare): Likewise.
17703 (rs6000_expand_float128_convert): Likewise.
17704 (output_profile_hook): Likewise.
17705 (rs6000_trampoline_init): Likewise.
17706 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17707 * config/sh/sh.c (sh_trampoline_init): Likewise.
17708 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17709 (sparc_emit_float_lib_cmp): Likewise.
17710 (sparc32_initialize_trampoline): Likewise.
17711 (sparc64_initialize_trampoline): Likewise.
17712 (sparc_profile_hook): Likewise.
17713 * config/spu/spu.c (ea_load_store): Likewise.
17714 * config/spu/spu.md (floatunssidf2): Likewise.
17715 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
17716 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
17717 * config/visium/visium.c (expand_block_move_4): Likewise.
17718 (expand_block_move_2): Likewise.
17719 (expand_block_move_1): Likewise.
17720 (expand_block_set_4): Likewise.
17721 (expand_block_set_2): Likewise.
17722 (expand_block_set_1): Likewise.
17723 (visium_trampoline_init): Likewise.
17724 (visium_profile_hook): Likewise.
17725 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
17726 (xtensa_setup_frame_addresses): Likewise.
17727 (xtensa_trampoline_init): Likewise.
17728 * except.c (sjlj_emit_function_enter): Likewise.
17729 (sjlj_emit_function_exit): Likewise.
17730 * explow.c (allocate_dynamic_stack_space): Likewise.
17731 (probe_stack_range): Likewise.
17732 * expr.c (convert_mode_scalar): Likewise.
17733 * optabs.c (expand_binop): Likewise.
17734 (expand_twoval_binop_libfunc): Likewise.
17735 (expand_unop): Likewise.
17736 (prepare_cmp_insn): Likewise.
17737 (prepare_float_lib_cmp): Likewise.
17738 (expand_float): Likewise.
17739 (expand_fix): Likewise.
17740 (expand_fixed_convert): Likewise.
17741 (maybe_emit_sync_lock_test_and_set): Likewise.
17742 (expand_atomic_compare_and_swap): Likewise.
17743 (expand_mem_thread_fence): Likewise.
17744 (expand_atomic_fetch_op): Likewise.
17746 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17748 * doc/generic.texi (OpenACC): Adjust URL.
17749 * doc/invoke.texi (C Dialect Options): Ditto.
17751 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17753 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17754 predicate for operand 1. Add (m,<S>) constraint.
17755 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17756 Prevent memory operand 1 with register operand 2.
17758 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17760 PR rtl-optimization/82024
17761 * combine.c (try_combine): If the combination result is a PARALLEL,
17762 and we only need to retain the SET in there that would be placed
17763 at I2, check that we can place that at I3 instead, before doing so.
17765 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17768 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
17770 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
17771 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17774 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17775 Jeff Law <law@redhat.com>
17777 * varasm.c (bss_initializer_p): Do not put constants into .bss
17778 (categorize_decl_for_section): Handle bss_initializer_p returning
17779 false when DECL_INITIAL is NULL.
17781 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17784 * config/s390/s390.c (s390_can_inline_p): New function.
17786 2017-09-01 Jeff Law <law@redhat.com>
17788 PR tree-optimization/82052
17789 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17790 Always initialize the returned slot after a hash table miss
17791 when INSERT is true.
17793 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
17795 * config/s390/s390.md (mem_signal_fence): Remove.
17796 * doc/md.texi (mem_signal_fence): Remove.
17797 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17799 * target-insns.def (mem_signal_fence): Remove.
17801 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17804 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17807 * asan.c (create_odr_indicator): Strip name encoding from assembler
17808 name before appending it after __odr_asan_.
17810 2017-09-01 Martin Liska <mliska@suse.cz>
17812 PR tree-optimization/82059
17813 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17814 frequency only when an edge is redirected.
17816 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17818 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17819 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17820 (arc_conditional_register_usage): Remove ARC600 lp_count
17822 (arc_file_start): Emit Tag_ARC_CPU_variation.
17823 (arc_can_use_doloop_p): New conditions to use ZOLs.
17824 (hwloop_fail): New function.
17825 (hwloop_optimize): Likewise.
17826 (hwloop_pattern_reg): Likewise.
17827 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17828 (arc_reorg_loops): New function, calls reorg_loops.
17829 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17830 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17832 (arc_loop_hazard): Remove function, functionality moved into
17834 (arc_hazard): Remove arc_loop_hazard call.
17835 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17836 into hwloop_optimize.
17837 (arc_label_align): Remove ZOL handling.
17838 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17839 * config/arc/arc.md (doloop_begin): Remove pattern.
17840 (doloop_begin_i): Likewise.
17841 (doloop_end_i): Likewise.
17842 (doloop_fallback): Likewise.
17843 (doloop_fallback_m): Likewise.
17844 (doloop_end): Reimplement expand.
17845 (arc_lp): New pattern for LP instruction.
17846 (loop_end): New pattern.
17847 (loop_fail): Likewise.
17848 (decrement_and_branch_until_zero): Likewise.
17849 * config/arc/arc.opt (mlpc-width): New option.
17850 * doc/invoke.texi (mlpc-width): Document option.
17852 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17854 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17855 (arc_ccfsm_advance): Fix checking for delay slots.
17856 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17858 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17860 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17862 (movhi_insn): Update store instruction constraint which are saving
17863 6-bit short immediates.
17864 (movsi_insn): Consider also short scaled load operations.
17865 (zero_extendhisi2_i): Use Usd constraint instead of T.
17866 (extendhisi2_i): Add q constraint.
17867 (arc_clzsi2): Add type and length attributes.
17868 (arc_ctzsi2): Likewise.
17869 * config/arc/constraints.md (Usc): Update constraint, the
17870 assembler can parse two relocations for a single instruction.
17872 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17874 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17875 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17877 2017-08-31 Olivier Hainque <hainque@adacore.com>
17879 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17880 match as powerpc-wrs-vxworks*.
17882 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17884 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17885 register constraint for by-element operand.
17886 (aarch64_mls_elt_merge<mode>): Likewise.
17888 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17890 * config/arc/arc.c (arc_can_follow_jump): Check for short
17893 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17895 * config.gcc: Use g.opt for arc.
17896 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17897 functionality moved to ...
17898 (legitimate_scaled_address_p): New function, ...here.
17899 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17900 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17901 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17903 (arc_override_options): Handle G option.
17904 (arc_output_pic_addr_const): Correct function definition.
17905 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17906 (arc_decl_anon_ns_mem_p): Delete.
17907 (arc_in_small_data_p): Overhaul this function to take into
17908 consideration the value given via G option.
17909 (arc_rewrite_small_data_1): Renamed and corrected old
17910 arc_rewrite_small_data function.
17911 (arc_rewrite_small_data): New function.
17912 (small_data_pattern): Don't use pic_offset_table_rtx.
17913 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17914 * config/arc/simdext.md (movmisalignv2hi): Use
17915 prepare_move_operands function.
17917 (movmisalign*): Likewise.
17918 * doc/invoke.texi (ARC options): Document -G option.
17920 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17922 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17924 * config/arc/arc.c (arc_print_operand): Output scalled address for
17925 sdata whenever is possible.
17926 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17927 load/stores are available.
17928 (compact_sda_memory_operand): Check for the alignment required by
17929 code density instructions.
17930 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17932 * config/arc/constraints.md (Usd): Update constraint.
17933 (Us0): New constraint.
17934 (Usc): Update constraint.
17936 2017-08-31 Richard Biener <rguenther@suse.de>
17938 PR middle-end/82054
17939 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17940 function only once.
17942 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17944 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17945 Resize type_signature.
17947 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17948 Alan Hayward <alan.hayward@arm.com>
17949 David Sherwood <david.sherwood@arm.com>
17951 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17952 subregs whose inner modes can be stored in GPRs.
17953 (aarch64_classify_index): Likewise.
17955 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17956 Alan Hayward <alan.hayward@arm.com>
17957 David Sherwood <david.sherwood@arm.com>
17959 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17960 (V_INT_EQUIV): ...this.
17961 (v_cmp_result): Rename to...
17962 (v_int_equiv): ...this.
17963 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17964 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17965 (copysign<mode>3): Likewise.
17966 (aarch64_simd_bsl<mode>_internal): Likewise.
17967 (aarch64_simd_bsl<mode>): Likewise.
17968 (vec_cmp<mode><mode>): Likewise.
17969 (vcond<mode><mode>): Likewise.
17970 (vcond<v_cmp_mixed><mode>): Likewise.
17971 (vcondu<mode><v_cmp_mixed>): Likewise.
17972 (aarch64_cm<optab><mode>): Likewise.
17973 (aarch64_cmtst<mode>): Likewise.
17974 (aarch64_fac<optab><mode>): Likewise.
17975 (vec_perm_const<mode>): Likewise.
17976 (vcond_mask_<mode><v_cmp_result>): Rename to...
17977 (vcond_mask_<mode><v_int_equiv>): ...this.
17978 (vec_cmp<mode><v_cmp_result>): Rename to...
17979 (vec_cmp<mode><v_int_equiv>): ...this.
17981 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17982 Alan Hayward <alan.hayward@arm.com>
17983 David Sherwood <david.sherwood@arm.com>
17985 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17987 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17988 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17989 (UNSPEC_LD4_DREG): New unspecs.
17990 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17991 (aarch64_ld2<mode>_dreg_be): Replace with...
17992 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17994 (aarch64_ld3<mode>_dreg_le)
17995 (aarch64_ld3<mode>_dreg_be): Replace with...
17996 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17998 (aarch64_ld4<mode>_dreg_le)
17999 (aarch64_ld4<mode>_dreg_be): Replace with...
18000 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18003 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18005 PR tree-optimization/81987
18006 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18007 insert an initializer in a location not dominated by the stride
18010 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18012 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18013 on the entire header of the finally block in the fallthru case.
18015 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18017 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18019 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18021 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18022 rs6000_emit_move_from_cr and call renamed function.
18023 (rs6000_emit_prologue): Call renamed functions.
18024 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18025 movesi_from_cr, remove volatile CRs.
18027 2017-08-30 Jon Beniston <jon@beniston.com>
18028 Richard Biener <rguenther@suse.de>
18030 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18031 of VECTOR_MODE_P check.
18032 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18033 element vector types.
18035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18037 * df.h (df_read_modify_subreg_p): Remove in favor of...
18038 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18039 const_rtx instead of an rtx.
18040 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18041 * df-problems.c (df_word_lr_mark_ref): Likewise.
18042 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18043 (mark_pseudo_reg_dead): Likewise.
18044 (mark_ref_dead): Likewise.
18045 * reginfo.c (init_subregs_of_mode): Likewise.
18046 * sched-deps.c (sched_analyze_1): Likewise.
18047 * df-scan.c (df_def_record_1): Likewise.
18048 (df_uses_record): Likewise.
18049 (df_read_modify_subreg_p): Remove in favor of...
18050 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18051 const_rtx instead of an rtx.
18053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18054 Alan Hayward <alan.hayward@arm.com>
18055 David Sherwood <david.sherwood@arm.com>
18057 * rtl.h (partial_subreg_p): New function.
18058 * caller-save.c (save_call_clobbered_regs): Use it.
18059 * calls.c (expand_call): Likewise.
18060 * combine.c (combinable_i3pat): Likewise.
18061 (simplify_set): Likewise.
18062 (make_extraction): Likewise.
18063 (make_compound_operation_int): Likewise.
18064 (gen_lowpart_or_truncate): Likewise.
18065 (force_to_mode): Likewise.
18066 (make_field_assignment): Likewise.
18067 (reg_truncated_to_mode): Likewise.
18068 (record_truncated_value): Likewise.
18069 (move_deaths): Likewise.
18070 * cse.c (record_jump_cond): Likewise.
18071 (cse_insn): Likewise.
18072 * cselib.c (cselib_lookup_1): Likewise.
18073 * expmed.c (extract_bit_field_using_extv): Likewise.
18074 * function.c (assign_parm_setup_reg): Likewise.
18075 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18076 * ira-build.c (create_insn_allocnos): Likewise.
18077 * lra-coalesce.c (merge_pseudos): Likewise.
18078 * lra-constraints.c (match_reload): Likewise.
18079 (simplify_operand_subreg): Likewise.
18080 (curr_insn_transform): Likewise.
18081 * lra-lives.c (process_bb_lives): Likewise.
18082 * lra.c (new_insn_reg): Likewise.
18083 (lra_substitute_pseudo): Likewise.
18084 * regcprop.c (mode_change_ok): Likewise.
18085 (maybe_mode_change): Likewise.
18086 (copyprop_hardreg_forward_1): Likewise.
18087 * reload.c (push_reload): Likewise.
18088 (find_reloads): Likewise.
18089 (find_reloads_subreg_address): Likewise.
18090 * reload1.c (alter_reg): Likewise.
18091 (eliminate_regs_1): Likewise.
18092 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18094 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18096 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18097 back to if statements, including unpack.
18099 2017-08-30 Martin Liska <mliska@suse.cz>
18101 PR inline-asm/82001
18102 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18104 (func_checker::compare_asm_inputs_outputs): ... this function.
18105 (func_checker::compare_gimple_asm): Use the function to compare
18106 also ASM constrains.
18107 * ipa-icf-gimple.h: Rename the function.
18109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18110 Alan Hayward <alan.hayward@arm.com>
18111 David Sherwood <david.sherwood@arm.com>
18113 * coretypes.h (complex_mode): New type.
18114 * gdbhooks.py (build_pretty_printer): Handle it.
18115 * machmode.h (complex_mode): New class.
18116 (complex_mode::includes_p): New function.
18117 (is_complex_int_mode): Likewise.
18118 (is_complex_float_mode): Likewise.
18119 * genmodes.c (get_mode_class): Handle complex mode classes.
18120 * function.c (expand_function_end): Use is_complex_int_mode.
18122 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18123 Alan Hayward <alan.hayward@arm.com>
18124 David Sherwood <david.sherwood@arm.com>
18126 * coretypes.h (scalar_mode_pod): New typedef.
18127 * gdbhooks.py (build_pretty_printer): Handle it.
18128 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18129 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18130 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18131 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18132 as_a <scalar_mode>.
18134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18135 Alan Hayward <alan.hayward@arm.com>
18136 David Sherwood <david.sherwood@arm.com>
18138 * machmode.h (mode_for_vector): Take a scalar_mode instead
18140 * stor-layout.c (mode_for_vector): Likewise.
18141 * explow.c (promote_mode): Use as_a <scalar_mode>.
18142 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18145 Alan Hayward <alan.hayward@arm.com>
18146 David Sherwood <david.sherwood@arm.com>
18148 * target.def (preferred_simd_mode): Take a scalar_mode
18149 instead of a machine_mode.
18150 * targhooks.h (default_preferred_simd_mode): Likewise.
18151 * targhooks.c (default_preferred_simd_mode): Likewise.
18152 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18153 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18154 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18155 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18156 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18157 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18158 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18159 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18161 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18162 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18163 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18164 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18165 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18166 * doc/tm.texi: Regenerate.
18167 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18171 Alan Hayward <alan.hayward@arm.com>
18172 David Sherwood <david.sherwood@arm.com>
18174 * target.def (scalar_mode_supported_p): Take a scalar_mode
18175 instead of a machine_mode.
18176 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18177 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18178 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18179 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18180 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18181 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18182 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18183 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18184 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18185 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18186 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18187 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18188 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18189 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18191 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18192 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18193 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18194 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18195 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18197 * doc/tm.texi: Regenerate.
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 (opt_scalar_mode): New typedef.
18204 * gdbhooks.py (build_pretty_printers): Handle it.
18205 * machmode.h (mode_iterator::get_2xwider): Add overload for
18207 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18209 * expr.c (convert_mode_scalar): Likewise.
18210 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18211 * optabs.c (expand_float): Likewise.
18212 (expand_fix): Likewise.
18213 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18215 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18216 Alan Hayward <alan.hayward@arm.com>
18217 David Sherwood <david.sherwood@arm.com>
18219 * optabs.c (expand_float): Explicitly check for scalars before
18220 using a branching expansion.
18221 (expand_fix): Likewise.
18223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18224 Alan Hayward <alan.hayward@arm.com>
18225 David Sherwood <david.sherwood@arm.com>
18227 * expr.c (convert_mode): Split scalar handling out into...
18228 (convert_mode_scalar): ...this new function. Treat the modes
18231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18232 Alan Hayward <alan.hayward@arm.com>
18233 David Sherwood <david.sherwood@arm.com>
18235 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18237 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18240 Alan Hayward <alan.hayward@arm.com>
18241 David Sherwood <david.sherwood@arm.com>
18243 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18244 rather than a machine_mode.
18245 (fixed_from_string): Likewise.
18246 (fixed_convert): Likewise.
18247 (fixed_convert_from_int): Likewise.
18248 (fixed_convert_from_real): Likewise.
18249 (real_convert_from_fixed): Likewise.
18250 * fixed-value.c (fixed_from_double_int): Likewise.
18251 (fixed_from_string): Likewise.
18252 (fixed_convert): Likewise.
18253 (fixed_convert_from_int): Likewise.
18254 (fixed_convert_from_real): Likewise.
18255 (real_convert_from_fixed): Likewise.
18256 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18259 Alan Hayward <alan.hayward@arm.com>
18260 David Sherwood <david.sherwood@arm.com>
18262 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18263 of separate mode class checks. Do not allow vector modes here.
18264 (immed_wide_int_const): Use as_a <scalar_mode>.
18265 * explow.c (trunc_int_for_mode): Likewise.
18266 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18267 (wi::shwi): Likewise.
18268 (wi::min_value): Likewise.
18269 (wi::max_value): Likewise.
18270 * dwarf2out.c (loc_descriptor): Likewise.
18271 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18272 for CONST_WIDE_INT.
18274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18275 Alan Hayward <alan.hayward@arm.com>
18276 David Sherwood <david.sherwood@arm.com>
18278 * tree.h (SCALAR_TYPE_MODE): New macro.
18279 * expr.c (expand_expr_addr_expr_1): Use it.
18280 (expand_expr_real_2): Likewise.
18281 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18282 (fold_convert_const_fixed_from_int): Likewise.
18283 (fold_convert_const_fixed_from_real): Likewise.
18284 (native_encode_fixed): Likewise
18285 (native_encode_complex): Likewise
18286 (native_encode_vector): Likewise.
18287 (native_interpret_fixed): Likewise.
18288 (native_interpret_real): Likewise.
18289 (native_interpret_complex): Likewise.
18290 (native_interpret_vector): Likewise.
18291 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18292 (simd_clone_adjust_argument_types): Likewise.
18293 (simd_clone_init_simd_arrays): Likewise.
18294 (simd_clone_adjust): Likewise.
18295 * stor-layout.c (layout_type): Likewise.
18296 * tree.c (build_minus_one_cst): Likewise.
18297 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18298 * tree-inline.c (estimate_move_cost): Likewise.
18299 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18300 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18301 (vectorizable_reduction): Likewise.
18302 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18303 (vect_recog_mixed_size_cond_pattern): Likewise.
18304 (check_bool_pattern): Likewise.
18305 (adjust_bool_pattern): Likewise.
18306 (search_type_for_mask_1): Likewise.
18307 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18308 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18309 (vectorizable_load): Likewise.
18310 (vectorizable_store): Likewise.
18311 * ubsan.c (ubsan_encode_value): Likewise.
18312 * varasm.c (output_constant): Likewise.
18314 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18315 Alan Hayward <alan.hayward@arm.com>
18316 David Sherwood <david.sherwood@arm.com>
18318 * coretypes.h (scalar_mode): New class.
18319 * machmode.h (scalar_mode): Likewise.
18320 (scalar_mode::includes_p): New function.
18321 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18322 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18323 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18324 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18325 * expmed.c (store_bit_field_1): Likewise.
18326 (extract_bit_field_1): Likewise.
18327 * expr.c (write_complex_part): Likewise.
18328 (read_complex_part): Likewise.
18329 (emit_move_complex_push): Likewise.
18330 (expand_expr_real_2): Likewise.
18331 * function.c (assign_parm_setup_reg): Likewise.
18332 (assign_parms_unsplit_complex): Likewise.
18333 * optabs.c (expand_binop): Likewise.
18334 * rtlanal.c (subreg_get_info): Likewise.
18335 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18336 * varasm.c (output_constant_pool_2): Likewise.
18338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18339 Alan Hayward <alan.hayward@arm.com>
18340 David Sherwood <david.sherwood@arm.com>
18342 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18344 (expmed_mult_highpart_optab): Likewise.
18345 (expmed_mult_highpart): Likewise.
18347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18348 Alan Hayward <alan.hayward@arm.com>
18349 David Sherwood <david.sherwood@arm.com>
18351 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18352 instead of a machine_mode.
18353 (builtin_memset_read_str): Likewise.
18354 * builtins.c (c_readstr): Likewise.
18355 (builtin_memcpy_read_str): Likewise.
18356 (builtin_strncpy_read_str): Likewise.
18357 (builtin_memset_read_str): Likewise.
18358 (builtin_memset_gen_str): Likewise.
18359 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18360 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18361 instead of a machine_mode.
18362 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18364 (make_extraction): Likewise.
18365 (try_widen_shift_mode): Take and return scalar_int_modes instead
18367 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18368 a scalar_int_mode instead of a machine_mode.
18369 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18370 (avr_addr_space_pointer_mode): Likewise.
18371 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18372 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18373 (msp430_unwind_word_mode): Likewise.
18374 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18375 (spu_addr_space_pointer_mode): Likewise.
18376 (spu_addr_space_address_mode): Likewise.
18377 (spu_libgcc_cmp_return_mode): Likewise.
18378 (spu_libgcc_shift_count_mode): Likewise.
18379 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18380 (rl78_addr_space_pointer_mode): Likewise.
18381 (fl78_unwind_word_mode): Likewise.
18382 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18384 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18385 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18386 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18387 (mips_valid_pointer_mode): Likewise.
18388 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18389 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18390 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18392 (ft32_addr_space_address_mode): Likewise.
18393 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18394 scalar_int_mode instead of a machine_mode.
18395 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18397 (m32c_addr_space_address_mode): Likewise.
18398 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18399 (rs6000_eh_return_filter_mode): Likewise.
18400 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18401 (rs6000_eh_return_filter_mode): Likewise.
18402 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18403 (s390_libgcc_shift_count_mode): Likewise.
18404 (s390_unwind_word_mode): Likewise.
18405 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18407 * target.def (mode_rep_extended): Likewise.
18408 (valid_pointer_mode): Likewise.
18409 (addr_space.valid_pointer_mode): Likewise.
18410 (eh_return_filter_mode): Return a scalar_int_mode rather than
18412 (libgcc_cmp_return_mode): Likewise.
18413 (libgcc_shift_count_mode): Likewise.
18414 (unwind_word_mode): Likewise.
18415 (addr_space.pointer_mode): Likewise.
18416 (addr_space.address_mode): Likewise.
18417 * doc/tm.texi: Regenerate.
18418 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18420 (do_jump): Use scalar_int_mode for local variables.
18421 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18422 rather than a machine_mode.
18423 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18424 (scompare_loc_descriptor_wide): Likewise.
18425 (scompare_loc_descriptor_narrow): Likewise.
18426 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18428 * except.c (sjlj_emit_dispatch_table): Likewise.
18429 (expand_builtin_eh_copy_values): Likewise.
18430 * explow.c (convert_memory_address_addr_space_1): Likewise.
18431 Take a scalar_int_mode rather than a machine_mode.
18432 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18433 than a machine_mode.
18434 (memory_address_addr_space): Use scalar_int_mode for local variables.
18435 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18436 rather than a machine_mode.
18437 * expmed.c (mask_rtx): Likewise.
18438 (init_expmed_one_conv): Likewise.
18439 (expand_mult_highpart_adjust): Likewise.
18440 (extract_high_half): Likewise.
18441 (expmed_mult_highpart_optab): Likewise.
18442 (expmed_mult_highpart): Likewise.
18443 (expand_smod_pow2): Likewise.
18444 (expand_sdiv_pow2): Likewise.
18445 (emit_store_flag_int): Likewise.
18446 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18448 (extract_low_bits): Likewise.
18449 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18451 * expr.c (pieces_addr::adjust): Likewise.
18452 (can_store_by_pieces): Likewise.
18453 (store_by_pieces): Likewise.
18454 (clear_by_pieces_1): Likewise.
18455 (expand_expr_addr_expr_1): Likewise.
18456 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18457 (expand_expr_real_1): Likewise.
18458 (try_casesi): Likewise.
18459 * final.c (shorten_branches): Likewise.
18460 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18461 type of "mode" to machine_mode.
18462 * internal-fn.c (expand_arith_overflow_result_store): Take a
18463 scalar_int_mode rather than a machine_mode.
18464 (expand_mul_overflow): Use scalar_int_mode for local variables.
18465 * loop-doloop.c (doloop_modify): Likewise.
18466 (doloop_optimize): Likewise.
18467 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18468 than a machine_mode.
18469 (expand_doubleword_shift_condmove): Likewise.
18470 (expand_doubleword_shift): Likewise.
18471 (expand_doubleword_clz): Likewise.
18472 (expand_doubleword_popcount): Likewise.
18473 (expand_doubleword_parity): Likewise.
18474 (expand_absneg_bit): Use scalar_int_mode for local variables.
18475 (prepare_float_lib_cmp): Likewise.
18476 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18477 rather than a machine_mode.
18478 (convert_memory_address_addr_space): Likewise.
18479 (get_mode_bounds): Likewise.
18480 (get_address_mode): Return a scalar_int_mode rather than a
18482 * rtlanal.c (get_address_mode): Likewise.
18483 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18484 than a machine_mode.
18485 * targhooks.c (default_mode_rep_extended): Likewise.
18486 (default_valid_pointer_mode): Likewise.
18487 (default_addr_space_valid_pointer_mode): Likewise.
18488 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18489 than a machine_mode.
18490 (default_libgcc_cmp_return_mode): Likewise.
18491 (default_libgcc_shift_count_mode): Likewise.
18492 (default_unwind_word_mode): Likewise.
18493 (default_addr_space_pointer_mode): Likewise.
18494 (default_addr_space_address_mode): Likewise.
18495 * targhooks.h (default_eh_return_filter_mode): Likewise.
18496 (default_libgcc_cmp_return_mode): Likewise.
18497 (default_libgcc_shift_count_mode): Likewise.
18498 (default_unwind_word_mode): Likewise.
18499 (default_addr_space_pointer_mode): Likewise.
18500 (default_addr_space_address_mode): Likewise.
18501 (default_mode_rep_extended): Take a scalar_int_mode rather than
18503 (default_valid_pointer_mode): Likewise.
18504 (default_addr_space_valid_pointer_mode): Likewise.
18505 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18507 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18508 rather than a machine_mode.
18509 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18510 for local variables.
18511 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18512 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18513 than a machine_mode.
18515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18516 Alan Hayward <alan.hayward@arm.com>
18517 David Sherwood <david.sherwood@arm.com>
18519 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18520 the mode argument to scalar_int_mode.
18521 (do_jump_by_parts_zero_rtx): Likewise.
18522 (do_jump_by_parts_equality_rtx): Likewise.
18523 (do_jump_by_parts_greater): Take a mode argument.
18524 (do_jump_by_parts_equality): Likewise.
18525 (do_jump_1): Update calls accordingly.
18527 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18528 Alan Hayward <alan.hayward@arm.com>
18529 David Sherwood <david.sherwood@arm.com>
18531 * is-a.h (safe_dyn_cast): New function.
18532 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18533 (jump_table_for_label): Likewise.
18534 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18535 instead of an rtx_insn *.
18536 (shorten_branches): Use dyn_cast instead of LABEL_P and
18537 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18538 rtx_jump_table_data::get_data_mode.
18539 (final_scan_insn): Likewise.
18541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18542 Alan Hayward <alan.hayward@arm.com>
18543 David Sherwood <david.sherwood@arm.com>
18545 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18546 trying to combine a full-register integer set with a subreg
18549 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18550 Alan Hayward <alan.hayward@arm.com>
18551 David Sherwood <david.sherwood@arm.com>
18553 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18554 that is always either address_mode or pointer_mode.
18556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18557 Alan Hayward <alan.hayward@arm.com>
18558 David Sherwood <david.sherwood@arm.com>
18560 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18561 when the two are known to be equal.
18563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18564 Alan Hayward <alan.hayward@arm.com>
18565 David Sherwood <david.sherwood@arm.com>
18567 * simplify-rtx.c (simplify_const_unary_operation): Use
18568 is_a <scalar_int_mode> instead of checking for a nonzero
18569 precision. Forcibly convert op_mode to a scalar_int_mode
18570 in that case. More clearly differentiate the operand and
18571 result modes and use the former when deciding what the value
18572 of a count-bits operation should be. Use is_int_mode instead
18573 of checking for a MODE_INT. Remove redundant check for whether
18574 this mode has a zero precision.
18576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18577 Alan Hayward <alan.hayward@arm.com>
18578 David Sherwood <david.sherwood@arm.com>
18580 * optabs.c (widen_leading): Change the type of the mode argument
18581 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18582 (widen_bswap): Likewise.
18583 (expand_parity): Likewise.
18584 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18585 (expand_ffs): Likewise.
18586 (epand_unop): Check for scalar integer modes before calling the
18589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18590 Alan Hayward <alan.hayward@arm.com>
18591 David Sherwood <david.sherwood@arm.com>
18593 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18595 (expand_expr_real_1): Update call accordingly.
18597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18598 Alan Hayward <alan.hayward@arm.com>
18599 David Sherwood <david.sherwood@arm.com>
18601 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18602 value_mode arguments. Use scalar_int_mode internally.
18603 (store_bit_field_1): Rename the new integer mode from imode
18604 to op0_mode and use it instead of GET_MODE (op0). Update calls
18605 to store_split_bit_field, store_bit_field_using_insv and
18606 store_fixed_bit_field.
18607 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18608 Use scalar_int_mode internally. Use a bit count rather than a mode
18609 when calculating the largest bit size for get_best_mode.
18610 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18611 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18612 Remove assertion that OP0 has a scalar integer mode.
18613 (store_split_bit_field): Add op0_mode and value_mode arguments.
18614 Update calls to extract_fixed_bit_field.
18615 (extract_bit_field_using_extv): Add an op0_mode argument.
18616 Use scalar_int_mode internally.
18617 (extract_bit_field_1): Rename the new integer mode from imode to
18618 op0_mode and use it instead of GET_MODE (op0). Update calls to
18619 extract_split_bit_field, extract_bit_field_using_extv and
18620 extract_fixed_bit_field.
18621 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18622 to extract_split_bit_field and extract_fixed_bit_field_1.
18623 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18624 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18625 on the target mode.
18626 (extract_split_bit_field): Add an op0_mode argument. Update call
18627 to extract_fixed_bit_field.
18629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18630 Alan Hayward <alan.hayward@arm.com>
18631 David Sherwood <david.sherwood@arm.com>
18633 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18634 * explow.c (hard_function_value): Likewise.
18635 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18636 convert_to_mode call outside the loop.
18637 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18638 for the mode iterator. Require the mode specified by max_pieces
18640 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18642 (copy_blkmode_to_reg): Likewise.
18643 (set_storage_via_setmem): Likewise.
18644 * optabs.c (prepare_cmp_insn): Likewise.
18645 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18646 * stor-layout.c (finish_bitfield_representative): Likewise.
18648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18649 Alan Hayward <alan.hayward@arm.com>
18650 David Sherwood <david.sherwood@arm.com>
18652 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18653 * expr.c (convert_move): Use them.
18654 (convert_modes): Likewise.
18655 (store_expr_with_bounds): Likewise.
18657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18658 Alan Hayward <alan.hayward@arm.com>
18659 David Sherwood <david.sherwood@arm.com>
18661 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18662 parameter for the mode of "x". Remove the "known_x", "known_mode"
18663 and "known_ret" arguments. Change the type of the mode argument
18664 to scalar_int_mode.
18665 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18666 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18667 (reg_num_sign_bit_copies_for_combine): Likewise.
18668 * rtlanal.c (nonzero_bits1): Likewise.
18669 (num_sign_bit_copies1): Likewise.
18670 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18671 (reg_num_sign_bit_copies_general): Likewise.
18672 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18673 (reg_nonzero_bits_general): Likewise.
18675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18676 Alan Hayward <alan.hayward@arm.com>
18677 David Sherwood <david.sherwood@arm.com>
18679 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18680 than in subroutines. Return 1 for non-integer modes.
18681 (cached_num_sign_bit_copies): Change the type of the mode parameter
18682 to scalar_int_mode.
18683 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18684 classes. Handle CONST_INT_P first and then check whether X also
18685 has a scalar integer mode. Check the same thing for inner registers
18686 of a SUBREG and for values that are being extended or truncated.
18688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18689 Alan Hayward <alan.hayward@arm.com>
18690 David Sherwood <david.sherwood@arm.com>
18692 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18693 in subroutines. Return the mode mask for non-integer modes.
18694 (cached_nonzero_bits): Change the type of the mode parameter
18695 to scalar_int_mode.
18696 (nonzero_bits1): Likewise. Remove early exit for other mode
18697 classes. Handle CONST_INT_P first and then check whether X
18698 also has a scalar integer mode.
18700 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18701 Alan Hayward <alan.hayward@arm.com>
18702 David Sherwood <david.sherwood@arm.com>
18704 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18705 Return a scalar_int_mode and assert that the size is greater than
18707 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18708 (op_by_pieces_d::op_by_pieces_d): Likewise.
18709 (op_by_pieces_d::run): Likewise.
18710 (can_store_by_pieces): Likewise.
18712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18713 Alan Hayward <alan.hayward@arm.com>
18714 David Sherwood <david.sherwood@arm.com>
18716 * combine.c (extract_left_shift): Add a mode argument and update
18718 (make_compound_operation_int): Change the type of the mode parameter
18719 to scalar_int_mode and update the call to extract_left_shift.
18721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18722 Alan Hayward <alan.hayward@arm.com>
18723 David Sherwood <david.sherwood@arm.com>
18725 * combine.c (simplify_and_const_int): Change the type of the mode
18726 parameter to scalar_int_mode.
18727 (simplify_and_const_int_1): Likewise. Update recursive call.
18729 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18730 Alan Hayward <alan.hayward@arm.com>
18731 David Sherwood <david.sherwood@arm.com>
18733 * combine.c (simplify_compare_const): Check that the mode is a
18734 scalar_int_mode (rather than VOIDmode) before testing its
18736 (simplify_comparison): Move COMPARISON_P handling out of the
18737 loop and restrict the latter part of the loop to scalar_int_modes.
18738 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
18739 and when considering SUBREG_REGs. Use is_int_mode instead of
18740 checking GET_MODE_CLASS against MODE_INT.
18742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18743 Alan Hayward <alan.hayward@arm.com>
18744 David Sherwood <david.sherwood@arm.com>
18746 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18747 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18748 shift_unit_mode and for modes involved in scalar shifts.
18750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18751 Alan Hayward <alan.hayward@arm.com>
18752 David Sherwood <david.sherwood@arm.com>
18754 * combine.c (force_int_to_mode): New function, split out from...
18755 (force_to_mode): ...here. Keep xmode up-to-date and use it
18756 instead of GET_MODE (x).
18758 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18759 Alan Hayward <alan.hayward@arm.com>
18760 David Sherwood <david.sherwood@arm.com>
18762 * optabs-query.h (extraction_insn::struct_mode): Change type to
18763 opt_scalar_int_mode and update comment.
18764 (extraction_insn::field_mode): Change type to scalar_int_mode.
18765 (extraction_insn::pos_mode): Likewise.
18766 * combine.c (make_extraction): Update accordingly.
18767 * optabs-query.c (get_traditional_extraction_insn): Likewise.
18768 (get_optab_extraction_insn): Likewise.
18769 * recog.c (simplify_while_replacing): Likewise.
18770 * expmed.c (narrow_bit_field_mem): Change the type of the mode
18771 parameter to opt_scalar_int_mode.
18773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18774 Alan Hayward <alan.hayward@arm.com>
18775 David Sherwood <david.sherwood@arm.com>
18777 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18778 to a scalar_int_mode instead of a machine_mode.
18779 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
18780 (get_best_mode): Return a boolean and use a pointer argument to store
18781 the selected mode. Replace the limit mode parameter with a bit limit.
18782 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
18783 for the values returned by bit_field_mode_iterator::next_mode.
18784 (store_bit_field): Update call to get_best_mode.
18785 (store_fixed_bit_field): Likewise.
18786 (extract_fixed_bit_field): Likewise.
18787 * expr.c (optimize_bitfield_assignment_op): Likewise.
18788 * fold-const.c (optimize_bit_field_compare): Likewise.
18789 (fold_truth_andor_1): Likewise.
18790 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
18791 Update for new type of m_mode.
18792 (get_best_mode): As above.
18794 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18795 Alan Hayward <alan.hayward@arm.com>
18796 David Sherwood <david.sherwood@arm.com>
18798 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18799 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18800 (store_bit_field): Check is_a <scalar_int_mode> before calling
18801 strict_volatile_bitfield_p.
18802 (extract_bit_field): Likewise.
18804 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18805 Alan Hayward <alan.hayward@arm.com>
18806 David Sherwood <david.sherwood@arm.com>
18808 * target.def (cstore_mode): Return a scalar_int_mode.
18809 * doc/tm.texi: Regenerate.
18810 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18811 * targhooks.h (default_cstore_mode): Likewise.
18812 * targhooks.c (default_cstore_mode): Likewise, using a forced
18814 * expmed.c (emit_cstore): Expect the target of the cstore to be
18817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18818 Alan Hayward <alan.hayward@arm.com>
18819 David Sherwood <david.sherwood@arm.com>
18821 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18823 (niter_desc): Likewise mode.
18824 (iv_analyze): Add a mode parameter.
18826 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18827 and change its type to scalar_int_mode.
18828 * loop-iv.c: Update commentary at head of file.
18829 (iv_constant): Pass the mode paraeter before the rtx it describes
18830 and change its type to scalar_int_mode. Remove VOIDmode handling.
18831 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18832 (iv_extend): Likewise.
18833 (shorten_into_mode): Likewise.
18834 (iv_add): Use scalar_int_mode.
18835 (iv_mult): Likewise.
18836 (iv_shift): Likewise.
18837 (canonicalize_iv_subregs): Likewise.
18838 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18839 it describes and change its mode to scalar_int_mode. Also change
18840 the type of the returned inner_mode to scalar_int_mode.
18841 (get_biv_step): Likewise, turning outer_mode from a pointer
18842 into a direct parameter. Update call to get_biv_step_1.
18843 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18844 iv_constant and get_biv_step.
18845 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18846 and change its type to scalar_int_mode. Don't initialise iv->mode
18847 to VOIDmode and remove later checks for its still being VOIDmode.
18848 Update calls to iv_analyze_op and iv_analyze_expr. Check
18849 is_a <scalar_int_mode> when changing the mode under consideration.
18850 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18851 Update call to iv_analyze_expr.
18852 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18853 inner register is not also a scalar_int_mode. Update call to
18855 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18856 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18857 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18858 separate mode class checks. Update calls to iv_analyze. Remove
18859 fix-up of VOIDmodes after iv_analyze_biv.
18860 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18861 don't have a scalar_int_mode. Update call to biv_p.
18863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18864 Alan Hayward <alan.hayward@arm.com>
18865 David Sherwood <david.sherwood@arm.com>
18867 * cfgexpand.c (convert_debug_memory_address): Use
18868 as_a <scalar_int_mode>.
18869 * combine.c (expand_compound_operation): Likewise.
18870 (make_extraction): Likewise.
18871 (change_zero_ext): Likewise.
18872 (simplify_comparison): Likewise.
18873 * cse.c (cse_insn): Likewise.
18874 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18875 (mem_loc_descriptor): Likewise.
18876 (loc_descriptor): Likewise.
18877 * expmed.c (init_expmed_one_mode): Likewise.
18878 (synth_mult): Likewise.
18879 (emit_store_flag_1): Likewise.
18880 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18881 of a comparison with size.
18882 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18883 (reduce_to_bit_field_precision): Likewise.
18884 * function.c (expand_function_end): Likewise.
18885 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18886 * loop-doloop.c (doloop_modify): Likewise.
18887 * optabs.c (expand_binop): Likewise.
18888 (expand_unop): Likewise.
18889 (expand_copysign_absneg): Likewise.
18890 (prepare_cmp_insn): Likewise.
18891 (maybe_legitimize_operand): Likewise.
18892 * recog.c (const_scalar_int_operand): Likewise.
18893 * rtlanal.c (get_address_mode): Likewise.
18894 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18895 (simplify_cond_clz_ctz): Likewise.
18896 * tree-nested.c (get_nl_goto_field): Likewise.
18897 * tree.c (build_vector_type_for_mode): Likewise.
18898 * var-tracking.c (use_narrower_mode): Likewise.
18900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18901 Alan Hayward <alan.hayward@arm.com>
18902 David Sherwood <david.sherwood@arm.com>
18904 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18905 * builtins.c (expand_builtin_signbit): Use it.
18906 * cfgexpand.c (expand_debug_expr): Likewise.
18907 * dojump.c (do_jump): Likewise.
18908 (do_compare_and_jump): Likewise.
18909 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18910 * expmed.c (make_tree): Likewise.
18911 * expr.c (expand_expr_real_2): Likewise.
18912 (expand_expr_real_1): Likewise.
18913 (try_casesi): Likewise.
18914 * fold-const-call.c (fold_const_call_ss): Likewise.
18915 * fold-const.c (unextend): Likewise.
18916 (extract_muldiv_1): Likewise.
18917 (fold_single_bit_test): Likewise.
18918 (native_encode_int): Likewise.
18919 (native_encode_string): Likewise.
18920 (native_interpret_int): Likewise.
18921 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18922 * internal-fn.c (expand_addsub_overflow): Likewise.
18923 (expand_neg_overflow): Likewise.
18924 (expand_mul_overflow): Likewise.
18925 (expand_arith_overflow): Likewise.
18926 * match.pd: Likewise.
18927 * stor-layout.c (layout_type): Likewise.
18928 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18929 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18930 * tree-ssanames.c (get_range_info): Likewise.
18931 * tree-switch-conversion.c (array_value_type) Likewise.
18932 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18933 (vect_recog_divmod_pattern): Likewise.
18934 (vect_recog_mixed_size_cond_pattern): Likewise.
18935 * tree-vrp.c (extract_range_basic): Likewise.
18936 (simplify_float_conversion_using_ranges): Likewise.
18937 * tree.c (int_fits_type_p): Likewise.
18938 * ubsan.c (instrument_bool_enum_load): Likewise.
18939 * varasm.c (mergeable_string_section): Likewise.
18940 (narrowing_initializer_constant_valid_p): Likewise.
18941 (output_constant): Likewise.
18943 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18944 Alan Hayward <alan.hayward@arm.com>
18945 David Sherwood <david.sherwood@arm.com>
18947 * machmode.h (NARROWEST_INT_MODE): New macro.
18948 * expr.c (alignment_for_piecewise_move): Use it instead of
18949 GET_CLASS_NARROWEST_MODE (MODE_INT).
18950 (push_block): Likewise.
18951 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18953 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18956 Alan Hayward <alan.hayward@arm.com>
18957 David Sherwood <david.sherwood@arm.com>
18959 * postreload.c (move2add_valid_value_p): Change the type of the
18960 mode parameter to scalar_int_mode.
18961 (move2add_use_add2_insn): Add a mode parameter and use it instead
18963 (move2add_use_add3_insn): Likewise.
18964 (reload_cse_move2add): Update accordingly.
18966 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18967 Alan Hayward <alan.hayward@arm.com>
18968 David Sherwood <david.sherwood@arm.com>
18970 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18972 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18973 * optabs.c (expand_unop): Likewise.
18975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18976 Alan Hayward <alan.hayward@arm.com>
18977 David Sherwood <david.sherwood@arm.com>
18979 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18980 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18981 (popcount_loc_descriptor): Likewise.
18982 (bswap_loc_descriptor): Likewise.
18983 (rotate_loc_descriptor): Likewise.
18984 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18985 calling the functions above.
18987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18988 Alan Hayward <alan.hayward@arm.com>
18989 David Sherwood <david.sherwood@arm.com>
18991 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18993 (try_combine): Likewise.
18994 (simplify_if_then_else): Likewise.
18995 * cse.c (cse_insn): Likewise.
18996 * dwarf2out.c (mem_loc_descriptor): Likewise.
18997 * emit-rtl.c (gen_lowpart_common): Likewise.
18998 * simplify-rtx.c (simplify_truncation): Likewise.
18999 (simplify_binary_operation_1): Likewise.
19000 (simplify_const_relational_operation): Likewise.
19001 (simplify_ternary_operation): Likewise.
19002 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19005 Alan Hayward <alan.hayward@arm.com>
19006 David Sherwood <david.sherwood@arm.com>
19008 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19009 * reload.c (push_reload): Likewise.
19010 (find_reloads): Likewise.
19012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19013 Alan Hayward <alan.hayward@arm.com>
19014 David Sherwood <david.sherwood@arm.com>
19016 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19017 (make_compound_operation_int): Likewise.
19018 (change_zero_ext): Likewise.
19019 * expr.c (convert_move): Likewise.
19020 (convert_modes): Likewise.
19021 * fwprop.c (forward_propagate_subreg): Likewise.
19022 * loop-iv.c (get_biv_step_1): Likewise.
19023 * optabs.c (widen_operand): Likewise.
19024 * postreload.c (move2add_valid_value_p): Likewise.
19025 * recog.c (simplify_while_replacing): Likewise.
19026 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19027 (simplify_binary_operation_1): Likewise. Remove redundant
19028 mode equality check.
19030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19031 Alan Hayward <alan.hayward@arm.com>
19032 David Sherwood <david.sherwood@arm.com>
19034 * combine.c (combine_simplify_rtx): Add checks for
19035 is_a <scalar_int_mode>.
19036 (simplify_if_then_else): Likewise.
19037 (make_field_assignment): Likewise.
19038 (simplify_comparison): Likewise.
19039 * ifcvt.c (noce_try_bitop): Likewise.
19040 * loop-invariant.c (canonicalize_address_mult): Likewise.
19041 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
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 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19048 is_a <scalar_int_mode> instead of != BLKmode.
19050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19051 Alan Hayward <alan.hayward@arm.com>
19052 David Sherwood <david.sherwood@arm.com>
19054 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19055 instead of != VOIDmode.
19056 * combine.c (if_then_else_cond): Likewise.
19057 (change_zero_ext): Likewise.
19058 * dwarf2out.c (mem_loc_descriptor): Likewise.
19059 (loc_descriptor): Likewise.
19060 * rtlanal.c (canonicalize_condition): Likewise.
19061 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19064 Alan Hayward <alan.hayward@arm.com>
19065 David Sherwood <david.sherwood@arm.com>
19067 * simplify-rtx.c (simplify_binary_operation_1): Use
19068 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19071 Alan Hayward <alan.hayward@arm.com>
19072 David Sherwood <david.sherwood@arm.com>
19074 * wide-int.h (int_traits<unsigned char>) New class.
19075 (int_traits<unsigned short>) Likewise.
19076 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19077 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19079 * combine.c (set_nonzero_bits_and_sign_copies): Use
19080 is_a <scalar_int_mode>.
19081 (find_split_point): Likewise.
19082 (combine_simplify_rtx): Likewise.
19083 (simplify_logical): Likewise.
19084 (expand_compound_operation): Likewise.
19085 (expand_field_assignment): Likewise.
19086 (make_compound_operation): Likewise.
19087 (extended_count): Likewise.
19088 (change_zero_ext): Likewise.
19089 (simplify_comparison): Likewise.
19090 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19091 (ucompare_loc_descriptor): Likewise.
19092 (minmax_loc_descriptor): Likewise.
19093 (mem_loc_descriptor): Likewise.
19094 (loc_descriptor): Likewise.
19095 * expmed.c (init_expmed_one_mode): Likewise.
19096 * lra-constraints.c (lra_constraint_offset): Likewise.
19097 * optabs.c (prepare_libcall_arg): Likewise.
19098 * postreload.c (move2add_note_store): Likewise.
19099 * reload.c (operands_match_p): Likewise.
19100 * rtl.h (load_extend_op): Likewise.
19101 * rtlhooks.c (gen_lowpart_general): Likewise.
19102 * simplify-rtx.c (simplify_truncation): Likewise.
19103 (simplify_unary_operation_1): Likewise.
19104 (simplify_binary_operation_1): Likewise.
19105 (simplify_const_binary_operation): Likewise.
19106 (simplify_const_relational_operation): Likewise.
19107 (simplify_subreg): Likewise.
19108 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19109 * var-tracking.c (adjust_mems): Likewise.
19110 (prepare_call_arguments): Likewise.
19112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19113 Alan Hayward <alan.hayward@arm.com>
19114 David Sherwood <david.sherwood@arm.com>
19116 * machmode.h (is_int_mode): New fuction.
19117 * combine.c (find_split_point): Use it.
19118 (combine_simplify_rtx): Likewise.
19119 (simplify_if_then_else): Likewise.
19120 (simplify_set): Likewise.
19121 (simplify_shift_const_1): Likewise.
19122 (simplify_comparison): Likewise.
19123 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19124 * cse.c (notreg_cost): Likewise.
19125 (cse_insn): Likewise.
19126 * cselib.c (cselib_lookup_1): Likewise.
19127 * dojump.c (do_jump_1): Likewise.
19128 (do_compare_rtx_and_jump): Likewise.
19129 * dse.c (get_call_args): Likewise.
19130 * dwarf2out.c (rtl_for_decl_init): Likewise.
19131 (native_encode_initializer): Likewise.
19132 * expmed.c (emit_store_flag_1): Likewise.
19133 (emit_store_flag): Likewise.
19134 * expr.c (convert_modes): Likewise.
19135 (store_field): Likewise.
19136 (expand_expr_real_1): Likewise.
19137 * fold-const.c (fold_read_from_constant_string): Likewise.
19138 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19139 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19140 * optabs.c (expand_binop): Likewise.
19141 (expand_unop): Likewise.
19142 (expand_abs_nojump): Likewise.
19143 (expand_one_cmpl_abs_nojump): Likewise.
19144 * simplify-rtx.c (mode_signbit_p): Likewise.
19145 (val_signbit_p): Likewise.
19146 (val_signbit_known_set_p): Likewise.
19147 (val_signbit_known_clear_p): Likewise.
19148 (simplify_relational_operation_1): Likewise.
19149 * tree.c (vector_type_mode): Likewise.
19151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19152 Alan Hayward <alan.hayward@arm.com>
19153 David Sherwood <david.sherwood@arm.com>
19155 * machmode.h (smallest_mode_for_size): Fix formatting.
19156 (smallest_int_mode_for_size): New function.
19157 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19158 instead of smallest_mode_for_size.
19159 * combine.c (make_extraction): Likewise.
19160 * config/arc/arc.c (arc_expand_movmem): Likewise.
19161 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19162 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19163 * config/s390/s390.c (s390_expand_insv): Likewise.
19164 * config/sparc/sparc.c (assign_int_registers): Likewise.
19165 * config/spu/spu.c (spu_function_value): Likewise.
19166 (spu_function_arg): Likewise.
19167 * coverage.c (get_gcov_type): Likewise.
19168 (get_gcov_unsigned_t): Likewise.
19169 * dse.c (find_shift_sequence): Likewise.
19170 * expmed.c (store_bit_field_1): Likewise.
19171 * expr.c (convert_move): Likewise.
19172 (store_field): Likewise.
19173 * internal-fn.c (expand_arith_overflow): Likewise.
19174 * optabs-query.c (get_best_extraction_insn): Likewise.
19175 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19176 * stor-layout.c (layout_type): Likewise.
19177 (initialize_sizetypes): Likewise.
19178 * targhooks.c (default_get_mask_mode): Likewise.
19179 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19181 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19182 Alan Hayward <alan.hayward@arm.com>
19183 David Sherwood <david.sherwood@arm.com>
19185 * machmode.h (opt_mode::else_blk): New function.
19186 (int_mode_for_mode): Declare.
19187 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19188 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19190 * cfgexpand.c (expand_debug_expr): Likewise.
19191 * combine.c (gen_lowpart_or_truncate): Likewise.
19192 (gen_lowpart_for_combine): Likewise.
19193 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19194 * config/avr/avr.c (avr_to_int_mode): Likewise.
19195 (avr_out_plus_1): Likewise.
19196 (avr_out_plus): Likewise.
19197 (avr_out_round): Likewise.
19198 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19199 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19200 (s390_expand_vcond): Likewise.
19201 * config/spu/spu.c (spu_split_immediate): Likewise.
19202 (spu_expand_mov): Likewise.
19203 * dse.c (get_stored_val): Likewise.
19204 * expmed.c (store_bit_field_1): Likewise.
19205 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19207 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19208 (extract_low_bits): Likewise.
19209 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19210 handling rather than repeating the check.
19211 (emit_group_store): Likewise.
19212 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19213 * optabs.c (expand_absneg_bit): Likewise.
19214 (expand_copysign_absneg): Likewise.
19215 (expand_copysign_bit): Likewise.
19216 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19217 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19218 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19219 * var-tracking.c (prepare_call_arguments): Likewise.
19220 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19221 int_mode_for_mode instead of mode_for_size.
19222 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19225 Alan Hayward <alan.hayward@arm.com>
19226 David Sherwood <david.sherwood@arm.com>
19228 * machmode.h (int_mode_for_size): New function.
19229 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19230 instead of mode_for_size.
19231 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19233 * combine.c (expand_field_assignment): Use int_mode_for_size
19234 instead of mode_for_size.
19235 (make_extraction): Likewise.
19236 (simplify_shift_const_1): Likewise.
19237 (simplify_comparison): Likewise.
19238 * dojump.c (do_jump): Likewise.
19239 * dwarf2out.c (mem_loc_descriptor): Likewise.
19240 * emit-rtl.c (init_derived_machine_modes): Likewise.
19241 * expmed.c (flip_storage_order): Likewise.
19242 (convert_extracted_bit_field): Likewise.
19243 * expr.c (copy_blkmode_from_reg): Likewise.
19244 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19245 * internal-fn.c (expand_mul_overflow): Likewise.
19246 * lower-subreg.c (simple_move): Likewise.
19247 * optabs-libfuncs.c (init_optabs): Likewise.
19248 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19249 * tree.c (vector_type_mode): Likewise.
19250 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19251 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19252 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19253 * tree-vect-stmts.c (vectorizable_load): Likewise.
19254 (vectorizable_store): Likewise.
19256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19257 Alan Hayward <alan.hayward@arm.com>
19258 David Sherwood <david.sherwood@arm.com>
19260 * coretypes.h (pod_mode): New type.
19261 (scalar_int_mode_pod): New typedef.
19262 * machmode.h (pod_mode): New class.
19263 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19264 * genmodes.c (emit_mode_int_n): Update accordingly.
19265 * lower-subreg.h (target_lower_subreg): Change type to
19266 scalar_int_mode_pod.
19267 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19268 scalar_int_mode_pod.
19270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19271 Alan Hayward <alan.hayward@arm.com>
19272 David Sherwood <david.sherwood@arm.com>
19274 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19275 machine_mode to scalar_int_mode.
19276 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19277 (rs6000_option_override_internal): Remove cast to int.
19278 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19279 machine_mode to scalar_int_mode.
19280 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19281 (rs6000_option_override_internal): Remove cast to int.
19282 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19283 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19285 * config/s390/s390.c (s390_expand_builtin): Likewise.
19286 * coretypes.h (scalar_int_mode): New type.
19287 (opt_scalar_int_mode): New typedef.
19288 * machmode.h (scalar_int_mode): New class.
19289 (scalar_int_mode::includes_p): New function.
19290 (byte_mode): Change type to scalar_int_mode.
19291 (word_mode): Likewise.
19292 (ptr_mode): Likewise.
19293 * emit-rtl.c (byte_mode): Likewise.
19294 (word_mode): Likewise.
19295 (ptr_mode): Likewise.
19296 (init_derived_machine_modes): Update accordingly.
19297 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19298 and MODE_PARTIAL_INT.
19299 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19300 opt_scalar_int_mode.
19302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19303 Alan Hayward <alan.hayward@arm.com>
19304 David Sherwood <david.sherwood@arm.com>
19306 * target.def (libgcc_floating_mode_supported_p): Take a
19308 * doc/tm.texi: Regenerate.
19309 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19311 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19312 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19316 Alan Hayward <alan.hayward@arm.com>
19317 David Sherwood <david.sherwood@arm.com>
19319 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19320 * doc/tm.texi: Regenerate.
19321 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19322 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19323 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19324 * targhooks.h (default_floatn_mode): Likewise.
19325 * targhooks.c (default_floatn_mode): Likewise.
19326 * tree.c (build_common_tree_nodes): Update accordingly.
19328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19329 Alan Hayward <alan.hayward@arm.com>
19330 David Sherwood <david.sherwood@arm.com>
19332 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19333 (mode_iterator::iterate_p): Likewise.
19334 (mode_iterator::get_wider): Likewise.
19335 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19338 Alan Hayward <alan.hayward@arm.com>
19339 David Sherwood <david.sherwood@arm.com>
19341 * coretypes.h (opt_scalar_float_mode): New typedef.
19342 * machmode.h (float_mode_for_size): New function.
19343 * emit-rtl.c (double_mode): Delete.
19344 (init_emit_once): Use float_mode_for_size.
19345 * stor-layout.c (layout_type): Likewise.
19346 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19349 Alan Hayward <alan.hayward@arm.com>
19350 David Sherwood <david.sherwood@arm.com>
19352 * output.h (assemble_real): Take a scalar_float_mode.
19353 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19354 * config/arm/arm.md (consttable_4): Likewise.
19355 (consttable_8): Likewise.
19356 (consttable_16): Likewise.
19357 * config/mips/mips.md (consttable_float): Likewise.
19358 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19359 * varasm.c (assemble_real): Take a scalar_float_mode.
19360 (output_constant_pool_2): Update accordingly.
19361 (output_constant): Likewise.
19363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19364 Alan Hayward <alan.hayward@arm.com>
19365 David Sherwood <david.sherwood@arm.com>
19367 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19368 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19369 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19370 (native_encode_real): Likewise.
19371 (native_interpret_real): Likewise.
19372 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19373 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19376 Alan Hayward <alan.hayward@arm.com>
19377 David Sherwood <david.sherwood@arm.com>
19379 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19380 <scalar_float_mode>. Simplify.
19381 (gen_extend_conv_libfunc): Likewise.
19383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19384 Alan Hayward <alan.hayward@arm.com>
19385 David Sherwood <david.sherwood@arm.com>
19387 * coretypes.h (scalar_float_mode): New type.
19388 * machmode.h (mode_traits::from_int): Use machine_mode if
19389 USE_ENUM_MODES is defined.
19390 (is_a): New function.
19392 (dyn_cast): Likewise.
19393 (scalar_float_mode): New class.
19394 (scalar_float_mode::includes_p): New function.
19395 (is_float_mode): Likewise.
19396 * gdbhooks.py (MachineModePrinter): New class.
19397 (build_pretty_printer): Use it for scalar_float_mode.
19398 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19399 (format_helper::format_helper): Turn into a template.
19400 * genmodes.c (get_mode_class): New function.
19401 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19402 or machine_mode if none.
19403 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19404 as_a <scalar_float_mode>.
19405 * dwarf2out.c (mem_loc_descriptor): Likewise.
19406 (insert_float): Likewise.
19407 (add_const_value_attribute): Likewise.
19408 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19409 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19410 (expand_unop): Update accordingly.
19411 (expand_abs_nojump): Likewise.
19412 (expand_copysign_absneg): Take a scalar_float_mode.
19413 (expand_copysign_bit): Likewise.
19414 (expand_copysign): Update accordingly.
19416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19417 Alan Hayward <alan.hayward@arm.com>
19418 David Sherwood <david.sherwood@arm.com>
19420 * coretypes.h (opt_mode): New class.
19421 * machmode.h (opt_mode): Likewise.
19422 (opt_mode::else_void): New function.
19423 (opt_mode::require): Likewise.
19424 (opt_mode::exists): Likewise.
19425 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19426 (GET_MODE_2XWIDER_MODE): Likewise.
19427 (mode_iterator::get_wider): Update accordingly.
19428 (mode_iterator::get_2xwider): Likewise.
19429 (mode_iterator::get_known_wider): Likewise, turning into a template.
19430 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19431 forcing a wider mode to exist.
19432 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19433 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19434 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19435 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19436 * lower-subreg.c (init_lower_subreg): Likewise.
19437 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19438 on the final iteration.
19439 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19440 a wider mode exists before asking for a move pattern.
19441 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19442 forcing a wider mode to exist.
19443 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19444 returning false if no such mode exists.
19445 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19446 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19447 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19448 Avoid checking for a MODE_INT if we already know the mode is not a
19450 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19451 forcing a wider mode to exist.
19452 (expmed_mult_highpart_optab): Likewise.
19453 (expmed_mult_highpart): Likewise.
19454 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19456 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19457 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19458 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19459 * internal-fn.c (expand_mul_overflow): Update use of
19460 GET_MODE_2XWIDER_MODE.
19461 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19462 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19463 GET_MODE_WIDER_MODE.
19464 (convert_plusminus_to_widen): Likewise.
19465 * tree-switch-conversion.c (array_value_type): Likewise.
19466 * var-tracking.c (emit_note_insn_var_location): Likewise.
19467 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19468 Return false inside rather than outside the loop if no wider mode
19470 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19471 and GET_MODE_2XWIDER_MODE
19472 (can_compare_p): Use else_void.
19473 * gdbhooks.py (OptMachineModePrinter): New class.
19474 (build_pretty_printer): Use it for opt_mode.
19476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19477 Alan Hayward <alan.hayward@arm.com>
19478 David Sherwood <david.sherwood@arm.com>
19480 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19481 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19484 Alan Hayward <alan.hayward@arm.com>
19485 David Sherwood <david.sherwood@arm.com>
19487 * machmode.h (mode_traits): New structure.
19488 (get_narrowest_mode): New function.
19489 (mode_iterator::start): Likewise.
19490 (mode_iterator::iterate_p): Likewise.
19491 (mode_iterator::get_wider): Likewise.
19492 (mode_iterator::get_known_wider): Likewise.
19493 (mode_iterator::get_2xwider): Likewise.
19494 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19495 (FOR_EACH_MODE): Likewise.
19496 (FOR_EACH_MODE_FROM): Likewise.
19497 (FOR_EACH_MODE_UNTIL): Likewise.
19498 (FOR_EACH_WIDER_MODE): Likewise.
19499 (FOR_EACH_2XWIDER_MODE): Likewise.
19500 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19501 * combine.c (simplify_comparison): Likewise
19502 * config/i386/i386.c (type_natural_mode): Likewise.
19503 * cse.c (cse_insn): Likewise.
19504 * dse.c (find_shift_sequence): Likewise.
19505 * emit-rtl.c (init_derived_machine_modes): Likewise.
19506 (init_emit_once): Likewise.
19507 * explow.c (hard_function_value): Likewise.
19508 * expmed.c (extract_fixed_bit_field_1): Likewise.
19509 (extract_bit_field_1): Likewise.
19510 (expand_divmod): Likewise.
19511 (emit_store_flag_1): Likewise.
19512 * expr.c (init_expr_target): Likewise.
19513 (convert_move): Likewise.
19514 (alignment_for_piecewise_move): Likewise.
19515 (widest_int_mode_for_size): Likewise.
19516 (emit_block_move_via_movmem): Likewise.
19517 (copy_blkmode_to_reg): Likewise.
19518 (set_storage_via_setmem): Likewise.
19519 (compress_float_constant): Likewise.
19520 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19521 * optabs-query.c (get_best_extraction_insn): Likewise.
19522 * optabs.c (expand_binop): Likewise.
19523 (expand_twoval_unop): Likewise.
19524 (expand_twoval_binop): Likewise.
19525 (widen_leading): Likewise.
19526 (widen_bswap): Likewise.
19527 (expand_parity): Likewise.
19528 (expand_unop): Likewise.
19529 (prepare_cmp_insn): Likewise.
19530 (prepare_float_lib_cmp): Likewise.
19531 (expand_float): Likewise.
19532 (expand_fix): Likewise.
19533 (expand_sfix_optab): Likewise.
19534 * postreload.c (move2add_use_add2_insn): Likewise.
19535 * reg-stack.c (reg_to_stack): Likewise.
19536 * reginfo.c (choose_hard_reg_mode): Likewise.
19537 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19538 * stor-layout.c (mode_for_size): Likewise.
19539 (smallest_mode_for_size): Likewise.
19540 (mode_for_vector): Likewise.
19541 (finish_bitfield_representative): Likewise.
19542 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19543 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19544 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19545 * var-tracking.c (prepare_call_arguments): Likewise.
19547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19548 Alan Hayward <alan.hayward@arm.com>
19549 David Sherwood <david.sherwood@arm.com>
19551 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19552 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19553 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19554 * machmode.h (mode_size): Move earlier in file.
19555 (mode_precision): Likewise.
19556 (mode_inner): Likewise.
19557 (mode_nunits): Likewise.
19558 (mode_unit_size): Likewise.
19559 (unit_unit_precision): Likewise.
19560 (mode_wider): Likewise.
19561 (mode_2xwider): Likewise.
19562 (machine_mode): New class.
19563 (mode_to_bytes): New function.
19564 (mode_to_bits): Likewise.
19565 (mode_to_precision): Likewise.
19566 (mode_to_inner): Likewise.
19567 (mode_to_unit_size): Likewise.
19568 (mode_to_unit_precision): Likewise.
19569 (mode_to_nunits): Likewise.
19570 (GET_MODE_SIZE): Use mode_to_bytes.
19571 (GET_MODE_BITSIZE): Use mode_to_bits.
19572 (GET_MODE_PRECISION): Use mode_to_precision.
19573 (GET_MODE_INNER): Use mode_to_inner.
19574 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19575 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19576 (GET_MODE_NUNITS): Use mode_to_nunits.
19577 * system.h (ALWAYS_INLINE): New macro.
19578 * config/powerpcspe/powerpcspe-c.c
19579 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19580 int for arg1_mode and arg2_mode.
19582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19583 Alan Hayward <alan.hayward@arm.com>
19584 David Sherwood <david.sherwood@arm.com>
19586 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19587 Prefix mode names with E_ in case statements.
19588 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19589 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19590 (aarch64_split_simd_move): Likewise.
19591 (aarch64_gen_storewb_pair): Likewise.
19592 (aarch64_gen_loadwb_pair): Likewise.
19593 (aarch64_gen_store_pair): Likewise.
19594 (aarch64_gen_load_pair): Likewise.
19595 (aarch64_get_condition_code_1): Likewise.
19596 (aarch64_constant_pool_reload_icode): Likewise.
19597 (get_rsqrte_type): Likewise.
19598 (get_rsqrts_type): Likewise.
19599 (get_recpe_type): Likewise.
19600 (get_recps_type): Likewise.
19601 (aarch64_gimplify_va_arg_expr): Likewise.
19602 (aarch64_simd_container_mode): Likewise.
19603 (aarch64_emit_load_exclusive): Likewise.
19604 (aarch64_emit_store_exclusive): Likewise.
19605 (aarch64_expand_compare_and_swap): Likewise.
19606 (aarch64_gen_atomic_cas): Likewise.
19607 (aarch64_emit_bic): Likewise.
19608 (aarch64_emit_atomic_swap): Likewise.
19609 (aarch64_emit_atomic_load_op): Likewise.
19610 (aarch64_evpc_trn): Likewise.
19611 (aarch64_evpc_uzp): Likewise.
19612 (aarch64_evpc_zip): Likewise.
19613 (aarch64_evpc_ext): Likewise.
19614 (aarch64_evpc_rev): Likewise.
19615 (aarch64_evpc_dup): Likewise.
19616 (aarch64_gen_ccmp_first): Likewise.
19617 (aarch64_gen_ccmp_next): Likewise.
19618 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19619 (alpha_emit_xfloating_libcall): Likewise.
19620 (emit_insxl): Likewise.
19621 (alpha_arg_type): Likewise.
19622 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19623 (arc_preferred_simd_mode): Likewise.
19624 (arc_secondary_reload): Likewise.
19625 (get_arc_condition_code): Likewise.
19626 (arc_print_operand): Likewise.
19627 (arc_legitimate_constant_p): Likewise.
19628 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19629 * config/arc/arc.md (casesi_load): Likewise.
19630 (casesi_compact_jump): Likewise.
19631 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19632 (cc_use_register): Likewise.
19633 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19634 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19635 (arm_init_iwmmxt_builtins): Likewise.
19636 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19637 (neon_expand_vector_init): Likewise.
19638 (arm_attr_length_move_neon): Likewise.
19639 (maybe_get_arm_condition_code): Likewise.
19640 (arm_emit_vector_const): Likewise.
19641 (arm_preferred_simd_mode): Likewise.
19642 (arm_output_iwmmxt_tinsr): Likewise.
19643 (thumb1_output_casesi): Likewise.
19644 (thumb2_output_casesi): Likewise.
19645 (arm_emit_load_exclusive): Likewise.
19646 (arm_emit_store_exclusive): Likewise.
19647 (arm_expand_compare_and_swap): Likewise.
19648 (arm_evpc_neon_vuzp): Likewise.
19649 (arm_evpc_neon_vzip): Likewise.
19650 (arm_evpc_neon_vrev): Likewise.
19651 (arm_evpc_neon_vtrn): Likewise.
19652 (arm_evpc_neon_vext): Likewise.
19653 (arm_validize_comparison): Likewise.
19654 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19655 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19656 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19657 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19658 (c6x_preferred_simd_mode): Likewise.
19659 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19660 (epiphany_rtx_costs): Likewise.
19661 * config/epiphany/predicates.md (proper_comparison_operator):
19663 * config/frv/frv.c (condexec_memory_operand): Likewise.
19664 (frv_emit_move): Likewise.
19665 (output_move_single): Likewise.
19666 (output_condmove_single): Likewise.
19667 (frv_hard_regno_mode_ok): Likewise.
19668 (frv_matching_accg_mode): Likewise.
19669 * config/h8300/h8300.c (split_adds_subs): Likewise.
19670 (h8300_rtx_costs): Likewise.
19671 (h8300_print_operand): Likewise.
19672 (compute_mov_length): Likewise.
19673 (output_logical_op): Likewise.
19674 (compute_logical_op_length): Likewise.
19675 (compute_logical_op_cc): Likewise.
19676 (h8300_shift_needs_scratch_p): Likewise.
19677 (output_a_shift): Likewise.
19678 (compute_a_shift_length): Likewise.
19679 (compute_a_shift_cc): Likewise.
19680 (expand_a_rotate): Likewise.
19681 (output_a_rotate): Likewise.
19682 * config/i386/i386.c (classify_argument): Likewise.
19683 (function_arg_advance_32): Likewise.
19684 (function_arg_32): Likewise.
19685 (function_arg_64): Likewise.
19686 (function_value_64): Likewise.
19687 (ix86_gimplify_va_arg): Likewise.
19688 (ix86_legitimate_constant_p): Likewise.
19689 (put_condition_code): Likewise.
19690 (split_double_mode): Likewise.
19691 (ix86_avx256_split_vector_move_misalign): Likewise.
19692 (ix86_expand_vector_logical_operator): Likewise.
19693 (ix86_split_idivmod): Likewise.
19694 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19695 (ix86_build_const_vector): Likewise.
19696 (ix86_build_signbit_mask): Likewise.
19697 (ix86_match_ccmode): Likewise.
19698 (ix86_cc_modes_compatible): Likewise.
19699 (ix86_expand_branch): Likewise.
19700 (ix86_expand_sse_cmp): Likewise.
19701 (ix86_expand_sse_movcc): Likewise.
19702 (ix86_expand_int_sse_cmp): Likewise.
19703 (ix86_expand_vec_perm_vpermi2): Likewise.
19704 (ix86_expand_vec_perm): Likewise.
19705 (ix86_expand_sse_unpack): Likewise.
19706 (ix86_expand_int_addcc): Likewise.
19707 (ix86_split_to_parts): Likewise.
19708 (ix86_vectorize_builtin_gather): Likewise.
19709 (ix86_vectorize_builtin_scatter): Likewise.
19710 (avx_vpermilp_parallel): Likewise.
19711 (inline_memory_move_cost): Likewise.
19712 (ix86_tieable_integer_mode_p): Likewise.
19713 (x86_maybe_negate_const_int): Likewise.
19714 (ix86_expand_vector_init_duplicate): Likewise.
19715 (ix86_expand_vector_init_one_nonzero): Likewise.
19716 (ix86_expand_vector_init_one_var): Likewise.
19717 (ix86_expand_vector_init_concat): Likewise.
19718 (ix86_expand_vector_init_interleave): Likewise.
19719 (ix86_expand_vector_init_general): Likewise.
19720 (ix86_expand_vector_set): Likewise.
19721 (ix86_expand_vector_extract): Likewise.
19722 (emit_reduc_half): Likewise.
19723 (ix86_emit_i387_round): Likewise.
19724 (ix86_mangle_type): Likewise.
19725 (ix86_expand_round_sse4): Likewise.
19726 (expand_vec_perm_blend): Likewise.
19727 (canonicalize_vector_int_perm): Likewise.
19728 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
19729 (expand_vec_perm_1): Likewise.
19730 (expand_vec_perm_interleave3): Likewise.
19731 (expand_vec_perm_even_odd_pack): Likewise.
19732 (expand_vec_perm_even_odd_1): Likewise.
19733 (expand_vec_perm_broadcast_1): Likewise.
19734 (ix86_vectorize_vec_perm_const_ok): Likewise.
19735 (ix86_expand_vecop_qihi): Likewise.
19736 (ix86_expand_mul_widen_hilo): Likewise.
19737 (ix86_expand_sse2_abs): Likewise.
19738 (ix86_expand_pextr): Likewise.
19739 (ix86_expand_pinsr): Likewise.
19740 (ix86_preferred_simd_mode): Likewise.
19741 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19742 * config/i386/sse.md (*andnot<mode>3): Likewise.
19743 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19744 (*<code><mode>3): Likewise.
19745 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19746 (ia64_expand_atomic_op): Likewise.
19747 (ia64_arg_type): Likewise.
19748 (ia64_mode_to_int): Likewise.
19749 (ia64_scalar_mode_supported_p): Likewise.
19750 (ia64_vector_mode_supported_p): Likewise.
19751 (expand_vec_perm_broadcast): Likewise.
19752 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19753 (iq2000_function_arg_advance): Likewise.
19754 (iq2000_function_arg): Likewise.
19755 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19756 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19757 (m68k_libcall_value): Likewise.
19758 (m68k_function_value): Likewise.
19759 (sched_attr_op_type): Likewise.
19760 * config/mcore/mcore.c (mcore_output_move): Likewise.
19761 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19763 (microblaze_function_arg): Likewise.
19764 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19765 (mips_print_operand): Likewise.
19766 (mips_mode_ok_for_mov_fmt_p): Likewise.
19767 (mips_vector_mode_supported_p): Likewise.
19768 (mips_preferred_simd_mode): Likewise.
19769 (mips_expand_vpc_loongson_even_odd): Likewise.
19770 (mips_expand_vec_unpack): Likewise.
19771 (mips_expand_vi_broadcast): Likewise.
19772 (mips_expand_vector_init): Likewise.
19773 (mips_expand_vec_reduc): Likewise.
19774 (mips_expand_msa_cmp): Likewise.
19775 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19776 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19777 (cc_flags_for_mode): Likewise.
19778 * config/msp430/msp430.c (msp430_print_operand): Likewise.
19779 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
19780 (nds32_output_casesi_pc_relative): Likewise.
19781 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19782 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
19783 (nvptx_gen_unpack): Likewise.
19784 (nvptx_gen_pack): Likewise.
19785 (nvptx_gen_shuffle): Likewise.
19786 (nvptx_gen_wcast): Likewise.
19787 (nvptx_preferred_simd_mode): Likewise.
19788 * config/pa/pa.c (pa_secondary_reload): Likewise.
19789 * config/pa/predicates.md (base14_operand): Likewise.
19790 * config/powerpcspe/powerpcspe-c.c
19791 (altivec_resolve_overloaded_builtin): Likewise.
19792 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
19794 (rs6000_preferred_simd_mode): Likewise.
19795 (output_vec_const_move): Likewise.
19796 (rs6000_expand_vector_extract): Likewise.
19797 (rs6000_split_vec_extract_var): Likewise.
19798 (reg_offset_addressing_ok_p): Likewise.
19799 (rs6000_legitimate_offset_address_p): Likewise.
19800 (rs6000_legitimize_address): Likewise.
19801 (rs6000_emit_set_const): Likewise.
19802 (rs6000_const_vec): Likewise.
19803 (rs6000_emit_move): Likewise.
19804 (spe_build_register_parallel): Likewise.
19805 (rs6000_darwin64_record_arg_recurse): Likewise.
19806 (swap_selector_for_mode): Likewise.
19807 (spe_init_builtins): Likewise.
19808 (paired_init_builtins): Likewise.
19809 (altivec_init_builtins): Likewise.
19810 (do_load_for_compare): Likewise.
19811 (rs6000_generate_compare): Likewise.
19812 (rs6000_expand_float128_convert): Likewise.
19813 (emit_load_locked): Likewise.
19814 (emit_store_conditional): Likewise.
19815 (rs6000_output_function_epilogue): Likewise.
19816 (rs6000_handle_altivec_attribute): Likewise.
19817 (rs6000_function_value): Likewise.
19818 (emit_fusion_gpr_load): Likewise.
19819 (emit_fusion_p9_load): Likewise.
19820 (emit_fusion_p9_store): Likewise.
19821 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19822 (fusion_gpr_mem_load): Likewise.
19823 (fusion_addis_mem_combo_load): Likewise.
19824 (fusion_addis_mem_combo_store): Likewise.
19825 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19826 (fusion_gpr_mem_load): Likewise.
19827 (fusion_addis_mem_combo_load): Likewise.
19828 (fusion_addis_mem_combo_store): Likewise.
19829 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19831 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19832 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19833 (rs6000_preferred_simd_mode): Likewise.
19834 (output_vec_const_move): Likewise.
19835 (rs6000_expand_vector_extract): Likewise.
19836 (rs6000_split_vec_extract_var): Likewise.
19837 (reg_offset_addressing_ok_p): Likewise.
19838 (rs6000_legitimate_offset_address_p): Likewise.
19839 (rs6000_legitimize_address): Likewise.
19840 (rs6000_emit_set_const): Likewise.
19841 (rs6000_const_vec): Likewise.
19842 (rs6000_emit_move): Likewise.
19843 (rs6000_darwin64_record_arg_recurse): Likewise.
19844 (swap_selector_for_mode): Likewise.
19845 (paired_init_builtins): Likewise.
19846 (altivec_init_builtins): Likewise.
19847 (rs6000_expand_float128_convert): Likewise.
19848 (emit_load_locked): Likewise.
19849 (emit_store_conditional): Likewise.
19850 (rs6000_output_function_epilogue): Likewise.
19851 (rs6000_handle_altivec_attribute): Likewise.
19852 (rs6000_function_value): Likewise.
19853 (emit_fusion_gpr_load): Likewise.
19854 (emit_fusion_p9_load): Likewise.
19855 (emit_fusion_p9_store): Likewise.
19856 * config/rx/rx.c (rx_gen_move_template): Likewise.
19857 (flags_from_mode): Likewise.
19858 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19859 (s390_slb_comparison): Likewise.
19860 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19861 (s390_vector_mode_supported_p): Likewise.
19862 (s390_cc_modes_compatible): Likewise.
19863 (s390_match_ccmode_set): Likewise.
19864 (s390_canonicalize_comparison): Likewise.
19865 (s390_emit_compare_and_swap): Likewise.
19866 (s390_branch_condition_mask): Likewise.
19867 (s390_rtx_costs): Likewise.
19868 (s390_secondary_reload): Likewise.
19869 (__SECONDARY_RELOAD_CASE): Likewise.
19870 (s390_expand_cs): Likewise.
19871 (s390_preferred_simd_mode): Likewise.
19872 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19873 * config/sh/sh.c (sh_print_operand): Likewise.
19874 (dump_table): Likewise.
19875 (sh_secondary_reload): Likewise.
19876 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19877 * config/sh/sh.md (casesi_worker_1): Likewise.
19878 (casesi_worker_2): Likewise.
19879 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19880 (fcc_comparison_operator): Likewise.
19881 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19882 (emit_soft_tfmode_cvt): Likewise.
19883 (sparc_preferred_simd_mode): Likewise.
19884 (output_cbranch): Likewise.
19885 (sparc_print_operand): Likewise.
19886 (sparc_expand_vec_perm_bmask): Likewise.
19887 (vector_init_bshuffle): Likewise.
19888 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19889 (spu_vector_mode_supported_p): Likewise.
19890 (spu_expand_insv): Likewise.
19891 (spu_emit_branch_or_set): Likewise.
19892 (spu_handle_vector_attribute): Likewise.
19893 (spu_builtin_splats): Likewise.
19894 (spu_builtin_extract): Likewise.
19895 (spu_builtin_promote): Likewise.
19896 (spu_expand_sign_extend): Likewise.
19897 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19898 (tilegx_simd_int): Likewise.
19899 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19900 (tilepro_simd_int): Likewise.
19901 * config/v850/v850.c (const_double_split): Likewise.
19902 (v850_print_operand): Likewise.
19903 (ep_memory_offset): Likewise.
19904 * config/vax/vax.c (vax_rtx_costs): Likewise.
19905 (vax_output_int_move): Likewise.
19906 (vax_output_int_add): Likewise.
19907 (vax_output_int_subtract): Likewise.
19908 * config/visium/predicates.md (visium_branch_operator): Likewise.
19909 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19910 (visium_print_operand_address): Likewise.
19911 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19912 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19913 (xtensa_expand_conditional_branch): Likewise.
19914 (xtensa_copy_incoming_a7): Likewise.
19915 (xtensa_output_literal): Likewise.
19916 * dfp.c (decimal_real_maxval): Likewise.
19917 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19919 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19920 Alan Hayward <alan.hayward@arm.com>
19921 David Sherwood <david.sherwood@arm.com>
19923 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19924 (mode_nunits_inline): Likewise.
19925 (mode_inner_inline): Likewise.
19926 (mode_unit_size_inline): Likewise.
19927 (mode_unit_precision_inline): Likewise.
19928 (emit_insn_modes_h): Likewise. Also emit a #define of the
19930 (emit_mode_wider): Add an E_ prefix to mode names.
19931 (emit_mode_complex): Likewise.
19932 (emit_mode_inner): Likewise.
19933 (emit_mode_adjustments): Likewise.
19934 (emit_mode_int_n): Likewise.
19935 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19936 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19937 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19938 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19939 (CRC32_BUILTIN, ENTRY): Likewise.
19940 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19941 (aarch64_pop_regs): Likewise.
19942 (aarch64_process_components): Likewise.
19943 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19944 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19945 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19946 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19947 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19948 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19950 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19951 (pcmpistr): Likewise.
19952 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19953 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19954 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19956 * config/rl78/rl78.c (mduc_regs): Likewise.
19957 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19958 (htm_expand_builtin): Likewise.
19959 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19960 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19961 * config/xtensa/xtensa.c (print_operand): Likewise.
19962 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19963 (NUM_MODE_VECTOR_INT): Likewise.
19964 * genoutput.c (null_operand): Likewise.
19965 (output_operand_data): Likewise.
19966 * genrecog.c (print_parameter_value): Likewise.
19967 * lra.c (debug_operand_data): Likewise.
19969 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19971 * dwarf2out.c (scompare_loc_descriptor_wide)
19972 (scompare_loc_descriptor_narrow): New functions, split out from...
19973 (scompare_loc_descriptor): ...here.
19974 * expmed.c (emit_store_flag_int): New function, split out from...
19975 (emit_store_flag): ...here.
19977 2017-08-30 Richard Biener <rguenther@suse.de>
19979 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19980 (dwarf2out_early_finish): Move setting of AT_pubnames from
19981 early debug output to early finish.
19983 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19985 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19986 and -mdata-region to the assembler.
19988 2017-08-30 Richard Biener <rguenther@suse.de>
19990 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19992 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19993 (dwarf2out_early_global_decl): Only generate a DIE for the
19994 abstract origin if it doesn't already exist or is a declaration DIE.
19995 (resolve_addr): Do not add the linkage name twice when
19996 generating a stub DIE for the DW_TAG_GNU_call_site target.
19998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20000 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20001 Use machine_mode rather than int for arg1_mode.
20003 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20006 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20007 that the second argument of the built-in functions to unpack
20008 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20009 switch statement instead a lot of if statements.
20010 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20011 Allow 64-bit values to be in Altivec registers as well as
20012 traditional floating point registers.
20013 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20015 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20017 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20018 MAX_REGS_PER_ADDRESS == 1.
20020 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20022 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20023 * config/i386/i386.c (ix86_option_override_internal): Simplify
20024 setting of opts->x_flag_entry.
20026 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20027 Jakub Jelinek <jakub@redhat.com>
20028 Richard Biener <rguenther@suse.de>
20030 PR tree-optimization/81503
20031 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20032 folded constant fits in the target type; reorder tests for clarity.
20034 2017-08-29 Martin Liska <mliska@suse.cz>
20036 * passes.def: Include pass_lower_switch.
20037 * stmt.c (dump_case_nodes): Remove and move to
20038 tree-switch-conversion.
20039 (case_values_threshold): Likewise.
20040 (expand_switch_as_decision_tree_p): Likewise.
20041 (emit_case_decision_tree): Likewise.
20042 (expand_case): Likewise.
20043 (balance_case_nodes): Likewise.
20044 (node_has_low_bound): Likewise.
20045 (node_has_high_bound): Likewise.
20046 (node_is_bounded): Likewise.
20047 (emit_case_nodes): Likewise.
20048 (struct simple_case_node): New struct.
20049 (add_case_node): Remove.
20050 (emit_case_dispatch_table): Use vector instead of case_list.
20051 (reset_out_edges_aux): Remove.
20052 (compute_cases_per_edge): Likewise.
20053 (expand_case): Build list of simple_case_node.
20054 (expand_sjlj_dispatch_table): Use it.
20055 * tree-switch-conversion.c (struct case_node): Moved from
20056 stmt.c and adjusted.
20057 (emit_case_nodes): Likewise.
20058 (node_has_low_bound): Likewise.
20059 (node_has_high_bound): Likewise.
20060 (node_is_bounded): Likewise.
20061 (case_values_threshold): Likewise.
20062 (reset_out_edges_aux): Likewise.
20063 (compute_cases_per_edge): Likewise.
20064 (add_case_node): Likewise.
20065 (dump_case_nodes): Likewise.
20066 (balance_case_nodes): Likewise.
20067 (expand_switch_as_decision_tree_p): Likewise.
20068 (emit_jump): Likewise.
20069 (emit_case_decision_tree): Likewise.
20070 (try_switch_expansion): Likewise.
20071 (do_jump_if_equal): Likewise.
20072 (emit_cmp_and_jump_insns): Likewise.
20073 (fix_phi_operands_for_edge): New function.
20074 (record_phi_operand_mapping): Likewise.
20075 (class pass_lower_switch): New pass.
20076 (pass_lower_switch::execute): New function.
20077 (make_pass_lower_switch): Likewise.
20078 (conditional_probability):
20079 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20080 * tree-pass.h: Add make_pass_lower_switch.
20082 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20085 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20088 2017-08-29 Richard Biener <rguenther@suse.de>
20090 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20091 we do not add a DW_AT_inline attribute twice.
20092 (gen_subprogram_die): Remove code setting DW_AT_inline on
20093 DECL_ABSTRACT_P nodes.
20095 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20097 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20098 calls to internal functions.
20099 (gimplify_modify_expr): Likewise.
20100 * tree-call-cdce.c (use_internal_fn): Likewise.
20101 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20102 (convert_to_divmod): Set the nothrow flag.
20103 * tree-if-conv.c (predicate_mem_writes): Likewise.
20104 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20105 (vectorizable_call): Likewise.
20106 (vectorizable_store): Likewise.
20107 (vectorizable_load): Likewise.
20108 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20109 (vect_recog_mask_conversion_pattern): Likewise.
20111 2017-08-29 Martin Liska <mliska@suse.cz>
20114 * gcc.c (driver_handle_option): Add new argument.
20115 * opts-common.c (handle_option): Pass
20116 target_option_override_hook.
20117 * opts-global.c (lang_handle_option): Add new option.
20118 (set_default_handlers): Add new argument.
20119 (decode_options): Likewise.
20120 * opts.c (target_handle_option): Likewise.
20121 (common_handle_option): Call target_option_override_hook.
20122 * opts.h (struct cl_option_handler_func): Add hook for
20123 target option override.
20124 (struct cl_option_handlers): Likewise.
20125 (set_default_handlers): Add new argument.
20126 (decode_options): Likewise.
20127 (common_handle_option): Likewise.
20128 (target_handle_option): Likewise.
20129 * toplev.c (toplev::main): Pass targetm.target_option.override
20132 2017-08-29 Richard Biener <rguenther@suse.de>
20133 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20135 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20136 life to the active subtree.
20138 2017-08-28 Jeff Law <law@redhat.com>
20140 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20141 derive_equivalences.
20142 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20143 Code moved into....
20144 (edge_info::derive_equivalences): New private member function
20146 * tree-ssa-dom.c (class edge_info): Changed from a struct
20147 to a class. Add ctor/dtor, methods and data members.
20148 (edge_info::edge_info): Renamed from allocate_edge_info.
20149 Initialize additional members.
20150 (edge_info::~edge_info): New.
20151 (free_dom_edge_info): Delete the edge info.
20152 (record_edge_info): Use new class & associated member functions.
20153 Tighten forms for testing for edge equivalences.
20154 (record_temporary_equivalences): Iterate over the simple
20155 equivalences rather than assuming there's only one per edge.
20156 (cprop_into_successor_phis): Iterate over the simple
20157 equivalences rather than assuming there's only one per edge.
20158 (optimize_stmt): Use operand_equal_p rather than pointer
20159 equality for mini-DSE code.
20161 2017-08-28 Nathan Sidwell <nathan@acm.org>
20163 * gcc.c (execute): Fold SIGPIPE handling into switch
20164 statement. Adjust internal error message.
20166 2017-08-28 Richard Biener <rguenther@suse.de>
20169 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20170 Do nothing for removed DIEs.
20172 2017-08-28 Richard Biener <rguenther@suse.de>
20174 PR tree-optimization/81977
20175 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20178 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20181 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20182 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20183 using targetm.gen_mem_thread_fence.
20185 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20188 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20189 predicate to register_operand. Reorder operands.
20190 (*btr<mode>): Ditto.
20191 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20192 (*btr<mode>_mask): Ditto.
20194 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20196 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20197 * config/rs6000/xmmintrin.h: New file.
20198 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20200 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20203 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20204 parameter and_insn and return it.
20205 (recombine_lvx_pattern): Insert a copy to ensure availability of
20206 the base register of the copied masking operation at the point of
20207 the instruction replacement.
20208 (recombine_stvx_pattern): Likewise.
20210 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20212 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20213 undocumented switches.
20214 (-mpower9-dform-vector): Likewise.
20215 (-mpower9-dform): Likewise.
20216 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20217 comments to delete references to -mpower9-dform* switches.
20218 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20219 Delete reference to -mpower9-dform* switches, test for
20220 -mpower9-vector instead.
20221 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20222 (OTHER_P9_VECTOR_MASKS): Likewise.
20223 (POWERPC_MASKS): Likewise.
20224 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20225 tests against -mpower9-dform* to -mpower9-vector. Delete code
20226 that checked for -mpower9-dform* consistancy with other options.
20227 Add test for -mpower9-misc to enable other power9 switches.
20228 (rs6000_init_hard_regno_mode_ok): Likewise.
20229 (rs6000_option_override_internal): Likewise.
20230 (rs6000_emit_prologue): Likewise.
20231 (rs6000_emit_epilogue): Likewise.
20232 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20233 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20234 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20236 (emit_fusion_p9_store): Likewise.
20237 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20238 resetting these macros if the assembler does not support ISA 3.0
20240 (TARGET_P9_DFORM_VECTOR): Likewise.
20241 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20242 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20244 2017-08-25 Alan Modra <amodra@gmail.com>
20246 PR rtl-optimization/81747
20247 * cse.c (cse_extended_basic_block): Don't attempt to record
20248 equivalences for degenerate conditional branches that branch
20249 to their fall-through.
20251 2017-08-24 Martin Sebor <msebor@redhat.com>
20253 PR middle-end/81908
20254 * gimple-fold.c (size_must_be_zero_p): New function.
20255 (gimple_fold_builtin_memory_op): Call it.
20257 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20259 * config/rs6000/mm_malloc.h: New file.
20261 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20263 PR tree-optimization/81913
20264 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20265 analysis when either IVs in condition can wrap.
20267 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20269 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20270 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20272 2017-08-24 Richard Biener <rguenther@suse.de>
20275 * targhooks.c (default_target_can_inline_p): Properly
20276 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20277 is present and always compare.
20278 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20279 infer -mfpmath=sse from TARGET_SSE_P.
20280 (ix86_can_inline_p): Properly use target_option_default_node when
20281 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20283 2017-08-24 Richard Biener <rguenther@suse.de>
20286 * dwarf2out.c (output_die): Handle flag_generate_offload like
20288 (output_comp_unit): Likewise.
20289 (gen_array_type_die): Likewise.
20290 (dwarf2out_early_finish): Likewise.
20291 (note_variable_value_in_expr): Likewise.
20292 (dwarf2out_finish): Likewise. Adjust assert.
20293 * cgraphunit.c (symbol_table::compile): Move setting of
20294 flag_generate_offload earlier ...
20295 (symbol_table::finalize_compilation_unit): ... here, before
20296 early debug finalization.
20298 2017-08-24 Richard Biener <rguenther@suse.de>
20300 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20301 and ipa-fnsummary.h.
20302 (ix86_can_inline_p): When ix86_fpmath flags do not match
20303 check whether the callee uses FP math at all.
20305 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20307 PR middle-end/81931
20308 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20309 instead of TYPE_PRECISION.
20311 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20312 Alan Hayward <alan.hayward@arm.com>
20313 David Sherwood <david.sherwood@arm.com>
20315 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20317 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20319 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20321 (ix86_frame::stack_realign_allocate): New field.
20322 (struct machine_frame_state): Modify comments.
20323 (machine_frame_state::sp_realigned_fp_end): New field.
20324 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20325 layout calculation.
20326 (sp_valid_at): Add assertion to assure no attempt to access invalid
20327 offset of a realigned stack.
20328 (fp_valid_at): Likewise.
20329 (choose_baseaddr): Modify comments.
20330 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20331 ix86_expand_prologue.
20332 (ix86_expand_prologue): Modify stack realignment and allocation.
20333 (ix86_expand_epilogue): Modify comments.
20334 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20335 avx2_runtime, avx512f, and avx512f_runtime.
20337 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20339 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20340 (mstackrealign): Do not init to -1.
20341 * config/i386/i386.c (ix86_option_override_internal):
20342 Check opts_set, not opts when setting default value of
20343 opts->x_ix86_force_align_arg_pointer.
20345 2017-08-23 Richard Biener <rguenther@suse.de>
20347 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20348 lang_hooks.decl_printable_name.
20349 * print-rtl-function.c (print_rtx_function): Likewise.
20350 * tree-pretty-print.c (dump_function_header): Likewise.
20352 2017-08-23 Richard Biener <rguenther@suse.de>
20355 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20356 -g0 at compile-time.
20358 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20360 PR middle-end/19706
20361 * doc/sourcebuild.texi (Other hardware attributes):
20364 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20366 PR middle-end/19706
20367 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20368 Fix single-use check.
20370 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20372 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20374 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20376 * doc/install.texi: Modify to add more details on running selected
20379 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20381 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20382 is combined with -mabi=ms.
20383 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20386 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20388 PR tree-optimization/81488
20389 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20390 and cached_basis fields.
20391 (MAX_SPREAD): New constant.
20392 (alloc_cand_and_find_basis): Initialize new fields.
20393 (clear_visited): New function.
20394 (create_phi_basis_1): Rename from create_phi_basis, set visited
20395 and cached_basis fields.
20396 (create_phi_basis): New wrapper function.
20397 (phi_add_costs_1): Rename from phi_add_costs, add spread
20398 parameter, set visited field, short-circuit when limits reached.
20399 (phi_add_costs): New wrapper function.
20400 (record_phi_increments_1): Rename from record_phi_increments, set
20402 (record_phi_increments): New wrapper function.
20403 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20404 (phi_incr_cost): New wrapper function.
20405 (all_phi_incrs_profitable_1): Rename from
20406 all_phi_incrs_profitable, set visited field.
20407 (all_phi_incrs_profitable): New wrapper function.
20409 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20410 Alan Hayward <alan.hayward@arm.com>
20411 David Sherwood <david.sherwood@arm.com>
20413 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20414 that takes the outer and inner modes.
20415 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20416 comparison as the canonical test for a paradoxical subreg.
20417 * combine.c (simplify_set): Use paradoxical_subreg_p.
20418 (make_extraction): Likewise.
20419 (force_to_mode): Likewise.
20420 (rtx_equal_for_field_assignment_p): Likewise.
20421 (gen_lowpart_for_combine): Likewise.
20422 (simplify_comparison): Likewise.
20423 * cse.c (equiv_constant): Likewise.
20424 * expmed.c (store_bit_field_1): Likewise.
20425 * final.c (alter_subreg): Likewise.
20426 * fwprop.c (propagate_rtx): Likewise.
20427 (forward_propagate_subreg): Likewise.
20428 * ira-conflicts.c (ira_build_conflicts): Likewise.
20429 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20430 * lra-constraints.c (curr_insn_transform): Likewise.
20431 (split_reg): Likewise.
20432 * lra-eliminations.c (move_plus_up): Likewise.
20433 (lra_eliminate_regs_1): Likewise.
20434 * recog.c (general_operand): Likewise.
20435 * ree.c (combine_reaching_defs): Likewise.
20436 * reload.c (push_reload): Likewise.
20437 (find_reloads): Likewise.
20438 * reload1.c (elimination_effects): Likewise.
20439 (compute_reload_subreg_offset): Likewise.
20440 (choose_reload_regs): Likewise.
20441 * rtlanal.c (subreg_lsb_1): Likewise.
20442 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20443 (simplify_subreg): Likewise.
20444 * var-tracking.c (track_loc_p): Likewise.
20445 * emit-rtl.c (byte_lowpart_offset): Likewise.
20446 (paradoxical_subreg_p): Delete out-of-line definition.
20448 2017-08-22 Jeff Law <law@redhat.com>
20450 PR tree-optimization/81741
20451 PR tree-optimization/71947
20452 * tree-ssa-dom.c: Include tree-inline.h.
20453 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20454 equivalences if one is more expensive to compute than the other.
20455 * tree-ssa-scopedtables.h (class const_or_copies): Make
20456 record_const_or_copy_raw method private.
20457 (class avail_exprs_stack): New method simplify_binary_operation.
20458 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20459 avail_exprs_stack::simplify_binary_operation as needed.
20460 (avail_exprs_stack::simplify_binary_operation): New function.
20462 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20464 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20465 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20466 (DOT_SYMBOLS): Likewise.
20467 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20468 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20469 (RS6000_ABI_NAME): Likewise.
20470 (TARGET_CMODEL): Likewise.
20471 (TOC_SECTION_ASM_OP): Likewise.
20472 (SET_CMODEL): New macro.
20473 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20475 2017-08-22 Richard Biener <rguenther@suse.de>
20477 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20478 to allow for free-lang-data replacements similar to verify_type_variant.
20480 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20481 Alan Hayward <alan.hayward@arm.com>
20482 David Sherwood <david.sherwood@arm.com>
20484 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20485 VOIDmode for the LABEL_REF.
20487 2017-08-22 Richard Biener <rguenther@suse.de>
20489 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20491 2017-08-22 Richard Biener <rguenther@suse.de>
20493 * common.opt (feliminate-dwarf2-dups): Ignore.
20494 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20495 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20496 same_die_p_wrap, compute_section_prefix,
20497 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20498 (comdat_symbol_id, comdat_symbol_number): Likewise.
20499 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20501 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20502 (output_die): Mark unreachable path unreachable.
20503 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20504 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20505 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20506 (dwarf2out_early_finish): Likewise.
20508 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20510 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20512 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20515 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20516 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20517 (avr_attribute_table) <io, io_low, address>: Initialize
20518 .decl_required with true.
20520 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20522 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20523 undocumented debugging options.
20524 (-mvsx-scalar-double): Likewise.
20525 (-mallow-df-permute): Likewise.
20526 (-mvectorize-builtins): Likewise.
20527 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20528 (rs6000_builtin_vectorized_function): Likewise.
20529 (rs6000_builtin_md_vectorized_function): Likewise.
20530 (rs6000_opt_vars): Likewise.
20532 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20535 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20536 (*btrq_imm): Rename from *btrq.
20537 (*btcq_imm): Rename from *btcq.
20538 (btsc): New code attribute.
20539 (*<btsc><mode>): New insn pattern.
20540 (*btr<mode>): Ditto.
20541 (*<btsc><mode>_mask): New insn_and_split pattern.
20542 (*btr<mode>_mask): Ditto.
20544 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20545 Alan Hayward <alan.hayward@arm.com>
20546 David Sherwood <david.sherwood@arm.com>
20548 * function.c (pad_below): Simplify padding calculation.
20550 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20551 Alan Hayward <alan.hayward@arm.com>
20552 David Sherwood <david.sherwood@arm.com>
20554 * target.def (function_prologue): Remove frame size argument.
20555 (function_epilogue): Likewise.
20556 * doc/tm.texi: Regenerate.
20557 * final.c (final_start_function): Update call to function_prologue.
20558 (final_end_function): Update call to function_epilogue.
20559 (default_function_pro_epilogue): Remove frame size argument.
20560 * output.h (default_function_pro_epilogue): Likewise.
20561 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20562 (arm_output_function_prologue): Likewise.
20563 * config/frv/frv.c (frv_function_prologue): Likewise.
20564 (frv_function_epilogue): Likewise.
20565 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20566 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20567 (ia64_output_function_epilogue): Likewise.
20568 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20569 (m32r_output_function_epilogue): Likewise.
20570 * config/microblaze/microblaze.c (microblaze_function_prologue)
20571 (microblaze_function_epilogue): Likewise.
20572 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20573 (mips_output_function_epilogue): Likewise.
20574 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20575 (mmix_target_asm_function_epilogue): Likewise.
20576 * config/msp430/msp430.c (msp430_start_function): Likewise.
20577 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20578 (nds32_asm_function_epilogue): Likewise.
20579 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20580 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20581 (pa_output_function_epilogue): Likewise.
20582 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20583 (rs6000_output_function_epilogue): Likewise.
20584 * config/rl78/rl78.c (rl78_start_function): Likewise.
20585 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20586 (rs6000_output_function_epilogue): Likewise.
20587 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20588 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20589 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20590 (sparc_asm_function_epilogue): Likewise.
20592 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20594 * tree.h (type_has_mode_precision_p): New function.
20595 * convert.c (convert_to_integer_1): Use it.
20596 * expr.c (expand_expr_real_2): Likewise.
20597 (expand_expr_real_1): Likewise.
20598 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20599 * match.pd: Likewise.
20600 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20601 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20602 * tree-tailcall.c (process_assignment): Likewise.
20603 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20604 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20605 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20606 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20607 (vectorizable_assignment): Likewise.
20608 (vectorizable_shift): Likewise.
20609 (vectorizable_operation): Likewise.
20610 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20612 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20614 * match.pd: Add pow (C, x) simplification.
20616 2017-08-21 Richard Biener <rguenther@suse.de>
20618 PR tree-optimization/81900
20619 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20620 for blocks with abnormal predecessors.
20621 (compute_antic): Do not set visited flag prematurely.
20623 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20626 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20628 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20630 * stor-layout.h (vector_type_mode): Move to...
20631 * tree.h (vector_type_mode): ...here.
20632 * stor-layout.c (vector_type_mode): Move to...
20633 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20635 2017-08-21 Richard Biener <rguenther@suse.de>
20637 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20638 register_external_die hooks.
20639 (debug_false_tree_charstarstar_uhwistar): Declare.
20640 (debug_nothing_tree_charstar_uhwi): Likewise.
20641 * debug.c (do_nothing_debug_hooks): Adjust.
20642 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20643 (debug_nothing_tree_charstar_uhwi): Likewise.
20644 * dbxout.c (dbx_debug_hooks): Adjust.
20645 (xcoff_debug_hooks): Likewise.
20646 * sdbout.c (sdb_debug_hooks): Likewise.
20647 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20648 * dwarf2out.c (macinfo_label_base): New global.
20649 (dwarf2out_register_external_die): New function for the
20650 register_external_die hook.
20651 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20652 (dwarf2_debug_hooks): Use them.
20653 (dwarf2_lineno_debug_hooks): Adjust.
20654 (struct die_struct): Add with_offset flag.
20655 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20656 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20657 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20658 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20659 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20660 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20661 defining section names for the early LTO debug variants.
20662 (reset_indirect_string): New helper.
20663 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20664 (print_dw_val): Add support for offsetted symbol references.
20665 (get_ultimate_context): Split out from is_cxx.
20666 (is_cxx): Use get_ultimate_context.
20667 (is_fortran): Add decl overload.
20668 (compute_comp_unit_symbol): Split out worker from
20669 compute_section_prefix.
20670 (compute_section_prefix): Call compute_comp_unit_symbol and
20671 set comdat_type_p here.
20672 (output_die): Skip DIE symbol output for the LTO added one.
20673 Handle DIE symbol references with offset.
20674 (output_comp_unit): Guard section name mangling properly.
20675 For LTO debug sections emit a symbol at the section beginning
20676 which we use to refer to its DIEs.
20677 (add_abstract_origin_attribute): For DIEs registered via
20678 dwarf2out_register_external_die directly refer to the early
20679 DIE rather than indirectly through the shadow one we created.
20680 Remove obsolete call to dwarf2out_abstract_function for
20681 non-function/block origins.
20682 (gen_array_type_die): When generating early LTO debug do
20683 not emit DW_AT_string_length.
20684 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20685 late when in LTO. As suggested place a gcc_unreachable for
20686 the DECL_ABSTRACT_P case.
20687 (gen_subprogram_die): Avoid another specification DIE
20688 for early built declarations/definitions for the late LTO case.
20689 (gen_variable_die): Add type references for late duplicated VLA dies
20691 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20692 we have the abstract instance already.
20693 (process_scope_var): Adjust decl DIE contexts in LTO which
20694 first puts them in limbo.
20695 (gen_decl_die): Do not generate type DIEs late apart from
20696 types for VLAs or for decls we do not yet have a DIE. Do not
20697 call dwarf2out_abstract_function late.
20698 (dwarf2out_early_global_decl): Make sure to create DIEs
20699 for abstract instances of a decl first.
20700 (dwarf2out_late_global_decl): Adjust comment.
20701 (output_macinfo_op): With multiple macro sections use
20702 macinfo_label_base to distinguish labels.
20703 (output_macinfo): Likewise. Update macinfo_label_base.
20704 Pass in the line info label.
20705 (note_variable_value_in_expr): When generating LTO resolve
20706 all variable values here by generating DIEs as needed.
20707 (init_sections_and_labels): Add early LTO debug flag parameter
20708 and generate different sections and names if set. Add generation
20709 counter for the labels so we can have multiple of them.
20710 (reset_dies): Helper to allow DIEs to be output multiple times.
20711 (dwarf2out_finish): When outputting DIEs to the fat part of an
20712 LTO object first reset DIEs.
20713 (dwarf2out_early_finish): Output early DIEs when generating LTO.
20714 (modified_type_die): Check for decl_ultimate_origin being self
20716 (gen_type_die_with_usage): Likewise.
20717 (gen_typedef_die): Allow decl_ultimate_origin being self.
20718 (set_decl_abstract_flags): Remove.
20719 (set_block_abstract_flags): Likewise.
20720 (dwarf2out_abstract_function): Treat the early generated DIEs
20721 as the abstract copy and only add DW_AT_inline and
20722 DW_AT_artificial here and call set_decl_origin_self.
20723 If the DIE has an abstract origin don't do anything.
20724 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
20725 if we have none yet (Go fails to build one, PR78628).
20726 (variably_modified_type_p): Prevent endless recursion for Ada
20727 cyclic pointer types.
20728 * lto-streamer-in.c: Include debug.h.
20729 (dref_queue): New global.
20730 (lto_read_tree_1): Stream in DIE references.
20731 (lto_input_tree): Register DIE references.
20732 (input_function): Stream DECL_DEBUG_ARGS.
20733 * lto-streamer-out.c: Include debug.h.
20734 (lto_write_tree_1): Output DIE references.
20735 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
20736 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
20737 (output_function): Stream DECL_DEBUG_ARGS.
20738 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20739 Stream DECL_ABSTRACT_ORIGIN.
20740 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20741 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20742 DECL_CONTEXT for file-scope decls.
20743 * lto-streamer.h (struct dref_entry): Declare.
20744 (dref_queue): Likewise.
20745 * cfgexpand.c (pass_expand::execute): Do not call the
20746 outlining_inline_function hook here.
20747 * lto-wrapper.c (debug_obj): New global.
20748 (tool_cleanup): Unlink it if required.
20749 (debug_objcopy): New function.
20750 (run_gcc): Handle early debug sections in the IL files by
20751 extracting them to separate files, partially linkin them and
20752 feeding the result back as result to the linker.
20753 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20754 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20755 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20756 sections into a separate segment.
20757 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20759 (darwin_asm_dwarf_section): Likewise.
20760 (darwin_asm_output_dwarf_offset): Likewise.
20761 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20763 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20764 Alan Hayward <alan.hayward@arm.com>
20765 David Sherwood <david.sherwood@arm.com>
20767 * read-md.h (md_reader::record_potential_iterator_use): Replace
20768 pointer argument with an rtx and an index.
20769 * read-rtl.c (iterator_group::apply_iterator): Likewise.
20770 (apply_mode_iterator): Likewise.
20771 (apply_code_iterator): Likewise.
20772 (apply_int_iterator): Likewise.
20773 (apply_subst_iterator): Likewise.
20774 (record_iterator_use): Likewise.
20775 (record_attribute_use): Likewise.
20776 (md_reader::record_potential_iterator_use): Likewise. Update calls
20777 to record_iterator_use and apply_iterator.
20778 (iterator_use): Replace ptr with x and index.
20779 (attribute_use): Likewise.
20780 (apply_attribute_uses): Update calls to apply_iterator.
20781 (apply_iterators): Likewise. Update initialization of iterator_use.
20782 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
20783 and record_potential_iterator_use.
20784 (rtx_reader::read_rtx_operand): Likewise.
20786 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20787 Alan Hayward <alan.hayward@arm.com>
20788 David Sherwood <david.sherwood@arm.com>
20790 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
20793 2017-08-21 Richard Biener <rguenther@suse.de>
20795 PR middle-end/81884
20796 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
20797 at struct end conservatively when comparing common bases.
20799 2017-08-21 Richard Biener <rguenther@suse.de>
20801 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20802 (mem_ref_in_stmt): Remove.
20803 (determine_max_movement): Use ref index to get at the reference.
20804 (invariantness_dom_walker::before_dom_children): Deal with
20805 lim data already initialized.
20806 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20808 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20810 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20811 (TARGET_ISA_ROUND): Ditto.
20812 (TARGET_ROUND): Ditto.
20813 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20814 * config/i386/i386.md: Ditto.
20815 * config/i386/sse.md: Ditto.
20816 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20817 with OPTION_MASK_ISA_SSE4_1.
20819 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20822 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20823 __builtin_ia32_lzcnt_u16.
20825 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20828 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20829 (rs6000_set_current_function): Rewrite function to use it.
20831 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20834 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20835 and TYPE_WARN_IF_NOT_ALIGN.
20836 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20837 (handle_warn_if_not_align): New.
20838 (place_union_field): Call handle_warn_if_not_align.
20839 (place_field): Call handle_warn_if_not_align.
20840 Copy TYPE_WARN_IF_NOT_ALIGN.
20841 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20842 (layout_type): Likewise.
20843 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20845 (tree_decl_common): Add warn_if_not_align.
20846 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20847 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20848 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20849 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20850 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20851 * doc/extend.texi: Document warn_if_not_aligned attribute.
20852 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20854 2017-08-17 Martin Liska <mliska@suse.cz>
20857 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20858 (get_data_dependence): Use it as pointer type.
20859 (distribute_loop): Likewise.
20861 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20863 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20864 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20865 (p8_vmrgow_<mode>_direct): New define_insn.
20866 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20867 handle endianness for vmrgew and vmrgow permute patterns.
20869 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20871 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20872 * config/rs6000/rs6000-cpus.def: Remove comment.
20873 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20874 (POWERPC_MASKS): Likewise.
20875 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20876 use of TARGET_VSX_TIMODE.
20877 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20878 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20879 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20880 (rs6000_option_override_internal): Remove dead code.
20881 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20882 (rs6000_legitimize_reload_address): Likewise.
20883 (rs6000_legitimate_address_p): Likewise.
20884 (rs6000_opt_masks): Delete "vsx-timode".
20885 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20886 from function comment.
20887 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20888 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20889 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20891 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20892 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20893 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20894 (VSX_TI): Likewise.
20896 (define_peephole2): Likewise.
20898 2017-08-17 Martin Sebor <msebor@redhat.com>
20901 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20902 past the end of an array.
20903 (test_pp_format): Add test cases.
20905 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20907 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20908 missing ECF_NOTHROW flags.
20910 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20913 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20914 operands residing in integer registers.
20915 (*vsx_le_perm_load_<mode>): Likewise.
20916 (*vsx_le_perm_store_<mode>): Likewise.
20917 (define_peephole2): Add peepholes to optimize the above.
20919 2017-08-17 Marek Polacek <polacek@redhat.com>
20921 PR middle-end/81814
20922 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20923 to mimic what shorten_compare did. Change the return type to bool.
20924 (fold_cond_expr_with_comparison): Update call to
20925 operand_equal_for_comparison_p.
20926 (fold_ternary_loc): Likewise.
20928 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20930 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20932 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20934 2017-08-17 Richard Biener <rguenther@suse.de>
20936 * tree-ssa-structalias.c (solve_graph): When propagating
20937 to successors update the graphs succ edges and avoid duplicate work.
20939 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20942 * config/i386/i386.c (ix86_option_override_internal): Save target
20943 specific options after ix86_stack_protector_guard_reg was changed.
20945 2017-08-17 Richard Biener <rguenther@suse.de>
20947 PR tree-optimization/81827
20948 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20949 (new_var_info): Initialize it conservatively.
20950 (get_call_vi): Mark register vars.
20951 (new_scalar_tmp_constraint_exp): Likewise.
20952 (handle_rhs_call): Likewise.
20953 (handle_const_call): Likewise.
20954 (create_function_info_for): Likewise.
20955 (solve_constraints): Sort varinfos to separate register from
20956 non-register vars to pack points-to solution bitmaps during
20959 2017-08-17 Marek Polacek <polacek@redhat.com>
20961 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20963 2017-08-17 Richard Biener <rguenther@suse.de>
20965 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20966 to true when overflow is undefined and we saturated the result.
20968 2017-08-17 Alan Modra <amodra@gmail.com>
20971 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20972 Don't use store multiple if only one reg needs saving.
20973 (interesting_frame_related_regno): New function.
20974 (rs6000_frame_related): Don't emit frame info for regs that
20976 (rs6000_emit_epilogue): Likewise.
20978 2017-08-16 Nathan Sidwell <nathan@acm.org>
20980 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20981 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20982 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20983 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20984 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20985 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20986 (verify_type): Adjust for TYPE_BINFO move.
20987 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20988 process TYPE_BINFO directly.
20989 (hash_tree): Likewise.
20990 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20992 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20995 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20997 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20999 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21002 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21003 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21004 Add ix86_binary_operator_ok to insn constraint.
21005 (*iordi_1_bts): Ditto.
21006 (*xordi_1_btc): Ditto.
21007 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21008 Update corresponding peephole2 pattern.
21012 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21014 PR tree-optimization/81832
21015 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21016 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21018 2017-08-16 Marek Polacek <polacek@redhat.com>
21021 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21022 regarding size_zero_node.
21024 2017-08-16 Martin Liska <mliska@suse.cz>
21027 * config.gcc: Respect previously set extra_objs in case
21030 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21032 PR tree-optimization/81835
21033 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21034 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21035 not depend on the phi.
21037 2017-08-16 Alan Modra <amodra@gmail.com>
21039 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21042 2017-08-16 Alan Modra <amodra@gmail.com>
21044 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21045 (save_reg_p): ..into this. Update all callers.
21046 (first_reg_to_save): Simplify.
21048 2017-08-16 Alan Modra <amodra@gmail.com>
21050 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21053 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21057 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21058 constant count if that count is less than 32.
21060 2017-08-15 Nathan Sidwell <nathan@acm.org>
21062 * gcc.c (execute): Emit friendlier message if inferior is killed
21063 by an external cause.
21065 2017-08-15 Richard Biener <rguenther@suse.de>
21067 PR tree-optimization/81790
21068 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21069 CONSTRUCTORs from simplifying and VN.
21071 2017-08-14 Martin Sebor <msebor@redhat.com>
21073 * builtin-attrs.def: Add comments.
21075 2017-08-14 Martin Sebor <msebor@redhat.com>
21078 * doc/extend.texi (attribute nonstring): Document new attribute.
21080 2017-08-14 Martin Sebor <msebor@redhat.com>
21083 * tree-diagnostic.c (default_tree_printer): Handle %G.
21084 * gimple-pretty-print.h (percent_G_format): Declare new function.
21085 * gimple-pretty-print.c (percent_G_format): Define.
21086 * tree-pretty-print.c (percent_K_format): Add argument.
21088 2017-08-14 Martin Sebor <msebor@redhat.com>
21090 PR translation/79998
21091 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21092 Remove a stray space.
21094 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21097 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21098 (*iordi_1_bts): Ditto.
21099 (*xordi_1_btc): Ditto.
21101 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21104 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21105 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21107 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21108 quoted strings, and make more translator-friendly.
21109 (darwin_rs6000_override_options): Likewise.
21110 (rs6000_option_override_internal): Likewise.
21111 (rs6000_return_in_memory): Fix overlong line.
21112 (init_cmulative_args): Use quoted strings, and make more
21113 translator-friendly.
21114 (rs6000_pass_by_reference): Fix overlong line.
21115 (def_builtin): Use quoted strings.
21116 (altivec_expand_predicate_builtin): Use quoted strings, and make
21117 more translator-friendly.
21118 (htm_expand_builtin): Use quoted strings.
21119 (cpu_expand_builtin): Use quoted strings, and make more
21120 translator-friendly.
21121 (altivec_expand_builtin): Likewise.
21122 (paired_expand_predicate_builtin): Likewise.
21123 (rs6000_invalid_builtin): Likewise.
21124 (builtin_function_type): Use quoted strings.
21125 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21126 more translator-friendly.
21127 (rs6000_trampoline_init): Likewise.
21128 (rs6000_handle_altivec_attribute): Likewise.
21129 (rs6000_inner_target_options): Use quoted strings.
21130 (rs6000_disable_incompatible_switches): Likewise.
21131 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21132 strings, and make more translator-friendly.
21133 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21135 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21137 PR tree-optimization/81799
21138 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21139 cond_expr to simple gimple operand.
21141 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21143 PR middle-end/46932
21144 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21146 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21150 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21151 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21152 TARGET_GASISR_PROLOGUES.
21153 * config/avr/avr.c (avr_option_override): Same.
21154 (avr_pass_pre_proep::execute): Same.
21156 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21159 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21160 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21162 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21164 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21165 define_insn_and_split. Split to a memory load from 0 in
21166 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21167 using PTR mode iterator.
21168 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21169 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21170 (*add_tp_<mode>): Redefine as define_insn_and_split.
21171 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21172 address space. Merge with *add_tp_x32 using PTR mode iterator.
21173 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21174 Split to an add with a memory load from 0 in
21175 DEFAULT_TLS_SEG_REG address space.
21177 2017-08-12 Andrew Pinski <apinski@cavium.com>
21179 * config/aarch64/aarch64-option-extensions.def (rdma):
21180 Fix feature string to what Linux prints out in /proc/cpuinfo.
21182 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21185 * dwarf2out.c (modified_type_die): For C typedef types that have
21186 an ultimate origin, process the ultimate origin instead of the
21188 (gen_typedef_die): Assert that input DECLs have no ultimate
21190 (gen_type_die_with_usage): For typedef variants that have an
21191 ultimate origin, just call gen_decl_die on the original DECL.
21192 (process_scope_var): Avoid creating DIEs for local typedefs and
21193 concrete static variables.
21195 2017-08-12 Alan Modra <amodra@gmail.com>
21199 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21200 match gnu-user.h startfile.
21201 (ENDFILE_LINUX_SPEC): Similarly.
21203 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21206 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21208 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21210 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21211 * config/aarch64/aarch64.md (mov<mode>): Change.
21212 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21213 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21214 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21216 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21218 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21219 for storage order barriers.
21221 2017-08-11 Martin Liska <mliska@suse.cz>
21224 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21225 variables of void type.
21227 2017-08-11 Martin Liska <mliska@suse.cz>
21229 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21230 TARGET_SUPPORTS_ALIASES.
21231 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21232 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21233 (optimize_weakref): Likewise.
21234 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21235 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21236 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21238 2017-08-11 Martin Liska <mliska@suse.cz>
21241 * config/i386/i386.c (make_resolver_func): Do complete
21242 refactoring of the function.
21244 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21247 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21248 * config/i386/i386.c (ix86_stack_protect_guard): Use
21249 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21250 * doc/invoke.texi (x86 Options): Document
21251 -mstack-protector-guard-symbol= option.
21253 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21255 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21256 * config/i386/i386.c (ix86_split_stack_guard): New function.
21257 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21258 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21259 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21260 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21261 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21262 (split_stack_space_check): Call ix86_split_stack_guard.
21264 2017-08-10 Martin Sebor <msebor@redhat.com>
21266 * print-tree.c (print_node): Print location using the established
21268 Replace spaces with colons.
21269 (debug_raw, debug): Ditto.
21271 2017-08-10 Martin Sebor <msebor@redhat.com>
21274 * pretty-print.c (pp_format): Correct the handling of %s precision.
21276 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21279 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21281 (ix86_finalize_stack_frame_flags): This. Also clear
21282 frame_pointer_needed if -fno-omit-frame-pointer is used without
21284 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21285 with ix86_finalize_stack_frame_flags.
21286 (ix86_expand_epilogue): Likewise.
21287 (ix86_expand_split_stack_prologue): Likewise.
21288 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21290 2017-08-10 Martin Liska <mliska@suse.cz>
21293 * c-attribs.c (handle_target_attribute):
21294 Report warning for an empty string argument of target attribute.
21296 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21299 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21301 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21302 or DECL_NONLOCAL labels.
21303 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21304 or DECL_NONLOCAL labels here.
21306 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21308 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21309 to indicate when early gimple folding has been disabled.
21310 (rs6000_gimple_fold_builtin): Add debug content.
21311 (rs6000_invalid_builtin): Fix whitespace.
21312 (rs6000_expand_builtin): Fix whitespace.
21313 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21315 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21318 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21319 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21321 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21323 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21324 (qdf24xx): Likewise.
21325 * config/aarch64/aarch64.md: Include falkor.md.
21326 * config/aarch64/falkor.md: New.
21328 2017-08-09 Marek Polacek <polacek@redhat.com>
21331 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21332 * diagnostic.c (emit_diagnostic): Add a comment.
21333 (emit_diagnostic_valist): New function.
21335 2017-08-09 Marek Polacek <polacek@redhat.com>
21338 * input.c (make_location): New overload.
21339 * input.h (make_location): Declare.
21341 2017-08-08 Alan Modra <amodra@gmail.com>
21342 H.J. Lu <hongjiu.lu@intel.com>
21345 * gcc.c (NO_PIE_SPEC): Delete.
21346 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21348 (LINK_PIE_SPEC): ..to here.
21349 (LINK_COMMAND_SPEC): Support -no-pie.
21350 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21351 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21353 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21354 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21355 (ENDFILE_CRTEND_SPEC): Similarly.
21357 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21360 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21361 (mstack-protector-guard-offset=): Ditto.
21362 * config/i386/i386.c (ix86_option_override): Handle
21363 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21365 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21366 ix86_stack_protect_guard_offset variables.
21367 (TARGET_STACK_PROTECT_GUARD): Always define.
21368 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21369 and -mstack-protector-guard-offset= options.
21371 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21373 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21374 boundary case for the last candidate.
21376 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21378 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21379 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21381 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21383 PR middle-end/19706
21384 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21385 * config/aarch64/aarch64-builtins.c
21386 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21387 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21388 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21390 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21391 Andrew Pinski <pinskia@gmail.com>
21393 PR middle-end/19706
21394 * internal-fn.def (XORSIGN): New.
21395 * optabs.def (xorsign_optab): New.
21396 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21397 (convert_expand_mult_copysign): New.
21398 (pass_optimize_widening_mul::execute): Call
21399 convert_expand_mult_copysign.
21401 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21403 PR tree-optimization/81354
21404 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21405 Insert on edges rather than explicitly creating landing pads.
21406 (analyze_candidates_and_replace): Commit edge inserts.
21408 2017-08-08 Richard Biener <rguenther@suse.de>
21410 PR middle-end/81719
21411 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21412 (expand_simple_operations): Also look through ADDR_EXPRs with
21413 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21415 2017-08-08 Richard Biener <rguenther@suse.de>
21417 PR tree-optimization/81723
21418 * tree-vect-slp.c (struct bst_traits): New hash traits.
21419 (bst_fail): New global.
21420 (vect_build_slp_tree_2): New worker, split out from ...
21421 (vect_build_slp_tree): ... this now wrapping it with using
21422 bst_fail set to cache SLP tree build fails. Properly handle
21424 (vect_analyze_slp_instance): Allocate and free bst_fail.
21426 2017-08-08 Martin Liska <mliska@suse.cz>
21429 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21430 LABEL_DECLs that can be from a different function.
21432 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21434 PR tree-optimization/81744
21435 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21436 loop's number of iterations.
21438 2017-08-08 Martin Liska <mliska@suse.cz>
21440 * asan.c: Include header files.
21441 * attribs.c (build_decl_attribute_variant): New function moved
21443 (build_type_attribute_qual_variant): Likewise.
21444 (cmp_attrib_identifiers): Likewise.
21445 (simple_cst_list_equal): Likewise.
21446 (omp_declare_simd_clauses_equal): Likewise.
21447 (attribute_value_equal): Likewise.
21448 (comp_type_attributes): Likewise.
21449 (build_type_attribute_variant): Likewise.
21450 (lookup_ident_attribute): Likewise.
21451 (remove_attribute): Likewise.
21452 (merge_attributes): Likewise.
21453 (merge_type_attributes): Likewise.
21454 (merge_decl_attributes): Likewise.
21455 (merge_dllimport_decl_attributes): Likewise.
21456 (handle_dll_attribute): Likewise.
21457 (attribute_list_equal): Likewise.
21458 (attribute_list_contained): Likewise.
21459 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21460 (lookup_attribute_by_prefix): Likewise.
21461 * bb-reorder.c: Include header files.
21462 * builtins.c: Likewise.
21463 * calls.c: Likewise.
21464 * cfgexpand.c: Likewise.
21465 * cgraph.c: Likewise.
21466 * cgraphunit.c: Likewise.
21467 * convert.c: Likewise.
21468 * dwarf2out.c: Likewise.
21469 * final.c: Likewise.
21470 * fold-const.c: Likewise.
21471 * function.c: Likewise.
21472 * gimple-expr.c: Likewise.
21473 * gimple-fold.c: Likewise.
21474 * gimple-pretty-print.c: Likewise.
21475 * gimple.c: Likewise.
21476 * gimplify.c: Likewise.
21477 * hsa-common.c: Likewise.
21478 * hsa-gen.c: Likewise.
21479 * internal-fn.c: Likewise.
21480 * ipa-chkp.c: Likewise.
21481 * ipa-cp.c: Likewise.
21482 * ipa-devirt.c: Likewise.
21483 * ipa-fnsummary.c: Likewise.
21484 * ipa-inline.c: Likewise.
21485 * ipa-visibility.c: Likewise.
21487 * lto-cgraph.c: Likewise.
21488 * omp-expand.c: Likewise.
21489 * omp-general.c: Likewise.
21490 * omp-low.c: Likewise.
21491 * omp-offload.c: Likewise.
21492 * omp-simd-clone.c: Likewise.
21493 * opts-global.c: Likewise.
21494 * passes.c: Likewise.
21495 * predict.c: Likewise.
21496 * sancov.c: Likewise.
21497 * sanopt.c: Likewise.
21498 * symtab.c: Likewise.
21499 * toplev.c: Likewise.
21500 * trans-mem.c: Likewise.
21501 * tree-chkp.c: Likewise.
21502 * tree-eh.c: Likewise.
21503 * tree-into-ssa.c: Likewise.
21504 * tree-object-size.c: Likewise.
21505 * tree-parloops.c: Likewise.
21506 * tree-profile.c: Likewise.
21507 * tree-ssa-ccp.c: Likewise.
21508 * tree-ssa-live.c: Likewise.
21509 * tree-ssa-loop.c: Likewise.
21510 * tree-ssa-sccvn.c: Likewise.
21511 * tree-ssa-structalias.c: Likewise.
21512 * tree-ssa.c: Likewise.
21513 * tree-streamer-in.c: Likewise.
21514 * tree-vectorizer.c: Likewise.
21515 * tree-vrp.c: Likewise.
21516 * tsan.c: Likewise.
21517 * ubsan.c: Likewise.
21518 * varasm.c: Likewise.
21519 * varpool.c: Likewise.
21520 * tree.c: Remove functions moved to attribs.[ch].
21521 * tree.h: Likewise.
21522 * config/aarch64/aarch64.c: Add attrs.h header file.
21523 * config/alpha/alpha.c: Likewise.
21524 * config/arc/arc.c: Likewise.
21525 * config/arm/arm.c: Likewise.
21526 * config/avr/avr.c: Likewise.
21527 * config/bfin/bfin.c: Likewise.
21528 * config/c6x/c6x.c: Likewise.
21529 * config/cr16/cr16.c: Likewise.
21530 * config/cris/cris.c: Likewise.
21531 * config/darwin.c: Likewise.
21532 * config/epiphany/epiphany.c: Likewise.
21533 * config/fr30/fr30.c: Likewise.
21534 * config/frv/frv.c: Likewise.
21535 * config/ft32/ft32.c: Likewise.
21536 * config/h8300/h8300.c: Likewise.
21537 * config/i386/winnt.c: Likewise.
21538 * config/ia64/ia64.c: Likewise.
21539 * config/iq2000/iq2000.c: Likewise.
21540 * config/lm32/lm32.c: Likewise.
21541 * config/m32c/m32c.c: Likewise.
21542 * config/m32r/m32r.c: Likewise.
21543 * config/m68k/m68k.c: Likewise.
21544 * config/mcore/mcore.c: Likewise.
21545 * config/microblaze/microblaze.c: Likewise.
21546 * config/mips/mips.c: Likewise.
21547 * config/mmix/mmix.c: Likewise.
21548 * config/mn10300/mn10300.c: Likewise.
21549 * config/moxie/moxie.c: Likewise.
21550 * config/msp430/msp430.c: Likewise.
21551 * config/nds32/nds32-isr.c: Likewise.
21552 * config/nds32/nds32.c: Likewise.
21553 * config/nios2/nios2.c: Likewise.
21554 * config/nvptx/nvptx.c: Likewise.
21555 * config/pa/pa.c: Likewise.
21556 * config/pdp11/pdp11.c: Likewise.
21557 * config/powerpcspe/powerpcspe.c: Likewise.
21558 * config/riscv/riscv.c: Likewise.
21559 * config/rl78/rl78.c: Likewise.
21560 * config/rx/rx.c: Likewise.
21561 * config/s390/s390.c: Likewise.
21562 * config/sh/sh.c: Likewise.
21563 * config/sol2.c: Likewise.
21564 * config/sparc/sparc.c: Likewise.
21565 * config/spu/spu.c: Likewise.
21566 * config/stormy16/stormy16.c: Likewise.
21567 * config/tilegx/tilegx.c: Likewise.
21568 * config/tilepro/tilepro.c: Likewise.
21569 * config/v850/v850.c: Likewise.
21570 * config/vax/vax.c: Likewise.
21571 * config/visium/visium.c: Likewise.
21572 * config/xtensa/xtensa.c: Likewise.
21574 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21577 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21578 constraints since the -mupper-regs-* switches have been
21580 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21581 into a vector from a double word element that was extracted from
21582 another vector, and eliminate extra XXPERMDI instructions.
21583 (vsx_concat_<mode>_2): Likewise.
21584 (vsx_concat_<mode>_3): Likewise.
21585 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21586 concat to allow optimizing inserts from previous extracts.
21588 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21590 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21591 memory reference to a SSP offset in TLS address space.
21592 (ix86_print_operand) <case '@'>: Remove.
21593 (ix86_print_operand_punct_valid_p): Remove '@' code.
21594 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21595 UNSPEC_SP_TLS_TEST.
21596 (stack_tls_protect_set_<mode>): Remove.
21597 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21598 (stack_tls_protect_test_<mode>): Remove.
21599 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21601 2017-08-07 Olivier Hainque <hainque@adacore.com>
21604 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21606 2017-08-07 Douglas Rupp <rupp@adacore.com>
21608 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21609 variable was referenced as multidir in command.
21611 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21614 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21617 2017-08-07 Martin Liska <mliska@suse.cz>
21619 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21620 * config/rl78/rl78.c: Add include of attribs.h.
21621 * config/sh/sh.c: Likewise.
21622 * config/v850/v850.c: Likewise.
21624 2017-08-07 Tom de Vries <tom@codesourcery.com>
21626 PR middle-end/78266
21627 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21629 2017-08-07 Martin Liska <mliska@suse.cz>
21631 * config/mips/mips.c: Include attribs.h.
21633 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21636 * doc/invoke.texi: Document change in behvaior for -Ofast for
21639 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21641 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21643 (aarch64_pop_regs): Likewise.
21644 (aarch64_gen_load_pair): Likewise.
21645 (aarch64_save_callee_saves): Likewise.
21646 (aarch64_restore_callee_saves): Likewise.
21648 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21650 * config/i386/i386.c: Revert the last change.
21652 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21655 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21657 (ix86_finalize_stack_frame_flags): This. Also clear
21658 frame_pointer_needed if -fno-omit-frame-pointer is used without
21660 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21661 with ix86_finalize_stack_frame_flags.
21662 (ix86_expand_epilogue): Likewise.
21663 (ix86_expand_split_stack_prologue): Likewise.
21665 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21668 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21669 to P_AES for Westmere.
21671 2017-08-07 Jonathan Yong <10walls@gmail.com>
21673 * config/i386/mingw.opt (fset-stack-executable): Removed.
21674 * config/i386/cygming.opt (fset-stack-executable): Moved
21676 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21678 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21680 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21682 2017-08-07 Marek Polacek <polacek@redhat.com>
21684 PR middle-end/81737
21685 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21687 2017-08-07 Martin Liska <mliska@suse.cz>
21689 * attribs.h (canonicalize_attr_name): New function.
21690 (cmp_attribs): Move from c-format.c and adjusted.
21691 (is_attribute_p): Moved from tree.h.
21692 * tree-inline.c: Add new includes.
21693 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21694 (private_is_attribute_p): Remove.
21695 (private_lookup_attribute): Likewise.
21696 (private_lookup_attribute_by_prefix): Simplify.
21697 (remove_attribute): Use is_attribute_p.
21698 * tree.h: Remove removed declarations.
21700 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21702 PR middle-end/81698
21703 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21704 instead of computing it in the function. Formatting fix.
21705 (expand_case): Don't rely on default_edge being the first edge,
21706 clear it if removing it, pass default_edge to
21707 emit_case_dispatch_table.
21708 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21711 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
21713 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
21714 insn in the function, emit NOP after the insn.
21716 2017-08-06 Tom de Vries <tom@codesourcery.com>
21718 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
21721 2017-08-06 Tom de Vries <tom@codesourcery.com>
21723 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
21726 2017-08-04 Yury Gribov <tetra2005@gmail.com>
21728 PR tree-optimization/57371
21729 * match.pd: New pattern.
21731 2017-08-04 Marek Polacek <polacek@redhat.com>
21733 PR middle-end/81695
21734 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
21735 perform the computation in offset_int.
21737 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21739 PR tree-optimization/81136
21740 * tree-vectorizer.h: Include tree-hash-traits.h.
21741 (vec_base_alignments): New typedef.
21742 (vec_info): Add a base_alignments field.
21743 (vect_record_base_alignments): Declare.
21744 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21746 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21747 (create_data_ref): Add an is_conditional_in_stmt argument.
21748 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21749 the is_conditional_in_stmt field.
21750 (data_ref_loc): Add an is_conditional_in_stmt field.
21751 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21752 (find_data_references_in_stmt): Update call to create_data_ref.
21753 (graphite_find_data_references_in_stmt): Likewise.
21754 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21755 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21756 (vect_record_base_alignment): New function.
21757 (vect_record_base_alignments): Likewise.
21758 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21759 for nested statements even if we fail to compute a misalignment.
21760 Use pooled base alignments for unconditional references.
21761 (vect_find_same_alignment_drs): Compare base addresses instead
21763 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21764 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
21766 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21768 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
21769 Add an explicit name for the enum. Use auto_vec for slp_instances
21770 and grouped_stores.
21771 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
21773 (_bb_vec_info): Add a constructor and destructor.
21774 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21775 (destroy_loop_vec_info): Delete.
21776 (vect_destroy_datarefs): Likewise.
21777 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21778 (vec_info::vec_info): New function.
21779 (vec_info::~vec_info): Likewise.
21780 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
21781 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
21782 destroy_loop_vec_info.
21783 * tree-vect-loop.c (new_loop_vec_info): Replace with...
21784 (_loop_vec_info::_loop_vec_info): ...this.
21785 (destroy_loop_vec_info): Replace with...
21786 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
21787 the stmt_vec_infos. Leave handling of vec_info information to its
21788 destructor. Remove explicit vector releases.
21789 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
21790 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
21791 * tree-vect-slp.c (new_bb_vec_info): Replace with...
21792 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
21793 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
21794 (destroy_bb_vec_info): Replace with...
21795 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
21796 information to its destructor.
21797 (vect_slp_analyze_bb_1): Use new and delete instead of
21798 new_bb_vec_info and destroy_bb_vec_info.
21799 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21802 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21804 * tree-data-ref.h (subscript): Add access_fn field.
21805 (data_dependence_relation): Add could_be_independent_p.
21806 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21807 (same_access_functions): Move to tree-data-ref.c.
21808 * tree-data-ref.c (ref_contains_union_access_p): New function.
21809 (access_fn_component_p): Likewise.
21810 (access_fn_components_comparable_p): Likewise.
21811 (dr_analyze_indices): Add a reference to access_fn_component_p.
21812 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21814 (constant_access_functions): Likewise.
21815 (add_other_self_distances): Likewise.
21816 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21817 (initialize_data_dependence_relation): Use XCNEW and remove
21818 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21819 of access functions that have the same type. Allow the
21820 subsequence to end with different bases in some circumstances.
21821 Record the chosen access functions in SUB_ACCESS_FN.
21822 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21823 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21824 (subscript_dependence_tester_1): Likewise dra and drb.
21825 (build_classic_dist_vector): Update calls accordingly.
21826 (subscript_dependence_tester): Likewise.
21827 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21828 DDR_COULD_BE_INDEPENDENT_P.
21829 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21830 comp_alias_ddrs instead of may_alias_ddrs.
21831 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21833 (vect_analyze_data_ref_dependence): Use it if
21834 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21835 distance vectors if that fails.
21836 (dependence_distance_ge_vf): New function.
21837 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21838 LOOP_VINFO_MAY_ALIAS_DDRS.
21840 2017-08-04 Richard Biener <rguenther@suse.de>
21842 PR middle-end/81705
21843 * fold-const.c (fold_binary_loc): Properly restrict
21844 minus_var0 && minus_var1 case when associating undefined overflow
21847 2017-08-04 Tom de Vries <tom@codesourcery.com>
21849 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21851 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21853 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21854 Don't start diagnostic messages with a capital letter.
21855 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21857 (rs6000_invalid_builtin): Likewise.
21858 (rs6000_trampoline_init): Likewise.
21860 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21863 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21864 after setting changeable df flags.
21866 2017-08-03 Richard Biener <rguenther@suse.de>
21868 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21869 up if the use is in USE - X.
21871 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21873 * toplev.c (dumpfile.h): New include.
21874 (internal_error_reentered): New static function. Use it...
21875 (internal_error_function): ...here to handle reentered internal_error.
21877 2017-08-03 Richard Biener <rguenther@suse.de>
21879 PR middle-end/81148
21880 * fold-const.c (split_tree): Add minus_var and minus_con
21881 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21882 here but always use minus_*.
21883 (associate_trees): Assert we never associate with MINUS_EXPR
21884 and NULL first operand. Do not recurse for PLUS_EXPR operands
21885 when associating as MINUS_EXPR either.
21886 (fold_binary_loc): Track minus_var and minus_con.
21888 2017-08-03 Tom de Vries <tom@codesourcery.com>
21891 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21892 ACCEL_COMPILER, apply finish_options on
21893 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21895 2017-08-03 Tom de Vries <tom@codesourcery.com>
21898 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21899 function_entry_patch_area_size > 0.
21901 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21904 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21905 instead of 10??LU, perform unit multiplication in wide_int,
21906 don't change alloc_object_size_limit if the limit is larger
21909 PR tree-optimization/81655
21910 PR tree-optimization/81588
21911 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21912 the case when ranges[i].low and high are 1 for unsigned type with
21915 PR middle-end/81052
21916 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21917 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21919 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21921 * tree-vrp.h: Add include guard.
21923 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21926 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21927 (ud2): New insn pattern.
21928 * config/i386/i386.c (ix86_expand_epilogue):
21929 For naked functions, generate ud2 instead of trap insn.
21931 2017-08-02 Marek Polacek <polacek@redhat.com>
21934 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21936 2017-08-02 Tom de Vries <tom@codesourcery.com>
21937 Cesar Philippidis <cesar@codesourcery.com>
21939 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21940 Add missing edge probabilities.
21942 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21944 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21945 Correct endianness.
21947 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21949 PR middle-end/79499
21950 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21951 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21952 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21953 prologue_seq sequences - if any.
21955 2017-08-02 Richard Biener <rguenther@suse.de>
21957 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21958 via vectors if supported, integer extracts via punning if supported
21959 or otherwise vector extracts.
21961 2017-08-02 Richard Biener <rguenther@suse.de>
21963 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21965 (bitmap_insert_into_set): ... this.
21967 2017-08-02 Richard Biener <rguenther@suse.de>
21969 PR tree-optimization/81633
21971 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21973 PR tree-optimization/71752
21974 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21976 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21978 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21979 (machine_function::call_ms2sysv_pad_out): Remove field.
21980 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21981 (ix86_compute_frame_layout): Likewise.
21983 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21986 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21987 attribute with interrupt attribute.
21989 2017-08-01 Andrew Pinski <apinski@cavium.com>
21991 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21992 BIT_INSERT_EXPR's operand 1
21993 to see if the types precision matches.
21995 2017-08-01 Martin Liska <mliska@suse.cz>
21997 PR middle-end/70140
21998 * builtins.c (expand_builtin_memcpy_args): Remove.
21999 (expand_builtin_memcpy): Call newly added function
22000 expand_builtin_memory_copy_args.
22001 (expand_builtin_memcpy_with_bounds): Likewise.
22002 (expand_builtin_mempcpy): Remove last argument.
22003 (expand_builtin_mempcpy_with_bounds): Likewise.
22004 (expand_builtin_memory_copy_args): New function created from
22005 expand_builtin_mempcpy_args with small modifications.
22006 (expand_builtin_mempcpy_args): Remove.
22007 (expand_builtin_stpcpy): Remove unused argument.
22008 (expand_builtin): Likewise.
22009 (expand_builtin_with_bounds): Likewise.
22011 2017-08-01 Martin Liska <mliska@suse.cz>
22014 Make mempcpy more optimal (PR middle-end/70140).
22016 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22019 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22020 __builtin_vec_cmpne verify both arguments are compatible vectors
22021 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22022 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22023 move computation of aligned to after checking the argument types.
22027 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22028 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22031 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22032 Doug Rupp <rupp@adacore.com>
22033 Olivier Hainque <hainque@adacore.com>
22035 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22036 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22038 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22039 for TARGET_OS_CPP_BUILTIN.
22040 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22041 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22043 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22044 passing required abi options to the assembler for EABI configurations.
22045 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22046 of .text.hot and .text.unlikely sections for kernel modules when
22047 using ARM style exceptions.
22048 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22049 options. Add EXTRA_CC1_SPEC.
22050 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22052 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22054 (ARM_TARGET2_DWARF_FORMAT): Define.
22055 * config/arm/t-vxworks: Adjust multilib control to removal of the
22056 Diab command line options.
22058 2017-08-01 Martin Liska <mliska@suse.cz>
22060 PR gcov-profile/81561
22061 * gcov.c (unblock): Make unblocking safe as we need to preserve
22062 index correspondence of blocks and block_lists.
22064 2017-08-01 Richard Biener <rguenther@suse.de>
22066 PR tree-optimization/81181
22067 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22068 (compute_antic): ... end of iteration here.
22070 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22072 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22073 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22074 (ftree-slp-vectorize): Likewise.
22075 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22076 can no longer be set independent of flag_tree_loop_vectorize.
22077 * omp-general.c (emp_max_vf): Likewise.
22078 * opts.c (enable_fdo_optimizations): Remove references to
22079 flag_tree_vectorize, these are now implicit.
22080 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22081 and leave it for the options machinery.
22083 2017-08-01 Martin Liska <mliska@suse.cz>
22085 PR middle-end/70140
22086 * builtins.c (expand_builtin_memcpy_args): Remove.
22087 (expand_builtin_memcpy): Call newly added function
22088 expand_builtin_memory_copy_args.
22089 (expand_builtin_memcpy_with_bounds): Likewise.
22090 (expand_builtin_mempcpy): Remove last argument.
22091 (expand_builtin_mempcpy_with_bounds): Likewise.
22092 (expand_builtin_memory_copy_args): New function created from
22093 expand_builtin_mempcpy_args with small modifications.
22094 (expand_builtin_mempcpy_args): Remove.
22095 (expand_builtin_stpcpy): Remove unused argument.
22096 (expand_builtin): Likewise.
22097 (expand_builtin_with_bounds): Likewise.
22099 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22102 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22103 print "ds:" only for immediates in generic address space.
22105 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22108 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22109 (ix86_function_ok_for_sibcall): Return false for naked functions.
22111 2017-08-01 Richard Biener <rguenther@suse.de>
22113 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22114 (compute_antic): Seed worklist with exit block predecessors.
22115 * cfganal.c (dfs_find_deadend): For a cycle return the source
22116 of the edge closing it.
22118 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22120 * config/aarch64/aarch64.c
22121 (aarch64_can_const_movi_rtx_p): Move 0 check.
22123 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22125 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22126 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22128 * match.pd: Ditto in address comparison pattern.
22130 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22132 PR tree-optimization/81627
22133 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22134 closed ssa form for store-store chain.
22136 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22138 PR tree-optimization/81620
22139 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22140 for store-store chain.
22142 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22144 PR tree-optimization/81588
22145 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22146 ranges[i].in_p, invert comparison code ccode. For >/>=,
22147 swap rhs1 and rhs2 and comparison code unconditionally,
22148 for </<= don't do that. Don't swap rhs1/rhs2 again if
22149 ranges[i].in_p, instead invert comparison code ccode if
22150 opcode or oe->rank is BIT_IOR_EXPR.
22153 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22154 a direct optab to conversion optab.
22155 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22156 with GET_MODE_INNER as last argument instead of optab_handler.
22157 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22158 vector extraction if possible and optab is available.
22159 * expr.c (store_constructor): Use convert_optab_handler instead
22160 of optab_handler. Use vector initialization from smaller
22161 vectors if possible and optab is available.
22162 * tree-vect-stmts.c (vectorizable_load): Likewise.
22163 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22164 now have two modes.
22165 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22166 of vec_init from half-sized vectors with the same element mode.
22167 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22168 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22169 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22170 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22171 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22172 after mode in gen_vec_extract* calls.
22173 (vec_extract<mode>): Renamed to ...
22174 (vec_extract<mode><ssescalarmodelower>): ... this.
22175 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22176 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22177 element mode after mode in gen_vec_init* calls.
22178 (VEC_INIT_HALF_MODE): New mode iterator.
22179 (vec_init<mode>): Renamed to ...
22180 (vec_init<mode><ssescalarmodelower>): ... this.
22181 (vec_init<mode><ssehalfvecmodelower>): New expander.
22182 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22183 (vec_extractv2sfsf): ... this.
22184 (vec_initv2sf): Renamed to ...
22185 (vec_initv2sfsf): ... this.
22186 (vec_extractv2si): Renamed to ...
22187 (vec_extractv2sisi): ... this.
22188 (vec_initv2si): Renamed to ...
22189 (vec_initv2sisi): ... this.
22190 (vec_extractv4hi): Renamed to ...
22191 (vec_extractv4hihi): ... this.
22192 (vec_initv4hi): Renamed to ...
22193 (vec_initv4hihi): ... this.
22194 (vec_extractv8qi): Renamed to ...
22195 (vec_extractv8qiqi): ... this.
22196 (vec_initv8qi): Renamed to ...
22197 (vec_initv8qiqi): ... this.
22198 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22199 (vec_init<mode>): Renamed to ...
22200 (vec_init<mode><VEC_base_l>): ... this.
22201 (vec_extract<mode>): Renamed to ...
22202 (vec_extract<mode><VEC_base_l>): ... this.
22203 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22204 (vec_initv2sfsf): ... this.
22205 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22206 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22207 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22208 element mode after mode in gen_vec_init* calls.
22209 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22210 (vec_init<mode><Vel>): ... this.
22211 (vec_extract<mode>): Renamed to ...
22212 (vec_extract<mode><Vel>): ... this.
22213 * config/aarch64/iterators.md (Vel): New mode attribute.
22214 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22215 Add element mode after mode in gen_vec_extract* calls.
22216 * config/s390/vector.md (non_vec_l): New mode attribute.
22217 (vec_extract<mode>): Renamed to ...
22218 (vec_extract<mode><non_vec_l>): ... this.
22219 (vec_init<mode>): Renamed to ...
22220 (vec_init<mode><non_vec_l>): ... this.
22221 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22222 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22224 * config/arm/iterators.md (V_elem_l): New mode attribute.
22225 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22226 (vec_extract<mode><V_elem_l>): ... this.
22227 (vec_extractv2di): Renamed to ...
22228 (vec_extractv2didi): ... this.
22229 (vec_init<mode>): Renamed to ...
22230 (vec_init<mode><V_elem_l>): ... this.
22231 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22232 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22233 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22234 Add element mode after gen_vec_extract* calls.
22235 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22236 (vec_init<mode><unitmode>): ... this.
22237 (vec_extract<mode>): Renamed to ...
22238 (vec_extract<mode><unitmode>): ... this.
22239 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22240 (vec_init<mode><unitmode>): ... this.
22241 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22242 (vec_initv2sfsf): ... this.
22243 (vec_extractv2sf): Renamed to ...
22244 (vec_extractv2sfsf): ... this.
22245 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22246 Add element mode after gen_vec_extract* calls.
22247 * config/mips/mips.md (unitmode): New mode iterator.
22248 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22249 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22250 * config/spu/spu.md (inner_l): New mode attribute.
22251 (vec_init<mode>): Renamed to ...
22252 (vec_init<mode><inner_l>): ... this.
22253 (vec_extract<mode>): Renamed to ...
22254 (vec_extract<mode><inner_l>): ... this.
22255 * config/sparc/sparc.md (veltmode): New mode iterator.
22256 (vec_init<VMALL:mode>): Renamed to ...
22257 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22258 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22259 (vec_initv2sisi): ... this.
22260 (vec_initv2sf): Renamed to ...
22261 (vec_initv2sfsf): ... this.
22262 (vec_extractv2sf): Renamed to ...
22263 (vec_extractv2sfsf): ... this.
22264 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22265 (vec_init<mode>): Renamed to ...
22266 (vec_init<mode><VEC_base_l>): ... this.
22267 (vec_extract<mode>): Renamed to ...
22268 (vec_extract<mode><VEC_base_l>): ... this.
22269 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22270 (vec_initv2sfsf): ... this.
22271 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22272 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22273 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22274 gen_vec_init* calls.
22276 2017-08-01 Richard Biener <rguenther@suse.de>
22278 PR tree-optimization/81297
22279 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22280 TREE_OVERFLOW from INTEGER_CSTs.
22282 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22284 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22286 2017-07-31 Carl Love <cel@us.ibm.com>
22288 * config/rs6000/rs6000-c: Add support for built-in functions
22289 vector signed char vec_xl_be (signed long long, signed char *);
22290 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22291 vector signed int vec_xl_be (signed long long, signed int *);
22292 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22293 vector signed long long vec_xl_be (signed long long, signed long long *);
22294 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22295 vector signed short vec_xl_be (signed long long, signed short *);
22296 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22297 vector double vec_xl_be (signed long long, double *);
22298 vector float vec_xl_be (signed long long, float *);
22299 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22300 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22301 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22303 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22304 (altivec_expand_builtin): Add switch statement to call
22305 altivec_expand_xl_be for each builtin.
22306 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22307 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22308 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22309 __builtin_vsx_le_be_v16qi.
22310 * doc/extend.texi: Update the built-in documentation file for the
22311 new built-in functions.
22313 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22316 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22318 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22320 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22322 * config.gcc: Add z14.
22323 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22324 CPU model numbers for z13s and z14.
22325 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22327 * config/s390/s390-opts.h (enum processor_type): Rename
22328 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22329 * config/s390/s390.c (processor_table): Add field for CPU name to
22330 be passed to Binutils.
22331 (s390_asm_output_machine_for_arch): Use the new field in
22332 processor_table for Binutils.
22333 (s390_expand_builtin): Replace arch12 with z14.
22334 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22335 (s390_get_sched_attrmask): Likewise.
22336 (s390_get_unit_mask): Likewise.
22337 * config/s390/s390.opt: Add z14 to processor_type enum.
22339 2017-07-31 Martin Jambor <mjambor@suse.cz>
22342 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22343 regardless of optimization level.
22345 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22346 Martin Liska <mliska@suse.cz>
22348 * predict.def: Remove old comment and adjust probability.
22349 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22350 PREDICT statements.
22352 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22355 * config/i386/i386.c (ix86_function_naked): New function.
22356 (ix86_can_use_return_insn_p): Return false for naked functions.
22357 (ix86_expand_prologue): Skip prologue for naked functions.
22358 (ix86_expand_epilogue): Skip epilogue for naked functions
22359 and emit trap instruction.
22360 (ix86_warn_func_return): New function.
22361 (ix86_attribute_table): Add "naked" attribute specification.
22362 (TARGET_WARN_FUNC_RETURN): Define.
22363 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22365 2017-07-31 Martin Liska <mliska@suse.cz>
22367 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22368 (dump_gimple_bb_header): Always dump BB info.
22369 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22371 2017-07-31 Martin Liska <mliska@suse.cz>
22374 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22375 also with current_function_decl non-null equality.
22377 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22380 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22381 change type to the element type, instead add eltype variable and
22382 use it where we are interested in the element type.
22384 PR tree-optimization/81603
22385 * ipa-polymorphic-call.c
22386 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22387 offset arithmetic in offset_int, bail out if the resulting bit offset
22388 doesn't fit into shwi.
22390 2017-07-31 Martin Liska <mliska@suse.cz>
22392 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22393 (gimplify_save_expr): Fix comment.
22395 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22398 * config/i386/i386.c (ix86_function_arg): Update arguments for
22400 (ix86_compute_frame_layout): Set the initial stack offset to
22401 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22402 INCOMING_FRAME_SP_OFFSET.
22403 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22404 stack before exception handler returns.
22405 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22406 the 'ERROR_CODE' for exception handler.
22408 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22410 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22411 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22412 (ASM_OUTPUT_REG_POP): Ditto.
22413 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22414 instead of asm_fprintf to output pure string.
22416 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22418 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22419 to imported_module_or_decl hook.
22420 (debug_nothing_tree_tree_tree_bool): Remove.
22421 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22422 * debug.c (do_nothing_debug_hooks): Use
22423 debug_nothing_tree_tree_tree_bool_bool instead of
22424 debug_nothing_tree_tree_tree_bool.
22425 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22426 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22427 * sdbout.c (sdb_debug_hooks): Likewise.
22428 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22429 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22431 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22432 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22433 attribute, don't add anything.
22435 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22437 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22438 (fold_build2_stat_loc): Likewise.
22439 (fold_build3_stat_loc): Likewise.
22440 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22441 (fold_build1_loc): Remove macro.
22442 (fold_build2_loc): Likewise.
22443 (fold_build3_loc): Likewise.
22445 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22447 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22448 (gimple_build_debug_bind_source_stat): Likewise.
22449 * gimple.h (gimple_build_debug_bind): Remove macro.
22450 (gimple_build_debug_bind_source): Likewise.
22452 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22454 * bitmap.c (bitmap_alloc): Adjust.
22455 (bitmap_gc_alloc): Likewise.
22456 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22458 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22460 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22461 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22462 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22463 (bitmap_gc_alloc_stat): Likewise.
22464 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22466 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22468 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22469 * rtl.h (shallow_copy_rtx): Remove macro.
22471 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22473 * emit-rtl.c (gen_raw_REG): Adjust.
22474 * gengenrtl.c (gendef): Likewise.
22475 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22476 * rtl.h (rtx_alloc): Remove macro.
22478 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22480 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22481 (build_tree_list_stat): Likewise.
22482 * tree.h (build_tree_list): Remove macro.
22483 (build_tree_list_vec): Likewise.
22485 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22487 * tree.c (make_vector_stat): Remove _stat from name.
22488 (build_vector_stat): Likewise.
22489 * tree.h (make_vector_stat): Remove macro.
22490 (build_vector_stat): Likewise.
22492 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22494 * tree.h (build_var_debug_value): Remove prototype.
22496 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22498 * tree.c (tree_cons_stat): Remove _stat from name.
22499 * tree.h (tree_cons): Remove macro.
22501 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22503 * tree.c (build_vl_exp_stat): Remove _stat from name.
22504 * tree.h (build_vl_exp): Remove macro.
22506 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22508 * tree.c (build_decl_stat): Remove _stat from name.
22509 * tree.h (build_decl): Remove macro.
22511 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22513 * gimple.c (gimple_build_with_ops_stat): Adjust.
22514 (gimple_alloc_stat): Remove _stat from name.
22515 * gimple.h (gimple_alloc): Remove macro.
22517 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22519 * tree.c (make_tree_vec_stat): Remove _stat from name.
22520 (grow_tree_vec_stat): Likewise.
22521 * tree.h (make_tree_vec_stat): Adjust prototype.
22522 (grow_tree_vec_stat): Likewise.
22523 (make_tree_vec): Remove macro.
22524 (grow_tree_vec): Likewise.
22526 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22528 * fold-const.c (fold_build1_stat_loc): Adjust.
22529 (fold_build2_stat_loc): Likewise.
22530 (fold_build3_stat_loc): Likewise.
22531 * tree.c (build0_stat): Remove _stat from name.
22532 (build1_stat): Likewise.
22533 (build2_stat): Likewise.
22534 (build3_stat): Likewise.
22535 (build4_stat): Likewise.
22536 (build5_stat): Likewise.
22537 * tree.h (build1_loc): Remove macro, and rename _stat function
22539 (build2_loc): Likewise.
22540 (build3_loc): Likewise.
22541 (build4_loc): Likewise.
22542 (build5_loc): Likewise.
22544 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22546 * tree.c (make_int_cst_stat): Remove _stat from name.
22547 * tree.h (make_int_cst_stat): Adjust prototype.
22548 (make_int_cst): Remove macro.
22550 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22552 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22553 * tree.h (make_tree_binfo_stat): Adjust prototype.
22554 (make_tree_binfo): Remove.
22556 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22558 * tree.c (copy_node_stat): Rename to copy_node.
22559 (build_distinct_type_copy): Adjust.
22560 * tree.h (copy_node_stat): Adjust prototype.
22561 (copy_node): Remove macro.
22563 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22565 * tree.c (make_node_stat): rename to make_node.
22566 (build_tree_list_stat): Adjust.
22567 (build0_stat): Likewise.
22568 (build2_stat): Likewise.
22569 (build3_stat): Likewise.
22570 (build4_stat): Likewise.
22571 (build5_stat): Likewise.
22572 (build_decl_stat): Likewise.
22573 * tree.h (make_node_stat): Adjust prototype.
22574 (make_node): remove macro.
22576 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22578 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22579 (PPC_FEATURE2_SCV): Likewise.
22580 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22582 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22584 * config/aarch64/aarch64.c
22585 (aarch64_internal_mov_immediate): Add new special pattern.
22586 * config/aarch64/aarch64.md (*movdi_aarch64):
22587 Add reg/32bit const mov case.
22589 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22590 Richard Sandiford <richard.sandiford@linaro.org>
22592 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22593 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22594 Add integer and movi cases.
22595 (movi-split-hf-df-sf split, fp16): New.
22596 (enabled): Added TARGET_FP_F16INST.
22597 * config/aarch64/iterators.md (GPF_HF): New.
22598 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22600 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22602 * config/aarch64/aarch64.c
22603 (aarch64_simd_container_mode): Add prototype.
22604 (aarch64_expand_mov_immediate): Add HI support.
22605 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22606 (aarch64_can_const_movi_rtx_p): New.
22607 (aarch64_preferred_reload_class):
22608 Remove restrictions of using FP registers for certain SIMD operations.
22609 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22610 (aarch64_valid_floating_const): Add integer move validation.
22611 (aarch64_simd_imm_scalar_p): Remove.
22612 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22613 (aarch64_legitimate_constant_p): Expand list of supported cases.
22614 * config/aarch64/aarch64-protos.h
22615 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22616 (aarch64_reinterpret_float_as_int): New.
22617 (aarch64_simd_imm_scalar_p): Remove.
22618 * config/aarch64/constraints.md (Uvi): New.
22619 (Dd): Split into Ds and new Dd.
22620 * config/aarch64/aarch64.md (*movsi_aarch64):
22622 (*movdi_aarch64): Add SIMD mov case.
22624 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22626 * tree-predcom.c: (struct chain): Handle store-store chain in which
22627 stores for elimination only store loop invariant values.
22628 (execute_pred_commoning_chain): Ditto.
22629 (prepare_initializers_chain_store_elim): Ditto.
22630 (prepare_finalizers): Ditto.
22631 (is_inv_store_elimination_chain): New function.
22632 (initialize_root_vars_store_elim_1): New function.
22634 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22636 * tree-predcom.c: Revise general description of the pass.
22637 (enum chain_type): New enum type for store elimination.
22638 (struct chain): New field supporting store elimination.
22639 (struct component): Ditto.
22640 (dump_chain): Dump store-stores chain.
22641 (release_chain): Release resources.
22642 (split_data_refs_to_components): Compute and create component
22643 contains only stores for elimination.
22644 (get_chain_last_ref_at): New function.
22645 (make_invariant_chain): Initialization.
22646 (make_rooted_chain): Specify chain type in parameter and record it.
22647 (add_looparound_copies): Skip for store-stores chain.
22648 (determine_roots_comp): Compute type of chain and pass it to
22650 (initialize_root_vars_store_elim_2): New function.
22651 (finalize_eliminated_stores): New function.
22652 (remove_stmt): Handle store for elimination.
22653 (execute_pred_commoning_chain): Execute predictive commoning on
22654 store-store chains.
22655 (determine_unroll_factor): Skip unroll for store-stores chain.
22656 (prepare_initializers_chain_store_elim): New function.
22657 (prepare_initializers_chain): Hanlde store-store chain.
22658 (prepare_finalizers_chain, prepare_finalizers): New function.
22659 (tree_predictive_commoning_loop): Return integer value indicating
22660 if loop is unrolled or lcssa form is corrupted.
22661 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22663 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22665 * tree-predcom.c (initialize_root): Delete.
22666 (execute_pred_commoning_chain): Initialize root vars and replace
22667 reference of non-combined chain directly, rather than call above
22670 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22672 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22673 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22675 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22677 * tree-predcom.c (struct chain): New field init_seq.
22678 (release_chain): Release init_seq.
22679 (prepare_initializers_chain): Record intialization stmts in above
22681 (insert_init_seqs): New function.
22682 (tree_predictive_commoning_loop): Call insert_init_seqs.
22684 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22686 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22688 2017-07-28 Richard Biener <rguenther@suse.de>
22690 * match.pd: Remove superfluous :c.
22691 * genmatch.c (simplify::id): Add member.
22692 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22694 (current_id): New global.
22695 (dt_node::parent): Move from ...
22696 (dt_operand::parent): ... here. Add for_id member.
22697 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22698 (decision_tree::find_node): Relax order requirement when
22699 merging DT_TRUE nodes to ones inbetween the current simplify
22700 and the one we try to merge with. Add diagnostic whenever
22701 we need to enforce pattern order by not merging.
22702 (decision_tree::insert): Set current_id.
22703 (decision_tree::print_node): Dump parent node and for_id.
22704 (parser::last_id): Add member.
22705 (parser::push_simplify): Assign unique id.
22706 (parser::parser): Initialize last_id.
22708 2017-07-28 Martin Liska <mliska@suse.cz>
22711 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22712 gimple_build_debug_bind.
22714 2017-07-28 Richard Biener <rguenther@suse.de>
22716 PR tree-optimization/81502
22717 * match.pd: Add pattern combining BIT_INSERT_EXPR with
22719 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
22721 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
22722 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
22723 for BIT_FIELD_REF args.
22724 * fold-const.c (make_bit_field_ref): Likewise.
22725 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
22727 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22730 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
22731 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
22732 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
22733 Or it into SANITIZER_UNDEFINED.
22734 * ubsan.c: Include gimple-fold.h and varasm.h.
22735 (ubsan_expand_ptr_ifn): New function.
22736 (instrument_pointer_overflow): New function.
22737 (maybe_instrument_pointer_overflow): New function.
22738 (instrument_object_size): Formatting fix.
22739 (pass_ubsan::execute): Call instrument_pointer_overflow
22740 and maybe_instrument_pointer_overflow.
22741 * internal-fn.c (expand_UBSAN_PTR): New function.
22742 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22743 * sanitizer.def (__ubsan_handle_pointer_overflow,
22744 __ubsan_handle_pointer_overflow_abort): New builtins.
22745 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22746 * internal-fn.def (UBSAN_PTR): New internal function.
22747 * opts.c (sanitizer_opts): Add pointer-overflow.
22748 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22749 * fold-const.c (build_range_check): Compute pointer range check in
22750 integral type if pointer arithmetics would be needed. Formatting
22753 2017-07-28 Martin Liska <mliska@suse.cz>
22756 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22757 parameters that are of a variable-length.
22759 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22761 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22763 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22764 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22765 (CRT_CALL_STATIC_FUNCTION): Likewise.
22766 (ASM_DEFAULT_SPEC): New define.
22767 (ASM_SPEC32): Likewise.
22768 (ASM_SPEC64): Likewise.
22769 (ASM_SPEC_COMMON): Likewise.
22770 (ASM_SPEC): Likewise.
22771 (INVALID_64BIT): Likewise.
22772 (LINK_OS_DEFAULT_SPEC): Likewise.
22773 (LINK_OS_SPEC32): Likewise.
22774 (LINK_OS_SPEC64): Likewise.
22775 (POWERPC_LINUX): Likewise.
22776 (PTRDIFF_TYPE): Likewise.
22777 (RESTORE_FP_PREFIX): Likewise.
22778 (RESTORE_FP_SUFFIX): Likewise.
22779 (SAVE_FP_PREFIX): Likewise.
22780 (SAVE_FP_SUFFIX): Likewise.
22781 (SIZE_TYPE): Likewise.
22782 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
22783 (TARGET_64BIT): Likewise.
22784 (TARGET_64BIT): Likewise.
22785 (TARGET_AIX): Likewise.
22786 (WCHAR_TYPE_SIZE): Likewise.
22787 (WCHAR_TYPE): Undefine.
22788 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
22789 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
22790 (CPP_OS_RTEMS_SPEC): Delete.
22791 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
22792 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
22794 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
22796 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22798 PR tree-optimization/81578
22799 * tree-parloops.c (build_new_reduction): Bail out if
22800 reduction_code isn't one of the standard OpenMP reductions.
22801 Move the details printing after that decision.
22803 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22805 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22806 related to reload_in_progress.
22807 (splat_input_operand): Likewise.
22808 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22810 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22812 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22813 (TARGET_INSTANTIATE_DECLS): Likewise.
22814 (legitimate_indexed_address_p): Delete reload_in_progress code.
22815 (rs6000_debug_legitimate_address_p): Likewise.
22816 (rs6000_eliminate_indexed_memrefs): Likewise.
22817 (rs6000_emit_le_vsx_store): Likewise.
22818 (rs6000_emit_move_si_sf_subreg): Likewise.
22819 (rs6000_emit_move): Likewise.
22820 (register_to_reg_type): Likewise.
22821 (rs6000_pre_atomic_barrier): Likewise.
22822 (rs6000_machopic_legitimize_pic_address): Likewise.
22823 (rs6000_allocate_stack_temp): Likewise.
22824 (rs6000_address_for_fpconvert): Likewise.
22825 (rs6000_address_for_altivec): Likewise.
22826 (rs6000_secondary_memory_needed_rtx): Delete function.
22827 (rs6000_check_sdmode): Likewise.
22828 (rs6000_alloc_sdmode_stack_slot): Likewise.
22829 (rs6000_instantiate_decls): Likewise.
22830 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22831 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22832 Delete reload_in_progress.
22833 (*vec_reload_and_plus_<mptrsize>): Likewise.
22834 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22835 (vsx_div_v2di): Likewise.
22836 (vsx_udiv_v2di): Likewise.
22838 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22840 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22841 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22842 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22843 (rs6000_debug_reg_global): Delete print of LRA status.
22844 (rs6000_option_override_internal): Delete dead LRA related code.
22845 (rs6000_lra_p): Delete function.
22846 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22848 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22850 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22851 * config/riscv/rtems.h: New file.
22853 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22854 Sudakshina Das <sudi.das@arm.com>
22856 * config/aarch64/aarch64.md
22857 (define_split for and<mode>3nr_compare): Move
22858 non aarch64_logical_operand to a register.
22859 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22860 register immediate operand to a register.
22861 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22863 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22865 PR middle-end/81564
22866 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22868 2017-07-27 Richard Biener <rguenther@suse.de>
22870 PR tree-optimization/81573
22871 PR tree-optimization/81494
22872 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22873 multi defuse cycle case.
22875 2017-07-27 Richard Biener <rguenther@suse.de>
22877 PR tree-optimization/81571
22878 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22881 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22883 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22884 earlier and only if MASK_FPU is set. Adjust formatting.
22886 2017-07-27 Martin Liska <mliska@suse.cz>
22888 * opt-functions.awk: Add validation of value of Init.
22889 * optc-gen.awk: Pass new argument.
22891 2017-07-27 Martin Liska <mliska@suse.cz>
22893 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22894 Fix wrong condition.
22896 2017-07-27 Martin Liska <mliska@suse.cz>
22898 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22899 BBs and edges seen by autoFDO.
22901 2017-07-27 Richard Biener <rguenther@suse.de>
22903 PR tree-optimization/81502
22904 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22905 with incompatible but same sized type.
22906 (execute_update_addresses_taken): Likewise.
22908 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22910 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22911 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22913 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22916 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22917 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22918 Change s_operand to memory_operand.
22920 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22922 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22923 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22924 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22925 Emit instructions rather than returning an expression. Handle TFmode
22926 and KFmode by casting to TImode.
22927 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22928 (rs6000_emit_le_vsx_store): Likewise.
22929 * config/rs6000/vsx.md (VSX_TI): New iterator.
22930 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22931 (*vsx_le_undo_permute_<mode>): Likewise.
22932 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22933 emit the split sequence.
22934 (*vsx_le_perm_store_<mode>): Likewise.
22936 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22938 PR tree-optimization/81555
22939 PR tree-optimization/81556
22940 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22941 if true, force CHANGED for the recursive invocation.
22942 (reassociate_bb): Remember original length of ops array, pass
22943 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22945 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22946 attributes for noipa attribute. For naked attribute use
22947 lookup_attribute first before lookup_attribute_spec.
22948 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22950 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22951 for functions with noipa attribute.
22952 (cgraph_externally_visible_p): Return true for functions with noipa
22954 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22955 for functions with noipa attribute.
22956 * doc/extend.texi: Document noipa function attribute.
22957 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22958 also for functions with noipa attribute.
22959 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22961 2017-07-26 Andrew Pinski <apinski@cavium.com>
22963 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22964 vec_unalign_load_cost and vec_unalign_store_cost.
22966 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22968 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22969 -mvsx-small-integer option.
22970 (ISA_3_0_MASKS_IEEE): Likewise.
22971 (OTHER_VSX_VECTOR_MASKS): Likewise.
22972 (POWERPC_MASKS): Likewise.
22973 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22974 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22975 code, only testing for DImode being allowed in non-VSX floating
22977 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22978 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22980 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22981 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22982 TARGET_P8_VECTOR test.
22983 (rs6000_secondary_reload_simple_move): Likewise.
22984 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22985 since TARGET_P9_VECTOR was already tested.
22986 (rs6000_opt_masks): Remove -mvsx-small-integer.
22987 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22988 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22990 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22991 test for TARGET_VEXTRACTUB was used, and that uses
22993 (p9 extract splitter): Likewise.
22994 (vsx_extract_<mode>_di_p9): Likewise.
22995 (vsx_extract_<mode>_store_p9): Likewise.
22996 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22997 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22998 the elimination of TARGET_VSX_SMALL_INTEGER.
22999 (vsx_extract_<mode>_p8): Likewise.
23000 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23001 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23002 (vsx_set_<mode>_p9): Likewise.
23003 (vsx_set_v4sf_p9): Likewise.
23004 (vsx_set_v4sf_p9_zero): Likewise.
23005 (vsx_insert_extract_v4sf_p9): Likewise.
23006 (vsx_insert_extract_v4sf_p9_2): Likewise.
23007 * config/rs6000/rs6000.md (sign extend splitter): Change
23008 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23009 (floatsi<mode>2_lfiwax_mem): Likewise.
23010 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23011 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23012 since a test for TARGET_P9_VECTOR was used.
23013 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23014 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23015 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23016 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23017 TARGET_P8_VECTOR test.
23018 (fix_trunc<mode>si2_stfiwx): Likewise.
23019 (fix_trunc<mode>si2_internal): Likewise.
23020 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23021 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23023 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23024 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23025 TARGET_P8_VECTOR test.
23026 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23027 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23028 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23030 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23031 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23032 since a test for TARGET_P9_VECTOR was used.
23033 (splitter for loading small constants): Likewise.
23035 2017-07-26 Andrew Pinski <apinski@cavium.com>
23037 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23040 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23043 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23044 (fp_valid_at): Likewise.
23046 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23048 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23049 (qdf24xx_addrcost_table): Likewise.
23050 (cortexa57_tunings): Update to use generic_branch_cost.
23051 (cortexa72_tunings): Likewise.
23052 (cortexa73_tunings): Likewise.
23053 (qdf24xx_tunings): Likewise.
23055 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23057 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23058 (thunderx2t99_branch_cost): Likewise.
23059 (cortexa35_tunings): Update to use generic_branch_cost.
23060 (cortexa53_tunings): Likewise.
23061 (cortexa57_tunings): Likewise.
23062 (cortexa72_tunings): Likewise.
23063 (cortexa73_tunings): Likewise.
23064 (thunderx2t99_tunings): Likewise.
23066 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23068 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23069 (sparc_option_override): Honour MASK_FSMULD.
23070 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23071 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23072 * config/sparc/sparc.opt (mfsmuld): New option.
23073 * doc/invoke.texi (mfsmuld): Document option.
23075 2017-07-26 Marek Polacek <polacek@redhat.com>
23077 PR middle-end/70992
23078 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23080 2017-07-26 Richard Biener <rguenther@suse.de>
23082 * gimple-match-head.c (do_valueize): Return OP if valueize
23084 (get_def): New helper to get at the def stmt of a SSA name
23085 if valueize allows.
23086 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23087 do_valueize to get at the def stmt.
23088 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23090 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23092 PR middle-end/46932
23093 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23095 2017-07-26 Martin Liska <mliska@suse.cz>
23098 * function.c (expand_function_start): Make expansion of
23099 nonlocal_goto_save_area after parm_birth_insn.
23101 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23103 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23104 from all CPU target flags enable members.
23106 2017-07-26 Richard Biener <rguenther@suse.de>
23108 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23109 (decision_tree::gen): Make 'type' const.
23110 (write_predicate): Likewise.
23112 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23114 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23115 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23116 (rs6000_option_override_internal): Likewise.
23117 (rs6000_expand_vector_set): Likewise.
23118 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23119 (TARGET_UPPER_REGS_SF): Likewise.
23120 (TARGET_UPPER_REGS_DI): Likewise.
23121 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23122 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23123 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23124 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23125 (Splitters for DI constants in Altivec registers): Likewise.
23126 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23127 (vsx_set_v4sf_p9): Likewise.
23128 (vsx_set_v4sf_p9_zero): Likewise.
23129 (vsx_insert_extract_v4sf_p9): Likewise.
23130 (vsx_insert_extract_v4sf_p9_2): Likewise.
23132 2017-07-25 Carl Love <cel@us.ibm.com>
23134 * doc/extend.texi: Update the built-in documentation file for the
23135 existing built-in functions
23136 vector signed char vec_cnttz (vector signed char);
23137 vector unsigned char vec_cnttz (vector unsigned char);
23138 vector signed short vec_cnttz (vector signed short);
23139 vector unsigned short vec_cnttz (vector unsigned short);
23140 vector signed int vec_cnttz (vector signed int);
23141 vector unsigned int vec_cnttz (vector unsigned int);
23142 vector signed long long vec_cnttz (vector signed long long);
23143 vector unsigned long long vec_cnttz (vector unsigned long long);
23145 2017-07-25 Andrew Pinski <apinski@cavium.com>
23147 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23148 accesses where the use is for the first operand of a BIT_INSERT.
23150 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23153 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23154 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23156 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23158 * config/i386/gstabs.h: Delete.
23159 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23161 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23163 * config/i386/i386.c (ix86_decompose_address): Do not check for
23164 register RTX when looking at index_reg or base_reg.
23165 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23167 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23169 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23170 to update EH info here.
23172 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23174 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23176 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23178 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23180 2017-07-25 Torsten Duwe <duwe@suse.de>
23182 * common.opt: Introduce -fpatchable-function-entry
23183 command line option, and its variables function_entry_patch_area_size
23184 and function_entry_patch_area_start.
23185 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23186 including a two-value parser.
23187 * target.def (print_patchable_function_entry): New target hook.
23188 * targhooks.h (default_print_patchable_function_entry): New function.
23189 * targhooks.c (default_print_patchable_function_entry): Likewise.
23190 * toplev.c (process_options): Switch off IPA-RA if
23191 patchable function entries are being generated.
23192 * varasm.c (assemble_start_function): Look at the
23193 patchable-function-entry command line switch and current
23194 function attributes and maybe generate NOP instructions by
23195 calling the print_patchable_function_entry hook.
23196 * doc/extend.texi: Document patchable_function_entry attribute.
23197 * doc/invoke.texi: Document -fpatchable_function_entry
23198 command line option.
23199 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23201 * doc/tm.texi: Re-generate.
23203 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23206 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23207 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23209 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23211 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23213 2017-07-25 Richard Biener <rguenther@suse.de>
23215 PR tree-optimization/81455
23216 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23217 not walk in cycles when looking for guards.
23219 2017-07-25 Richard Biener <rguenther@suse.de>
23221 PR tree-optimization/81529
23222 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23223 when optimizing backedge uses.
23225 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23227 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23229 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23230 to dl_section_ref. On AIX, append an expression to subtract
23231 the size of the section length to dl_section_ref.
23233 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23235 * configure.ac: If any of the config.* scripts fail, exit 1.
23236 * configure: Regenerate.
23238 2017-07-25 Richard Biener <rguenther@suse.de>
23240 PR middle-end/81546
23241 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23242 of immediate uses, be more verbose on errors.
23244 2017-07-25 Richard Biener <rguenther@suse.de>
23246 PR tree-optimization/81510
23247 * tree-vect-loop.c (vect_is_simple_reduction): When the
23248 reduction stmt is not inside the loop bail out.
23250 2017-07-25 Richard Biener <rguenther@suse.de>
23252 PR tree-optimization/81303
23253 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23254 profitability check against LOOP_VINFO_NITERSM1.
23256 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23258 * domwalk.c (cmp_bb_postorder): Simplify.
23259 (sort_bbs_postorder): New function. Use it...
23260 (dom_walker::walk): ...here to optimize common cases.
23262 2017-07-25 Martin Liska <mliska@suse.cz>
23265 * ipa-visibility.c (function_and_variable_visibility): Make the
23266 redirection just on target that supports aliasing.
23267 Fix GNU coding style.
23269 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23272 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23274 * config/arm/rtems.h: Likewise.
23275 * config/bfin/rtems.h: Likewise.
23276 * config/i386/rtemself.h: Likewise.
23277 * config/lm32/rtems.h: Likewise.
23278 * config/m32c/rtems.h: Likewise.
23279 * config/m68k/rtemself.h: Likewise.
23280 * config/microblaze/rtems.h: Likewise.
23281 * config/mips/rtems.h: Likewise.
23282 * config/moxie/rtems.h: Likewise.
23283 * config/nios2/rtems.h: Likewise.
23284 * config/powerpcspe/rtems.h: Likewise.
23285 * config/rs6000/rtems.h: Likewise.
23286 * config/rtems.h: Likewise.
23287 * config/sh/rtems.h: Likewise.
23288 * config/sh/rtemself.h: Likewise.
23289 * config/sparc/rtemself.h: Likewise.
23291 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23294 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23295 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23296 * tree-ssa-structalias.c (alias_get_name): Same.
23298 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23301 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23302 instructions if no du chain is found.
23304 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23306 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23308 2017-07-25 Richard Biener <rguenther@suse.de>
23310 PR middle-end/81505
23311 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23314 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23316 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23317 upper-regs options.
23318 (ISA_2_7_MASKS_SERVER): Likewise.
23319 (ISA_3_0_MASKS_IEEE): Likewise.
23320 (OTHER_P8_VECTOR_MASKS): Likewise.
23321 (OTHER_VSX_VECTOR_MASKS): Likewise.
23322 (POWERPC_MASKS): Likewise.
23323 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23324 duplicate list of options.
23325 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23326 explicit -mupper-regs options.
23327 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23328 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23329 alias for -mupper-regs-df.
23330 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23331 (rs6000_init_hard_regno_mode_ok): Likewise.
23332 (rs6000_option_override_internal): Likewise.
23333 (rs6000_opt_masks): Likewise.
23334 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23335 options in terms of whether -mvsx or -mpower8-vector was used.
23336 (TARGET_UPPER_REGS_DI): Likewise.
23337 (TARGET_UPPER_REGS_SF): Likewise.
23338 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23339 -mupper-regs-* options.
23341 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23343 * passes.c (emergency_dump_function): Print some empty lines and a
23344 header before the RTL dump.
23346 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23348 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23350 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23353 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23354 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23356 2017-07-24 Carl Love <cel@us.ibm.com>
23358 * config/rs6000/rs6000-c.c: Add support for built-in functions
23359 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23360 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23361 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23362 vec_extract_fp_from_shortl): Add defines for the two builtins.
23363 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23364 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23366 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23367 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23368 * doc/extend.texi: Update the built-in documentation file for the
23369 new built-in function.
23371 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23374 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23376 * doc/generic.texi: Likewise.
23377 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23378 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23380 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23382 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23383 (aarch64_mls_elt_merge<mode>): Likewise.
23385 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23387 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23388 having __cxa_atexit.
23390 2017-07-23 Michael Collison <michael.collison@arm.com>
23392 * config/arm/arm.c (arm_option_override): Deprecate
23393 use of -mstructure-size-boundary.
23394 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23395 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23397 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23400 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23401 Reduce cost estimate for direct moves.
23403 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23406 * config/i386/i386.c (ix86_option_override_internal): Disable
23407 BMI, BMI2 and TBM instructions for -m16.
23409 2017-07-21 Carl Love <cel@us.ibm.com>
23411 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23412 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23413 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23414 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23415 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23416 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23417 VMULOSW): New enum "unspec" values.
23418 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23419 altivec_vmulosw): New patterns.
23420 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23421 VMULOSW): Add definitions.
23423 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23425 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23426 (qdf24xx): Likewise.
23427 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23428 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23429 (AARCH64_FL_V8_1): Renumber.
23430 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23431 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23432 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23433 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23434 rdma to feature modifiers list.
23436 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23438 PR middle-end/56727
23439 * ipa-visibility (function_and_variable_visibility): Convert
23440 recursive PLT call to direct call if appropriate.
23442 2017-07-21 Andrew Pinski <apinski@cavium.com>
23444 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23445 operand 1 to see if the types precision matches.
23446 * fold-const.c (operand_equal_p): Likewise.
23448 2017-07-21 Richard Biener <rguenther@suse.de>
23450 PR tree-optimization/81303
23451 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23452 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23453 (vect_peeling_hash_get_lowest_cost): Adjust.
23454 (vect_enhance_data_refs_alignment): Likewise. Use
23455 vect_get_peeling_costs_all_drs to compute the penalty for no
23456 peeling to match up costs.
23458 2017-07-21 Richard Biener <rguenther@suse.de>
23460 PR tree-optimization/81500
23461 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23462 we didn't identify a reduction path.
23464 2017-07-21 Tom de Vries <tom@codesourcery.com>
23465 Cesar Philippidis <cesar@codesourcery.com>
23467 PR gcov-profile/81442
23468 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23471 2017-07-21 Tom de Vries <tom@codesourcery.com>
23474 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23476 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23477 nvptx_override_options_after_change.
23479 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23481 * dwarf2out.c (output_file_names): Avoid double testing for
23482 dwarf_version >= 5.
23484 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23486 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23488 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23490 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23492 (try_crossjump_to_edge): Do not punt on partitioned functions.
23494 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23496 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23497 Put all BBs reachable only via paths crossing cold region to cold
23499 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23501 2016-07-21 Richard Biener <rguenther@suse.de>
23503 PR tree-optimization/81303
23504 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23505 into account prologue and epilogue iterations when raising
23506 min_profitable_iters to sth at least covering one vector iteration.
23508 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23510 * config/arm/arm.c (arm_test_cpu_arch_dat):
23513 2017-07-20 Nathan Sidwell <nathan@acm.org>
23515 Remove TYPE_METHODS.
23516 * tree.h (TYPE_METHODS): Delete.
23517 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23518 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23519 (dbxout_type_methods): Scan TYPE_FIELDS.
23520 (dbxout_type): Don't check TYPE_METHODS here.
23521 * function.c (use_register_for_decl): Always ignore register for
23522 class types when not optimizing.
23523 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23524 * tree.c (free_lang_data_in_type): Stitch out member functions and
23525 templates from TYPE_FIELDS.
23526 (build_distinct_type_copy, verify_type_variant,
23527 verify_type): Member fns are on TYPE_FIELDS.
23528 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23529 * tree-pretty-print.c (dump_generic_node): Likewise.
23531 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23534 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23535 V2TImode and V4TImode.
23536 (ix86_expand_vector_extract): Likewise.
23537 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23538 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23539 (ssescalarmode): Handle V4TImode and V2TImode.
23540 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23541 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23542 (VEXTRACTI128_MODE): New mode iterator.
23543 (splitter for *vec_extractv?ti first element): New.
23544 (VEC_INIT_MODE): New mode iterator.
23545 (vec_init<mode>): Consolidate 3 expanders into one using
23546 VEC_INIT_MODE mode iterator.
23548 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23550 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23551 non_spilled_static_chain_regno_p.
23553 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23555 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23557 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23559 * bb-reorder.c (connect_traces): Allow copying of blocks within
23562 2017-07-20 Richard Biener <rguenther@suse.de>
23564 * gimple.h (gimple_phi_result): Add gphi * overload.
23565 (gimple_phi_result_ptr): Likewise.
23566 (gimple_phi_arg): Likewise. Adjust index assert to only
23567 allow actual argument accesses rather than all slots available
23569 (gimple_phi_arg_def): Add gphi * overload.
23570 * tree-phinodes.c (make_phi_node): Initialize only actual
23572 (resize_phi_node): Clear memory not covered by old node,
23573 do not initialize excess argument slots.
23574 (reserve_phi_args_for_new_edge): Initialize new argument slot
23577 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23579 PR tree-optimization/81388
23581 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23583 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23584 by removing computation of may_be_zero.
23586 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23587 Tom de Vries <tom@codesourcery.com>
23589 PR middle-end/81030
23590 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23591 when gimple level profile disagrees with what RTL expander did.
23593 2017-07-20 Richard Biener <rguenther@suse.de>
23595 PR tree-optimization/61171
23596 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23597 (vect_analyze_stmt): Add slp instance parameter.
23598 (vectorizable_reduction): Likewise.
23599 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23600 (vect_is_simple_reduction): Deal with chains not detected
23601 as SLP reduction chain, specifically not properly associated
23602 chains containing a mix of plus/minus.
23603 (get_reduction_op): Remove.
23604 (get_initial_defs_for_reduction): Simplify, pass in whether
23605 this is a reduction chain, pass in the SLP node for the PHIs.
23606 (vect_create_epilog_for_reduction): Get the SLP instance as
23608 (vectorizable_reduction): Get the SLP instance as arg.
23609 During analysis remember the SLP node with the PHIs in the
23610 instance. Simplify getting at the vectorized reduction PHIs.
23611 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23612 through SLP instance.
23613 (vect_slp_analyze_operations): Likewise.
23614 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23615 (vect_transform_stmt): Likewise.
23617 2017-07-20 Tom de Vries <tom@codesourcery.com>
23619 PR tree-optimization/81489
23620 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23621 read of phi arg location to before loop that modifies phi.
23623 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23625 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23628 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23630 PR middle-end/81331
23631 * except.c (execute): Fix ordering issue.
23633 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23635 PR rtl-optimization/81423
23636 * combine.c (make_compound_operation_int): Don't try to optimize
23637 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23639 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23641 PR rtl-optimization/81423
23642 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23643 with a constant that is -1 in the truncated to mode.
23645 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23647 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23648 (determine_unlikely_bbs): ... here.
23649 * predict.h (propagate_unlikely_bbs_forward): Declare.
23650 * cfgexpand.c (pass_expand::execute): Use it.
23651 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23653 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23654 propagate_unlikely_bbs_forward.
23656 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23658 PR middle-end/81331
23659 * except.c (maybe_add_nop_after_section_switch): New function.
23662 2017-07-19 Tom de Vries <tom@codesourcery.com>
23664 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23666 2017-07-19 Tom de Vries <tom@codesourcery.com>
23668 * gimple.h (gimple_phi_arg): Make assert more strict.
23670 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23672 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23673 * config/rs6000/mmintrin.h: New file.
23674 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23676 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23678 PR tree-optimization/81346
23679 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23681 2017-07-19 Tom de Vries <tom@codesourcery.com>
23683 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23685 2017-07-19 Tom de Vries <tom@codesourcery.com>
23687 * config/nvptx/nvptx-modes.def: Add V2DImode.
23688 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23689 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23690 (nvptx_output_mov_insn): Handle lack of mov.b128.
23691 (nvptx_print_operand): Handle 'H' and 'L' codes.
23692 (nvptx_vector_mode_supported): Allow V2DImode.
23693 (nvptx_preferred_simd_mode): New function.
23694 (nvptx_data_alignment): New function.
23695 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23696 nvptx_preferred_simd_mode.
23697 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23699 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23701 2017-07-19 Tom de Vries <tom@codesourcery.com>
23703 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23704 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23705 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23706 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23707 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23708 (mov<VECIM>_insn): New define_insn.
23709 (define_expand "mov<VECIM>): New define_expand.
23711 2017-07-19 Tom de Vries <tom@codesourcery.com>
23713 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
23715 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23717 PR tree-optimization/81346
23718 * fold-const.h (fold_div_compare, range_check_type): Declare.
23719 * fold-const.c (range_check_type): New function.
23720 (build_range_check): Use range_check_type.
23721 (fold_div_compare): No longer static, rewritten into
23722 a match.pd helper function.
23723 (fold_comparison): Don't call fold_div_compare here.
23724 * match.pd (X / C1 op C2): New optimization using fold_div_compare
23725 as helper function.
23727 2017-07-19 Nathan Sidwell <nathan@acm.org>
23729 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
23730 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
23731 * tree.c (find_decls_types_r, verify_type): Use
23732 TYPE_{MIN,MAX}_VALUE_RAW.
23733 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
23734 (hash_tree): Likewise.
23735 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
23737 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
23740 2017-07-18 Tom de Vries <tom@codesourcery.com>
23742 PR middle-end/81464
23743 * omp-expand.c (expand_omp_for_static_chunk): Handle
23744 equal-argument loop exit phi.
23746 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23749 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23750 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23751 operand 2 predicate.
23752 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23753 operand 2 predicate.
23754 (ror,rol -> rorx splitters): Use const_int_operand as
23755 operand 2 predicate.
23757 2017-06-18 Richard Biener <rguenther@suse.de>
23759 PR tree-optimization/81410
23760 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23761 the gap in the ! slp_perm SLP case after each group.
23763 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23765 PR middle-end/81463
23766 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
23769 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23771 PR middle-end/81462
23772 * predict.c (set_even_probabilities): Cleanup; do not affect
23773 probabilities that are already known.
23774 (combine_predictions_for_bb): Call even when count is set.
23776 2017-07-18 Nathan Sidwell <nathan@acm.org>
23778 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
23781 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23784 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
23785 optimization for loop niter analysis.
23787 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
23790 * config/avr/avr.c (avr_optimize_casesi): Don't use
23791 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
23793 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
23795 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
23796 body_cost_vec from _vect_peel_extended_info.
23797 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23798 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23801 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23803 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23805 2017-07-18 Richard Biener <rguenther@suse.de>
23807 PR tree-optimization/80620
23808 PR tree-optimization/81403
23809 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23810 info when re-using a VN table entry.
23812 2017-07-18 Richard Biener <rguenther@suse.de>
23814 PR tree-optimization/81418
23815 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23816 vectype_in. Verify that with lane-reducing reduction operations
23817 we have a single def-use cycle.
23819 2017-07-17 Carl Love <cel@us.ibm.com>
23821 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23823 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23824 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23825 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23826 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23827 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23828 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23829 VMULOSW): New enum "unspec" values.
23830 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23831 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23832 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23833 altivec_vmulosw): New patterns.
23834 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23835 VMULOSW): Add definitions.
23837 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23839 * config/alpha/alpha.c: Include predict.h.
23841 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23843 * tree-vrp.c (compare_assert_loc): Fix comparison function
23844 to return predictable results.
23846 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23848 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23850 (subdi3): Likewise.
23851 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23852 * doc/invoke.texi (mexpand-adddi): Update text.
23854 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23856 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23857 that also clobbers the CC register. The old expand code is moved
23859 (*arc_clzsi2): ... here.
23860 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23861 the CC register. The old expand code is moved to ...
23862 (arc_ctzsi2): ... here.
23864 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23866 * config/arc/arc.opt (mindexed-loads): Use initial value
23867 TARGET_INDEXED_LOADS_DEFAULT.
23868 (mauto-modify-reg): Use initial value
23869 TARGET_AUTO_MODIFY_REG_DEFAULT.
23870 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23871 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23872 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23873 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23875 2017-07-17 Martin Liska <mliska@suse.cz>
23878 * opts.c (finish_options): Do not allow -fgnu-tm
23879 w/ -fsanitize={kernel-,}address. Say sorry.
23881 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23884 * tree-loop-distribution.c (classify_partition): Only assert on
23885 numer of iterations.
23886 (merge_dep_scc_partitions): Delete prameter. Update function call.
23887 (distribute_loop): Remove code handling loop with unknown niters.
23888 (pass_loop_distribution::execute): Skip loop with unknown niters.
23890 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23893 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23894 function sort_partitions_by_post_order.
23896 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23898 PR tree-optimization/81374
23899 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23900 the max index of basic blocks, rather than number of basic blocks.
23902 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23904 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23906 (arc_legitimate_pic_operand_p): Likewise.
23907 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23909 (arc_needs_pcl_p): Likewise.
23910 (arc_legitimate_pc_offset_p): Likewise.
23911 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23912 function is also used in constrains.md.
23913 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23914 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23915 PLUS. Only return true/false in known cases, otherwise assert.
23916 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23917 is already called in arc_legitimate_constant_p.
23918 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23920 (LEGITIMATE_PIC_OPERAND_P): Use
23921 arc_raw_symbolic_reference_mentioned_p function.
23922 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23927 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23928 Andrew Burgess <andrew.burgess@embecosm.com>
23930 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23931 (arc_return_address_register): New function.
23932 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23933 (arc_handle_fndecl_attribute): Add naked attribute.
23934 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23935 (TARGET_WARN_FUNC_RETURN): Likewise.
23936 (arc_allocate_stack_slots_for_args): New function.
23937 (arc_warn_func_return): Likewise.
23938 (machine_function): Change type fn_type.
23939 (arc_compute_function_type): Consider new naked function type,
23940 change function return type.
23941 (arc_must_save_register): Adapt to handle new
23942 arc_compute_function_type's return type.
23943 (arc_expand_prologue): Likewise.
23944 (arc_expand_epilogue): Likewise.
23945 (arc_return_address_regs): Delete.
23946 (arc_return_address_register): New function.
23947 (arc_epilogue_uses): Use above function.
23948 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23949 (arc_function_type): Change encoding, add naked type.
23950 (ARC_INTERRUPT_P): Change to handle the new encoding.
23951 (ARC_FAST_INTERRUPT_P): Likewise.
23952 (ARC_NORMAL_P): Define.
23953 (ARC_NAKED_P): Likewise.
23954 (arc_compute_function_type): Delete prototype.
23955 * config/arc/arc.md (in_ret_delay_slot): Use
23956 arc_return_address_register function.
23957 (simple_return): Likewise.
23958 (p_return_i): Likewise.
23960 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23962 PR tree-optimization/81428
23963 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23964 can't be built for those types.
23966 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23968 Remove stuff dead since r239246.
23970 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23971 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23972 (avr_inform_devices): Remove dead stuff.
23974 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23976 * config/arm/arm_neon.h: Fix softp typo.
23978 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23980 PR tree-optimization/81365
23981 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23982 aggregate moves onto bb predecessor edges, make sure there are no
23983 loads that could alias the lhs in between the start of bb and the
23986 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23989 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23990 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23991 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23993 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23995 PR tree-optimization/81396
23996 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23997 (init_symbolic_number): Initialize it to 1.
23998 (perform_symbolic_merge): Add n_ops from both operands into the new
24000 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24001 without base_addr as useless if they need more than one operation.
24002 (bswap_replace): Handle !bswap case for NULL base_addr.
24004 2017-07-17 Tom de Vries <tom@codesourcery.com>
24007 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24010 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24012 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24013 conditional builtin define __FIX_LEON3FT_B2BST.
24015 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24017 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24018 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24021 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24023 PR rtl-optimization/81424
24024 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24025 to remove potential trapping from operands if -fnon-call-exceptions.
24027 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24029 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24030 profile_proability for scalling.
24031 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24033 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24035 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24037 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24039 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24040 fixpoint arithmetics.
24042 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24044 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24045 fixpoint arithmetics.
24047 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24049 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24050 fixpoint arithmetics.
24052 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24054 * profile-count.h (profile_probability::from_reg_br_prob_note,
24055 profile_probability::to_reg_br_prob_note): New functions.
24056 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24057 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24058 * predict.c (probability_reliable_p): Update.
24059 (edge_probability_reliable_p): Update.
24060 (br_prob_note_reliable_p): Update.
24061 (invert_br_probabilities): Update.
24062 (add_reg_br_prob_note): New function.
24063 (combine_predictions_for_insn): Update.
24064 * asan.c (asan_clear_shadow): Update.
24065 * cfgbuild.c (compute_outgoing_frequencies): Update.
24066 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24067 (update_br_prob_note): Update.
24068 (rtl_verify_edges): Update.
24069 (purge_dead_edges): Update.
24070 (fixup_reorder_chain): Update.
24071 * emit-rtl.c (try_split): Update.
24072 * ifcvt.c (cond_exec_process_insns): Update.
24073 (cond_exec_process_if_block): Update.
24074 (dead_or_predicable): Update.
24075 * internal-fn.c (expand_addsub_overflow): Update.
24076 (expand_neg_overflow): Update.
24077 (expand_mul_overflow): Update.
24078 * loop-doloop.c (doloop_modify): Update.
24079 * loop-unroll.c (compare_and_jump_seq): Update.
24080 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24081 * predict.h: Update.
24082 * reorg.c (mostly_true_jump): Update.
24084 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24085 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24086 * config/arc/arc.c: (emit_unlikely_jump): Update.
24087 * config/arm/arm.c: (emit_unlikely_jump): Update.
24088 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24089 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24090 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24091 (ix86_print_operand): Update.
24092 (ix86_split_fp_branch): Update.
24093 (predict_jump): Update.
24094 * config/ia64/ia64.c (ia64_print_operand): Update.
24095 * config/mmix/mmix.c (mmix_print_operand): Update.
24096 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24097 (rs6000_expand_split_stack_prologue): Update.
24098 * config/rs6000/rs6000.c: Update.
24099 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24100 (s390_expand_vec_movstr): Update.
24101 (s390_expand_cs_tdsi): Update.
24102 (s390_expand_split_stack_prologue): Update.
24103 * config/sh/sh.c (sh_print_operand): Update.
24104 (expand_cbranchsi4): Update.
24105 (expand_cbranchdi4): Update.
24106 * config/sparc/sparc.c (output_v9branch): Update.
24107 * config/spu/spu.c (get_branch_target): Update.
24108 (ea_load_store_inline): Update.
24109 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24110 * config/tilepro/tilepro.c: Update.
24112 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24114 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24115 (gimplify_save_expr): Likewise.
24117 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24119 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24121 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24123 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24125 * timevar.def (TV_IPA_FNSUMMARY): Define.
24127 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24129 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24130 to back store errata sensitive sequence from being generated.
24131 (sqrtdf2_fix): Likewise.
24133 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24135 * tree-ssa-threadupdate.c (compute_path_counts,
24136 update_joiner_offpath_counts): Use profile_probability.
24138 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24141 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24143 * config/arm/arm-c.c (arm_cpu_builtins): Define
24144 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24146 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24148 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24149 array entries to represent __ieee128 versions of the
24150 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24151 scalar_extract_sig, and scalar_insert_exp built-in functions.
24152 (altivec_resolve_overloaded_builtin): Add special case handling
24153 for the __builtin_scalar_insert_exp function, as represented by
24154 the P9V_BUILTIN_VEC_VSIEDP constant.
24155 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24156 exponent support for __ieee128 argument.
24157 (VSESQP): Add scalar extract signature support for __ieee128
24159 (VSTDCNQP): Add scalar test negative support for __ieee128
24161 (VSIEQP): Add scalar insert exponent support for __int128 argument
24162 with __ieee128 result.
24163 (VSIEQPF): Add scalar insert exponent support for __ieee128
24164 argument with __ieee128 result.
24165 (VSTDCQP): Add scalar test data class support for __ieee128
24167 (VSTDCNQP): Add overload support for scalar test negative with
24168 __ieee128 argument.
24169 (VSTDCQP): Add overload support for scalar test data class
24170 __ieee128 argument.
24171 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24172 UNSPEC_VSX_SXSIGDP.
24173 (UNSPEC_VSX_SIEXPQP): New constant.
24174 (xsxexpqp): New insn for VSX scalar extract exponent quad
24176 (xsxsigqp): New insn for VSX scalar extract significand quad
24178 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24179 precision with floating point argument.
24180 (xststdcqp): New expand for VSX scalar test data class quad
24182 (xststdcnegqp): New expand for VSX scalar test negative quad
24184 (xststdcqp): New insn to match expansions for VSX scalar test data
24185 class quad precision and VSX scalar test negative quad precision.
24186 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24187 special case operand checking to enforce that second operand of
24188 VSX scalar test data class with quad precision argument is a 7-bit
24190 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24191 prototypes and descriptions of __ieee128 versions of
24192 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24193 scalar_test_data_class, and scalar_test_neg built-in functions.
24195 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24197 PR tree-optimization/81162
24198 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24199 replace a negate with an add.
24201 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24203 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24205 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24207 * config/arm/arm-c.c (arm_cpu_builtins): Define
24208 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24210 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24212 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24213 (armv8-r): Set ARM Cortex-R52 as default CPU.
24214 * config/arm/arm-tables.opt: Regenerate.
24215 * config/arm/arm-tune.md: Regenerate.
24216 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24218 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24219 extension for -mcpu=cortex-r52.
24221 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24223 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24224 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24225 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24226 (fp-armv8): Define it as FP_ARMv8 only.
24227 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24228 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24230 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24231 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24232 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24233 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24234 than TARGET_FPU_ARMV8.
24235 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24236 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24237 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24239 * config/arm/neon.md (neon_vrint): Likewise.
24240 (neon_vcvt): Likewise.
24241 (neon_<fmaxmin_op><mode>): Likewise.
24242 (<fmaxmin><mode>3): Likewise.
24243 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24244 * config/arm/predicates.md (arm_cond_move_operator): Check against
24245 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24247 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24249 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24250 to top of function.
24252 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24254 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24255 loop in comment with memset.
24257 2017-07-14 Martin Liska <mliska@suse.cz>
24259 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24260 * dwarf2out.c (is_java): Remove the function.
24261 (output_pubname): Remove usage of the function.
24262 (lower_bound_default): Remove usage of DW_LANG_Java.
24263 (gen_compile_unit_die): Likewise.
24264 * gcc.c: Remove compiler defaults for .java and .zip files.
24265 * gimple-expr.c (remove_suffix): Change as there's no longer
24266 extension than 4-letter one.
24267 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24268 (gimplify_save_expr): Likewise.
24269 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24270 as it's possible even for other languages than Java.
24271 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24272 * lto-opts.c (lto_write_options): Remove reference to Java.
24273 * opts.c (strip_off_ending): Update file extension handling.
24274 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24275 * tree-eh.c (lower_resx): Likewise.
24276 * tree.c (free_lang_data_in_type): Remove dead code.
24277 (find_decls_types_r): Likewise.
24278 (build_common_builtin_nodes): Remove Java from a comment.
24279 (verify_type): Remove dead code.
24280 * varasm.c (assemble_external): Remove Java from a comment.
24282 2017-07-14 Martin Liska <mliska@suse.cz>
24284 * opts.c (finish_options): Add quotes.
24285 (common_handle_option): Likewise.
24287 2017-07-14 Martin Liska <mliska@suse.cz>
24289 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24290 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24291 Remove N_SO_PASCAL.
24292 * dwarf2out.c (lower_bound_default): Do not handle
24294 (gen_compile_unit_die): Likewise.
24295 * gcc.c: Remove default extension binding for GNU Pascal.
24296 * stmt.c: Remove Pascal language from a comment.
24297 * xcoffout.c: Likewise.
24299 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24302 * diagnostic-show-locus.c (fixit_cmp): New function.
24303 (layout::layout): Sort m_fixit_hints.
24304 (column_range::column_range): Assert that the values are valid.
24305 (struct char_span): New struct.
24306 (correction::overwrite): New method.
24307 (struct source_line): New struct.
24308 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24309 calls in terms of classes source_line and char_span, and
24310 correction::overwrite.
24311 (selftest::test_overlapped_fixit_printing_2): New function.
24312 (selftest::diagnostic_show_locus_c_tests): Call it.
24314 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24316 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24317 early if there is no lhs.
24319 2017-07-13 Martin Liska <mliska@suse.cz>
24321 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24322 (gen_reference_type_die): Likewise.
24323 * stor-layout.c: Remove Pascal-related comment.
24325 2017-07-13 Martin Liska <mliska@suse.cz>
24327 * opts.c (finish_options): Add quotes to error messages.
24328 (parse_sanitizer_options): Likewise.
24330 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24332 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24334 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24336 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24338 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24340 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24342 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24344 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24347 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24348 provides the hardware capability bits, define the macro
24349 __BUILTIN_CPU_SUPPORTS__.
24350 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24351 if GLIBC does not provide the hardware capability bits. Add a
24352 gcc_unreachable call if the built-in cpu function is neither
24353 __builtin_cpu_is nor __builtin_cpu_supports.
24354 (rs6000_get_function_versions_dispatcher): Change the warning
24355 that an old GLIBC is used which does not export the capability
24356 bits to be an error.
24357 * doc/extend.texi (target_clones attribute): Document the
24358 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24359 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24360 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24361 the macros defined by GCC if the newer GLIBC is available.
24363 2017-07-12 Jeff Law <law@redhat.com>
24365 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24366 remaining includes slightly.
24367 * config/riscv/riscv-builtins.c: Include profile-count.h.
24369 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24372 * config/avr/avr.c (avr_set_current_function): In diagnostic
24373 messages: Quote keywords and (parts of) identifiers.
24374 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24377 2017-07-12 Carl Love <cel@us.ibm.com>
24379 * config/rs6000/rs6000-c.c: Add support for built-in functions
24380 vector bool char vec_revb (vector bool char);
24381 vector bool short vec_revb (vector short char);
24382 vector bool int vec_revb (vector bool int);
24383 vector bool long long vec_revb (vector bool long long);
24384 * doc/extend.texi: Update the built-in documentation file for the
24385 new built-in functions.
24387 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24389 * config/s390/s390.md: Remove movcc splitter.
24391 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24393 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24394 load/store on condition.
24396 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24399 * config/avr/avr.c (avr_encode_section_info)
24400 [progmem && !TREE_READONLY]: Error if progmem object needs
24403 2017-07-11 Michael Collison <michael.collison@arm.com>
24405 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24408 2017-07-11 Carl Love <cel@us.ibm.com>
24410 * config/rs6000/rs6000-c.c: Add support for builtins
24411 vector unsigned int vec_parity_lsbb (vector signed int);
24412 vector unsigned int vec_parity_lsbb (vector unsigned int);
24413 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24414 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24415 vector unsigned long long vec_parity_lsbb (vector signed long long);
24416 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24417 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24418 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24419 * doc/extend.texi: Update the built-in documentation file for the
24420 new built-in functions.
24422 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24424 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24425 (layout::m_primary_loc): New field.
24426 (layout::layout): Initialize new field. Move location filtering
24427 logic from here to...
24428 (layout::maybe_add_location_range): ...this new method. Add
24429 support for filtering to just the lines already specified by other
24431 (layout::will_show_line_p): New method.
24432 (gcc_rich_location::add_location_if_nearby): New method.
24433 (selftest::test_add_location_if_nearby): New test function.
24434 (selftest::diagnostic_show_locus_c_tests): Call it.
24435 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24438 2017-07-11 Tom de Vries <tom@codesourcery.com>
24440 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24441 (bb_first_real_insn): New function.
24442 (nvptx_single): Add extra initialization of broadcasted condition
24445 2017-07-11 Nathan Sidwell <nathan@acm.org>
24447 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24449 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24451 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24452 Binutils doc as TEXI will mess them up.
24453 * doc/invoke.texi (AVR Options): Same here.
24455 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24457 * config/sparc/sparc.opt (mfix-ut700): New option.
24458 (mfix-gr712rc): Likewise.
24459 (sparc_fix_b2bst): New variable.
24460 * doc/invoke.texi (SPARC options): Document them.
24461 (ARM options): Fix warnings.
24462 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24463 instructions to prevent sequences that can trigger the store-store
24464 errata for certain LEON3FT processors.
24465 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24466 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24467 * config/sparc/sparc.md (fix_b2bst): New attribute.
24468 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24470 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24473 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24475 (*rsqrtsf2_sse): Ditto.
24477 (div<mode>3): Macroize insn from divdf3 and divsf3
24478 using MODEF mode iterator.
24480 2017-07-10 Martin Sebor <msebor@redhat.com>
24482 PR tree-optimization/80397
24483 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24484 instead of testing for equality to INTEGER_TYPE.
24486 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24488 * config.gcc: Remove uclibc from arc target spec.
24490 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24492 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24494 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24497 * lto-wrapper.c (remove_option): New function.
24498 (merge_and_complain): Merge PIC/PIE options more realistically.
24500 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24502 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24506 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24507 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24508 * config.in: Regenerate.
24509 * configure: Regenerate.
24510 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24511 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24512 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24513 (TARGET_GASISR_PROLOGUES): ...target mask.
24514 * common/config/avr/avr-common.c
24515 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24516 Set -mgas-isr-prologues.
24517 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24518 INSERT_PASS_BEFORE for it.
24519 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24520 * config/avr/avr.c (avr_option_override)
24521 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24522 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24523 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24524 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24525 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24527 (make_avr_pass_pre_proep): New function.
24528 (emit_push_sfr) <treg>: Add argument to function and use it
24529 instead of TMP_REG.
24530 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24531 and set machine->gasisr.yes.
24532 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24533 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24534 __gcc_isr.n_pushed to .L__stack_usage.
24535 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24536 (avr_asm_final_postscan_insn): ...this new static function.
24537 * config/avr/avr.h (machine_function)
24538 <is_no_gccisr, use_L__stack_usage>: New fields.
24539 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24540 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24541 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24542 (gasisr, *gasisr): New expander and insn.
24543 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24544 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24545 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24547 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24549 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24552 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24554 Move jump-tables out of .text again.
24557 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24558 (ASM_OUTPUT_ADDR_VEC): New function.
24559 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24560 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24561 INSN_ADDRESSes as asm comment.
24562 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24563 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24564 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24565 * config/avr/avr.md (*tablejump): Adjust comment.
24566 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24567 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24569 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24570 (avr_output_addr_vec): New proto.
24571 (avr_log_t) <insn_addresses>: New field.
24573 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24576 * config/i386/i386.c (ix86_function_arg_advance): Set
24577 outgoing_args_on_stack to true if there are outgoing arguments
24579 (ix86_function_arg): Likewise.
24580 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24581 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24582 * config/i386/i386.h (machine_function): Add
24583 outgoing_args_on_stack.
24585 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24587 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24588 supporting pthreds.
24589 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24591 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24593 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24594 (REAL_H): Remove $(MACHMODE_H).
24595 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24597 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24598 $(MACHMODE_H) and double-int.h.
24599 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24601 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24604 2017-07-07 Andrew Pinski <apinski@cavium.com>
24606 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24607 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24609 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24611 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24612 Add warning if GCC was not configured to link against a GLIBC that
24613 exports the hardware capability bits.
24614 (make_resolver_func): Make resolver function private and not a
24615 COMDAT function. Create the name with clone_function_name instead
24616 of make_unique_name.
24619 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24620 correct operand in doing the split.
24622 2017-07-07 Carl Love <cel@us.ibm.com>
24624 * config/rs6000/rs6000-c: Add support for built-in function
24625 vector unsigned short vec_pack_to_short_fp32 (vector float,
24627 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24628 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24629 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24630 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24631 (convert_4f32_8i16): Add define_expand.
24632 * doc/extend.texi: Update the built-in documentation file for the
24633 new built-in function.
24635 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24637 * config/sparc/m8.md: New file.
24638 * config/sparc/sparc.md: Include m8.md.
24640 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24642 * config/sparc/sparc.opt: New option -mvis4b.
24643 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24644 (sparc_option_override): Handle VIS4B.
24645 (enum sparc_builtins): Define
24646 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24647 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24648 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24649 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24650 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24651 (check_constant_argument): New function.
24652 (sparc_vis_init_builtins): Define builtins
24653 __builtin_vis_dictunpack{8,16,32},
24654 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24655 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24656 __builtin_vis_fpcmpde{8,16,32}shl and
24657 __builtin_vis_fpcmpur{8,16,32}shl.
24658 (sparc_expand_builtin): Check that the constant operands to
24659 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24660 constant and in range.
24661 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24663 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24664 (SPARC_IMM5_P): Likewise.
24665 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24666 (enabled): Handle vis4b.
24667 (UNSPEC_DICTUNPACK): New unspec.
24668 (UNSPEC_FPCMPSHL): Likewise.
24669 (UNSPEC_FPUCMPSHL): Likewise.
24670 (UNSPEC_FPCMPDESHL): Likewise.
24671 (UNSPEC_FPCMPURSHL): Likewise.
24672 (cpu_feature): New CPU feature `vis4b'.
24673 (dictunpack{8,16,32}): New insns.
24674 (FPCSMODE): New mode iterator.
24675 (fpcscond): New code iterator.
24676 (fpcsucond): Likewise.
24677 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24678 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24679 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24680 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24681 * config/sparc/constraints.md: Define constraints `q' for unsigned
24682 2-bit integer constants and `t' for unsigned 5-bit integer
24684 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24686 (imm5_operand_dictunpack16): Likewise.
24687 (imm5_operand_dictunpack32): Likewise.
24688 (imm2_operand): Likewise.
24689 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24690 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24691 ditunpack* and fpcmp*shl builtins.
24693 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24695 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24696 * config.in: Add HAVE_AS_SPARC6 define.
24697 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24699 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24701 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24702 (CPP_CPU_SPEC): Handle m8.
24703 (ASM_CPU_SPEC): Likewise.
24704 * config/sparc/sparc-opts.h (enum processor_type): Add
24706 * config/sparc/sparc.c (m8_costs): New struct.
24707 (sparc_option_override): Handle TARGET_CPU_m8.
24708 (sparc32_initialize_trampoline): Likewise.
24709 (sparc64_initialize_trampoline): Likewise.
24710 (sparc_issue_rate): Likewise.
24711 (sparc_register_move_cost): Likewise.
24712 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
24713 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
24714 (ASM_CPU64_DEFAULT_SPEC): Likewise.
24715 (CPP_CPU_SPEC): Handle M8.
24716 (ASM_CPU_SPEC): Likewise.
24717 (AS_M8_FLAG): Define.
24718 * config/sparc/sparc.md: Add m8 to the cpu attribute.
24719 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
24720 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
24722 * configure: Regenerate.
24723 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
24726 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24728 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
24730 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
24731 ("*movdi_insn_sp32"): Do not set v3pipe.
24732 ("*movsi_insn"): Likewise.
24733 ("*movdi_insn_sp64"): Likewise.
24734 ("*movsf_insn"): Likewise.
24735 ("*movdf_insn_sp32"): Likewise.
24736 ("*movdf_insn_sp64"): Likewise.
24737 ("*zero_extendsidi2_insn_sp64"): Likewise.
24738 ("*sign_extendsidi2_insn"): Likewise.
24739 ("*mov<VM32:mode>_insn"): Likewise.
24740 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24741 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24742 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24743 ("<vlop:code><VL:mode>3"): Likewise.
24744 ("*not_<vlop:code><VL:mode>3"): Likewise.
24745 ("*nand<VL:mode>_vis"): Likewise.
24746 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24747 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24748 ("one_cmpl<VL:mode>2"): Likewise.
24749 ("faligndata<VM64:mode>_vis"): Likewise.
24750 ("alignaddrsi_vis"): Likewise.
24751 ("alignaddrdi_vis"): Likweise.
24752 ("alignaddrlsi_vis"): Likewise.
24753 ("alignaddrldi_vis"): Likewise.
24754 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24755 ("bmaskdi_vis"): Likewise.
24756 ("bmasksi_vis"): Likewise.
24757 ("bshuffle<VM64:mode>_vis"): Likewise.
24758 ("cmask8<P:mode>_vis"): Likewise.
24759 ("cmask16<P:mode>_vis"): Likewise.
24760 ("cmask32<P:mode>_vis"): Likewise.
24761 ("pdistn<P:mode>_vis"): Likewise.
24762 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24764 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24766 * config/sparc/sparc.md ("subtype"): New insn attribute.
24767 ("*wrgsr_sp64"): Set insn subtype.
24768 ("*rdgsr_sp64"): Likewise.
24769 ("alignaddrsi_vis"): Likewise.
24770 ("alignaddrdi_vis"): Likewise.
24771 ("alignaddrlsi_vis"): Likewise.
24772 ("alignaddrldi_vis"): Likewise.
24773 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24774 ("fexpand_vis"): Likewise.
24775 ("fpmerge_vis"): Likewise.
24776 ("faligndata<VM64:mode>_vis"): Likewise.
24777 ("bshuffle<VM64:mode>_vis"): Likewise.
24778 ("cmask8<P:mode>_vis"): Likewise.
24779 ("cmask16<P:mode>_vis"): Likewise.
24780 ("cmask32<P:mode>_vis"): Likewise.
24781 ("fchksm16_vis"): Likewise.
24782 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
24783 ("fmean16_vis"): Likewise.
24784 ("fp<plusminus_insn>64_vis"): Likewise.
24785 ("<plusminus_insn>v8qi3"): Likewise.
24786 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24787 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
24788 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
24789 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
24790 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
24791 ("*movqi_insn"): Likewise.
24792 ("*movhi_insn"): Likewise.
24793 ("*movsi_insn"): Likewise.
24794 ("movsi_pic_gotdata_op"): Likewise.
24795 ("*movdi_insn_sp32"): Likewise.
24796 ("*movdi_insn_sp64"): Likewise.
24797 ("movdi_pic_gotdata_op"): Likewise.
24798 ("*movsf_insn"): Likewise.
24799 ("*movdf_insn_sp32"): Likewise.
24800 ("*movdf_insn_sp64"): Likewise.
24801 ("*zero_extendhisi2_insn"): Likewise.
24802 ("*zero_extendqihi2_insn"): Likewise.
24803 ("*zero_extendqisi2_insn"): Likewise.
24804 ("*zero_extendqidi2_insn"): Likewise.
24805 ("*zero_extendhidi2_insn"): Likewise.
24806 ("*zero_extendsidi2_insn_sp64"): Likewise.
24807 ("ldfsr"): Likewise.
24808 ("prefetch_64"): Likewise.
24809 ("prefetch_32"): Likewise.
24810 ("tie_ld32"): Likewise.
24811 ("tie_ld64"): Likewise.
24812 ("*tldo_ldub_sp32"): Likewise.
24813 ("*tldo_ldub1_sp32"): Likewise.
24814 ("*tldo_ldub2_sp32"): Likewise.
24815 ("*tldo_ldub_sp64"): Likewise.
24816 ("*tldo_ldub1_sp64"): Likewise.
24817 ("*tldo_ldub2_sp64"): Likewise.
24818 ("*tldo_ldub3_sp64"): Likewise.
24819 ("*tldo_lduh_sp32"): Likewise.
24820 ("*tldo_lduh1_sp32"): Likewise.
24821 ("*tldo_lduh_sp64"): Likewise.
24822 ("*tldo_lduh1_sp64"): Likewise.
24823 ("*tldo_lduh2_sp64"): Likewise.
24824 ("*tldo_lduw_sp32"): Likewise.
24825 ("*tldo_lduw_sp64"): Likewise.
24826 ("*tldo_lduw1_sp64"): Likewise.
24827 ("*tldo_ldx_sp64"): Likewise.
24828 ("*mov<VM32:mode>_insn"): Likewise.
24829 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24830 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24832 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24834 * config/sparc/sparc.md ("type"): New insn type viscmp.
24835 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24837 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24838 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24839 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24840 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24842 ("n7_vis_logical_11cycle"): Likewise.
24843 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24844 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24845 * config/sparc/niagara.md ("niag_vis"): Likewise.
24846 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24847 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24849 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24851 * config/sparc/sparc.md: New instruction type `bmask'.
24852 (bmaskdi_vis): Use the `bmask' type.
24853 (bmasksi_vis): Likewise.
24854 * config/sparc/ultra3.md (us3_array): Likewise.
24855 * config/sparc/niagara7.md (n7_array): Likewise.
24856 * config/sparc/niagara4.md (n4_array): Likewise.
24857 * config/sparc/niagara2.md (niag2_vis): Likewise.
24858 (niag3_vis): Likewise.
24859 * config/sparc/niagara.md (niag_vis): Likewise.
24861 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24863 * ipa-comdats.c: Remove optimize check from gate.
24864 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24865 for functions not optimized.
24866 (ipa_fn_summary_read): Skip optimize check.
24867 (ipa_fn_summary_write): Likewise.
24868 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24870 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24872 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24873 (check_callers): Give up on caller which is not optimized.
24874 (inline_small_functions): Likewise.
24875 (ipa_inline): Do not give up when not optimizing.
24876 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24877 away unoptimizes cdtors.
24878 (whole_program_function_and_variable_visibility): Do
24879 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24880 * ipa.c (process_references): Do not check optimize.
24881 (symbol_table::remove_unreachable_nodes): Update optimize check.
24882 (set_writeonly_bit): Update optimize check.
24883 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24884 (pass_ipa_single_use::gate): Remove.
24886 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24888 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24889 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24890 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24891 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24892 permute_load, permute_store, adjust_extract, adjust_splat,
24893 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24894 replace_swap_with_copy, dump_swap_insn_table,
24895 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24896 recombine_lvx_pattern, recombine_stvx_pattern,
24897 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24898 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24899 to file rs6000-p8swap.c.
24900 * config/rs6000/rs6000-p8swap.c: New file.
24901 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24902 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24903 and rs6000*-*-* targets.
24905 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24907 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24909 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24911 * lto-wrapper.c (merge_and_complain): Do not merge
24912 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24913 fsigned_zeros, ftrapping_math, fwrapv.
24914 (append_compiler_options): Do not track these options.
24915 (append_linker_options): Likewie
24917 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24919 * cgraphunit.c (cgraph_node::finalize_function): When
24920 !flag_toplevel_reorde set no_reorder flag.
24921 (varpool_node::finalize_decl): Likewise.
24922 (symbol_table::compile): Drop no toplevel reorder path.
24924 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24926 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24927 edges; zero probability is not better than uninitialized.
24929 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24931 * asan.h (asan_sanitize_allocas_p): Declare.
24932 * asan.c (asan_sanitize_allocas_p): New function.
24933 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24934 (handle_builtin_alloca): Likewise.
24935 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24936 if !asan_sanitize_allocas_p.
24937 * params.def (asan-instrument-allocas): Add new option.
24938 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24939 * opts.c (common_handle_option): Disable allocas sanitization for
24942 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24944 * asan.c: Include gimple-fold.h.
24945 (get_last_alloca_addr): New function.
24946 (handle_builtin_stackrestore): Likewise.
24947 (handle_builtin_alloca): Likewise.
24948 (asan_emit_allocas_unpoison): Likewise.
24949 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24950 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24951 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24952 (instrument_builtin_call): Pass gimple iterator to
24953 get_mem_refs_of_builtin_call.
24954 (last_alloca_addr): New global.
24955 * asan.h (asan_emit_allocas_unpoison): Declare.
24956 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24957 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24958 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24959 if function calls alloca.
24960 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24961 * gimple-fold.h (replace_call_with_value): Declare.
24962 * internal-fn.c: Include asan.h.
24963 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24964 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24966 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24968 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24969 (C_SELFTEST_FLAGS): New.
24970 (CPP_SELFTEST_FLAGS): New.
24971 (SELFTEST_DEPS): New, from deps of s-selftest.
24972 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24973 (CPP_SELFTEST_DEPS): New.
24974 (selftest): Add dependency on s-selftest-c++.
24975 (s-selftest): Rename to...
24976 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24977 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24978 than SELFTEST_FLAGS.
24979 (selftest-gdb): Rename to...
24980 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24982 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24983 (selftest-valgrind): Rename to...
24984 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24986 (selftest-valgrind): Reintroduce as an alias for
24987 selftest-c-valgrind.
24988 (s-selftest-c++): New.
24989 (selftest-c++-gdb): New.
24990 (selftest-c++-valgrind): New.
24992 2017-07-06 Olivier Hainque <hainque@adacore.com>
24994 * gcc.c (process_command): When deciding if undefined variables
24995 should be ignored when processing specs, accept "gcc -v" as well.
24997 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24999 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25000 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25002 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25004 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25005 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25006 * config/arm/arm-tables.opt: Regenerate.
25007 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25009 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25011 2017-07-06 Carl Love <cel@us.ibm.com>
25013 * ChangeLog: Clean up from mid air collision
25015 2017-07-06 Carl Love <cel@us.ibm.com>
25017 * config/rs6000/rs6000-c.c: Add support for built-in functions
25018 vector signed int vec_subc (vector signed int, vector signed int);
25019 vector signed __int128 vec_subc (vector signed __int128,
25020 vector signed __int128);
25021 vector unsigned __int128 vec_subc (vector unsigned __int128,
25022 vector unsigned __int128);
25023 vector signed int vec_sube (vector signed int, vector signed int,
25024 vector signed int);
25025 vector unsigned int vec_sube (vector unsigned int,
25026 vector unsigned int,
25027 vector unsigned int);
25028 vector signed __int128 vec_sube (vector signed __int128,
25029 vector signed __int128,
25030 vector signed__int128);
25031 vector unsigned __int128 vec_sube (vector unsigned __int128,
25032 vector unsigned __int128,
25033 vector unsigned __int128);
25034 vector signed int vec_subec (vector signed int, vector signed int,
25035 vector signed int);
25036 vector unsigned int vec_subec (vector unsigned int,
25037 vector unsigned int,
25038 vector unsigned int);
25039 vector signed __int128 vec_subec (vector signed __int128,
25040 vector signed __int128,
25041 vector signed__int128);
25042 vector unsigned __int128 vec_subec (vector unsigned __int128,
25043 vector unsigned __int128,
25044 vector unsigned __int128);
25045 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25046 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25047 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25048 BU_ALTIVEC_OVERLOAD_X definitions.
25049 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25050 * doc/extend.texi: Update the built-in documentation file for the new
25051 built-in functions.
25053 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25056 * diagnostic-show-locus.c (layout::layout): Use start and finish
25057 spelling location for the start and finish of each range.
25058 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25059 Add unused aspect param.
25060 * input.c (expand_location_1): Add "aspect" param, and use it
25061 to access the correct part of the location.
25062 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25064 (expand_location_to_spelling_point): Likewise.
25065 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25066 param, and pass it to expand_location_1.
25068 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25070 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25071 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25072 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25073 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25074 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25075 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25076 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25077 _mm_maskz_getmant_ss): New intrinsics.
25078 (__builtin_ia32_getexpss128_mask): Changed to ...
25079 __builtin_ia32_getexpss128_round ... this.
25080 (__builtin_ia32_getexpsd128_mask): Changed to ...
25081 __builtin_ia32_getexpsd128_round ... this.
25082 * config/i386/i386-builtin-types.def
25083 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25084 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25085 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25086 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25087 __builtin_ia32_getmantss_mask_round): New builtins.
25088 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25089 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25090 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25091 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25092 * config/i386/sse.md
25093 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25094 avx512f_sgetexp<mode><mask_scalar_name>
25095 <round_saeonly_scalar_name> ... this.
25096 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25097 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25098 vgetexp<ssescalarmodesuffix>
25099 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25100 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25101 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25102 avx512f_vgetmant<mode><mask_scalar_name>
25103 <round_saeonly_scalar_name> ... this.
25104 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25105 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25106 vgetmant<ssescalarmodesuffix>
25107 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25108 %0<mask_scalar_operand4>, %1, %2
25109 <round_saeonly_scalar_mask_op4>, %3} ... this.
25110 * config/i386/subst.md (mask_scalar_operand4,
25111 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25112 round_saeonly_scalar_nimm_predicate): New subst attributes.
25114 2017-07-06 Julia Koval <julia.koval@intel.com>
25116 * config/i386/i386.c (ix86_erase_embedded_rounding):
25117 Remove code for old rounding pattern.
25119 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25121 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25123 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25125 * doc/sourcebuild.texi (Test Directives, Variants of
25126 dg-require-support): Add documentation for dg-require-stack-check.
25128 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25130 * config/i386/subst.md (mask_scalar, round_scalar,
25131 round_saeonly_scalar): New meta-templates.
25132 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25133 round_scalar_mask_operand3, round_scalar_mask_op3,
25134 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25135 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25136 round_saeonly_scalar_constraint,
25137 round_saeonly_scalar_prefix): New subst attribute.
25138 * config/i386/sse.md
25139 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25140 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25141 <round_scalar_name> ... this.
25142 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25143 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25144 <round_scalar_name> ... this.
25145 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25146 <sse>_vm<code><mode>3<mask_scalar_name>
25147 <round_saeonly_scalar_name> ... this.
25148 (v<plusminus_mnemonic><ssescalarmodesuffix>
25149 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25150 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25151 v<plusminus_mnemonic><ssescalarmodesuffix>
25152 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25153 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25154 (v<multdiv_mnemonic><ssescalarmodesuffix>
25155 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25156 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25157 v<multdiv_mnemonic><ssescalarmodesuffix>
25158 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25159 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25160 (v<maxmin_float><ssescalarmodesuffix>
25161 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25162 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25163 v<maxmin_float><ssescalarmodesuffix>
25164 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25165 %0<mask_scalar_operand3>, %1, %<iptr>2
25166 <round_saeonly_scalar_mask_op3>} ... this.
25168 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25170 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25171 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25173 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25174 Alan Hayward <alan.hayward@arm.com>
25175 David Sherwood <david.sherwood@arm.com>
25177 * combine.c (simplify_if_then_else): Remove "enum" before
25179 * compare-elim.c (can_eliminate_compare): Likewise.
25180 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25182 (aarch64_lookup_simd_builtin_type): Likewise.
25183 (aarch64_simd_builtin_type): Likewise.
25184 (aarch64_init_simd_builtin_types): Likewise.
25185 (aarch64_simd_expand_args): Likewise.
25186 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25188 (aarch64_reverse_mask): Likewise.
25189 (aarch64_simd_emit_reg_reg_move): Likewise.
25190 (aarch64_gen_adjusted_ldpstp): Likewise.
25191 (aarch64_ccmp_mode_to_code): Likewise.
25192 (aarch64_operands_ok_for_ldpstp): Likewise.
25193 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25194 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25196 (aarch64_min_divisions_for_recip_mul): Likewise.
25197 (aarch64_reassociation_width): Likewise.
25198 (aarch64_get_condition_code_1): Likewise.
25199 (aarch64_simd_emit_reg_reg_move): Likewise.
25200 (aarch64_simd_attr_length_rglist): Likewise.
25201 (aarch64_reverse_mask): Likewise.
25202 (aarch64_operands_ok_for_ldpstp): Likewise.
25203 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25204 (aarch64_gen_adjusted_ldpstp): Likewise.
25205 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25207 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25208 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25209 (arm_lookup_simd_builtin_type): Likewise.
25210 (arm_simd_builtin_type): Likewise.
25211 (arm_init_simd_builtin_types): Likewise.
25212 (arm_expand_builtin_args): Likewise.
25213 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25214 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25215 (ft32_setup_incoming_varargs): Likewise.
25216 (ft32_function_arg): Likewise.
25217 (ft32_function_arg_advance): Likewise.
25218 (ft32_pass_by_reference): Likewise.
25219 (ft32_arg_partial_bytes): Likewise.
25220 (ft32_valid_pointer_mode): Likewise.
25221 (ft32_addr_space_pointer_mode): Likewise.
25222 (ft32_addr_space_legitimate_address_p): Likewise.
25223 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25225 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25226 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25227 (iamcu_alignment): Likewise.
25228 (canonicalize_vector_int_perm): Likewise.
25229 (ix86_noce_conversion_profitable_p): Likewise.
25230 (ix86_mpx_bound_mode): Likewise.
25231 (ix86_operands_ok_for_move_multiple): Likewise.
25232 * config/microblaze/microblaze-protos.h
25233 (microblaze_expand_conditional_branch_reg): Likewise.
25234 * config/microblaze/microblaze.c
25235 (microblaze_expand_conditional_branch_reg): Likewise.
25236 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25238 (rs6000_reassociation_width): Likewise.
25239 (rs6000_invalid_binary_op): Likewise.
25240 (fusion_p9_p): Likewise.
25241 (emit_fusion_p9_load): Likewise.
25242 (emit_fusion_p9_store): Likewise.
25243 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25245 (riscv_hard_regno_mode_ok_p): Likewise.
25246 (riscv_address_insns): Likewise.
25247 (riscv_split_symbol): Likewise.
25248 (riscv_legitimize_move): Likewise.
25249 (riscv_function_value): Likewise.
25250 (riscv_hard_regno_nregs): Likewise.
25251 (riscv_expand_builtin): Likewise.
25252 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25253 (riscv_build_integer): Likewise.
25254 (riscv_split_integer): Likewise.
25255 (riscv_legitimate_constant_p): Likewise.
25256 (riscv_cannot_force_const_mem): Likewise.
25257 (riscv_regno_mode_ok_for_base_p): Likewise.
25258 (riscv_valid_base_register_p): Likewise.
25259 (riscv_valid_offset_p): Likewise.
25260 (riscv_valid_lo_sum_p): Likewise.
25261 (riscv_classify_address): Likewise.
25262 (riscv_legitimate_address_p): Likewise.
25263 (riscv_address_insns): Likewise.
25264 (riscv_load_store_insns): Likewise.
25265 (riscv_force_binary): Likewise.
25266 (riscv_split_symbol): Likewise.
25267 (riscv_force_address): Likewise.
25268 (riscv_legitimize_address): Likewise.
25269 (riscv_move_integer): Likewise.
25270 (riscv_legitimize_const_move): Likewise.
25271 (riscv_legitimize_move): Likewise.
25272 (riscv_address_cost): Likewise.
25273 (riscv_subword): Likewise.
25274 (riscv_output_move): Likewise.
25275 (riscv_canonicalize_int_order_test): Likewise.
25276 (riscv_emit_int_order_test): Likewise.
25277 (riscv_function_arg_boundary): Likewise.
25278 (riscv_pass_mode_in_fpr_p): Likewise.
25279 (riscv_pass_fpr_single): Likewise.
25280 (riscv_pass_fpr_pair): Likewise.
25281 (riscv_get_arg_info): Likewise.
25282 (riscv_function_arg): Likewise.
25283 (riscv_function_arg_advance): Likewise.
25284 (riscv_arg_partial_bytes): Likewise.
25285 (riscv_function_value): Likewise.
25286 (riscv_pass_by_reference): Likewise.
25287 (riscv_setup_incoming_varargs): Likewise.
25288 (riscv_print_operand): Likewise.
25289 (riscv_elf_select_rtx_section): Likewise.
25290 (riscv_save_restore_reg): Likewise.
25291 (riscv_for_each_saved_reg): Likewise.
25292 (riscv_register_move_cost): Likewise.
25293 (riscv_hard_regno_mode_ok_p): Likewise.
25294 (riscv_hard_regno_nregs): Likewise.
25295 (riscv_class_max_nregs): Likewise.
25296 (riscv_memory_move_cost): Likewise.
25297 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25298 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25299 (rl78_addr_space_address_mode): Likewise.
25300 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25302 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25303 (rs6000_reassociation_width): Likewise.
25304 (rs6000_invalid_binary_op): Likewise.
25305 (fusion_p9_p): Likewise.
25306 (emit_fusion_p9_load): Likewise.
25307 (emit_fusion_p9_store): Likewise.
25308 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25309 (ok_for_simple_move_operands): Likewise.
25310 (ok_for_simple_move_strict_operands): Likewise.
25311 (ok_for_simple_arith_logic_operands): Likewise.
25312 (visium_legitimize_reload_address): Likewise.
25313 (visium_select_cc_mode): Likewise.
25314 (output_cbranch): Likewise.
25315 (visium_split_double_move): Likewise.
25316 (visium_expand_copysign): Likewise.
25317 (visium_expand_int_cstore): Likewise.
25318 (visium_expand_fp_cstore): Likewise.
25319 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25320 (visium_function_arg): Likewise.
25321 (visium_function_arg_advance): Likewise.
25322 (visium_libcall_value): Likewise.
25323 (visium_setup_incoming_varargs): Likewise.
25324 (visium_legitimate_constant_p): Likewise.
25325 (visium_legitimate_address_p): Likewise.
25326 (visium_legitimize_address): Likewise.
25327 (visium_secondary_reload): Likewise.
25328 (visium_register_move_cost): Likewise.
25329 (visium_memory_move_cost): Likewise.
25330 (prepare_move_operands): Likewise.
25331 (ok_for_simple_move_operands): Likewise.
25332 (ok_for_simple_move_strict_operands): Likewise.
25333 (ok_for_simple_arith_logic_operands): Likewise.
25334 (visium_function_value_1): Likewise.
25335 (rtx_ok_for_offset_p): Likewise.
25336 (visium_legitimize_reload_address): Likewise.
25337 (visium_split_double_move): Likewise.
25338 (visium_expand_copysign): Likewise.
25339 (visium_expand_int_cstore): Likewise.
25340 (visium_expand_fp_cstore): Likewise.
25341 (visium_split_cstore): Likewise.
25342 (visium_select_cc_mode): Likewise.
25343 (visium_split_cbranch): Likewise.
25344 (output_cbranch): Likewise.
25345 (visium_print_operand_address): Likewise.
25346 * expmed.c (flip_storage_order): Likewise.
25347 * expmed.h (emit_cstore): Likewise.
25348 (flip_storage_order): Likewise.
25349 * genrecog.c (validate_pattern): Likewise.
25350 * hsa-gen.c (gen_hsa_addr): Likewise.
25351 * internal-fn.c (expand_arith_overflow): Likewise.
25352 * ira-color.c (allocno_copy_cost_saving): Likewise.
25353 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25354 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25355 (process_invariant_for_inheritance): Likewise.
25356 * lra-eliminations.c (move_plus_up): Likewise.
25357 * omp-low.c (lower_oacc_reductions): Likewise.
25358 * simplify-rtx.c (simplify_subreg): Likewise.
25359 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25360 (TARGET_CHKP_BOUND_MODE): Likewise..
25361 * targhooks.c (default_chkp_bound_mode): Likewise.
25362 (default_setup_incoming_vararg_bounds): Likewise.
25363 * targhooks.h (default_chkp_bound_mode): Likewise.
25364 (default_setup_incoming_vararg_bounds): Likewise.
25365 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25366 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25367 (have_whole_vector_shift): Likewise.
25368 * tree-vect-stmts.c (vectorizable_load): Likewise.
25369 * doc/tm.texi: Regenerate.
25371 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25373 Graceful degrade if Binutils PR21472 is not available.
25376 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25377 .rodata in flash test fails.
25378 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25379 * confgure: Regenerate.
25380 * config.in: Regenerate.
25381 * config/avr/avr.c (avr_asm_named_section)
25382 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25383 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25384 (avr_asm_init_sections): Same.
25386 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25388 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25389 (fma<VH:mode>4_intrinsic): Likewise.
25390 (*fmsub<VCVTF:mode>4): Likewise.
25391 (*fmsub<VH:mode>4_intrinsic): Likewise.
25393 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25396 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25397 Don't depend on "optimize > 0".
25398 (out_movhi_r_mr, out_movqi_mr_r): Same.
25399 (out_movhi_mr_r, out_movqi_r_mr): Same.
25400 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25401 io_address_operand on "optimize > 0".
25403 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25405 * tree-loop-distribution.c: Add general explanantion on the pass.
25406 (generate_loops_for_partition): Mark distributed loop.
25407 (pg_add_dependence_edges): New parameter. Handle alias data
25408 dependence specially and record it in the parameter if asked.
25409 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25410 (init_partition_graph_vertices, add_partition_graph_edge): New.
25411 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25412 (free_partition_graph_vdata, build_partition_graph): New.
25413 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25414 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25415 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25416 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25417 (version_for_distribution_p, finalize_partitions): New.
25418 (distribute_loop): Handle alias data dependence specially. Factor
25419 out loop fusion code as functions and call these functions.
25421 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25423 * tree-loop-distribution.c (classify_partition): New parameter and
25424 better handle reduction statement.
25425 (rdg_build_partitions): Revise comment.
25426 (distribute_loop): Compute statements in all partitions and pass it
25427 to classify_partition.
25429 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25431 * tree-loop-distribution.c (enum partition_type): New.
25432 (struct partition): New field type.
25433 (partition_merge_into): Add parameter. Update partition type.
25434 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25435 (build_rdg_partition_for_vertex): Compute partition type.
25436 (rdg_build_partitions): Dump partition type.
25437 (distribute_loop): Update calls to partition_merge_into.
25439 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25441 * tree-loop-distribution.c (struct ddr_hasher): New.
25442 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25444 (classify_partition): Call get_data_dependence.
25445 (pg_add_dependence_edges): Ditto.
25446 (distribute_loop): Release data dependence hash table.
25448 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25450 * tree-loop-distribution.c (ref_base_address): Delete.
25451 (similar_memory_accesses): Rename ...
25452 (share_memory_accesses): ... to this. Check if partitions access
25453 the same memory reference.
25454 (distribute_loop): Call share_memory_accesses.
25456 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25458 * tree-loop-distribution.c (struct partition): New field recording
25459 its data reference.
25460 (partition_alloc, partition_free): Init and release data refs.
25461 (partition_merge_into): Merge data refs.
25462 (build_rdg_partition_for_vertex): Collect data refs for partition.
25463 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25465 (distribute_loop): Remve data refs from vertice data of partition
25468 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25470 * tree-loop-distribution.c (params.h): Include header file.
25471 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25472 (datarefs_vec): New global var.
25473 (create_rdg_vertices): Use datarefs_vec directly.
25474 (free_rdg): Don't free data references.
25475 (build_rdg): Update use. Don't free data references.
25476 (distribute_loop): Compute global variable for data references.
25477 Bail out if there are too many data references.
25479 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25481 * tree-loop-distribution.c (loop_nest): New global var.
25482 (build_rdg): Use loop directly, rather than loop nest.
25483 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25485 (distribute_loop): Compute global variable loop nest. Update use.
25487 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25489 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25490 (partition_merge_into): New parameter. Dump reason for fusion.
25491 (distribute_loop): Update use of partition_merge_into.
25493 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25495 * tree-loop-distribution.c (bb_top_order_index): New.
25496 (bb_top_order_index_size, bb_top_order_cmp): New.
25497 (stmts_from_loop): Use topological order.
25498 (pass_loop_distribution::execute): Compute and release topological
25499 order for basic blocks.
25501 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25503 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25506 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25508 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25509 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25510 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25511 * internal-fn.def (LOOP_DIST_ALIAS): New.
25512 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25513 (fold_loop_internal_call): ... this.
25514 (vect_loop_dist_alias_call): New function.
25515 (set_uid_loop_bbs): Call fold_loop_internal_call.
25516 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25519 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25522 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25523 Require dead FLAGS_REG at the beginning of a peephole.
25525 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25528 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25529 arguments in the call to __builtin_ia32_sbb_u32.
25530 (_subborrow_u64): Swap _X and _Y arguments in the call to
25531 __builtin_ia32_sbb_u64.
25533 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25536 * tree-vrp.c (compare_assert_loc): Turn into a function template
25537 with stable template parameter. Only test if a->e is NULL,
25538 !a->e == !b->e has been verified already. Use e == NULL or
25539 e != NULL instead of e or ! e tests. If stable is true, don't use
25540 iterative_hash_expr, on the other side allow a or b or both NULL
25541 and sort the NULLs last.
25542 (process_assert_insertions): Sort using compare_assert_loc<false>
25543 instead of compare_assert_loc, later sort using
25544 compare_assert_loc<true> before calling process_assert_insertions_for
25545 in a loop. Use break instead of continue once seen NULL pointer.
25547 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25549 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25550 Cortex-R7 and Cortex-R8 processors.
25552 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25554 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25555 uninitialized while src is not.
25557 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25559 * common/config/arm/arm-common.c: Adjust include path for
25561 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25562 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25563 (arm-cpu-data.h): Likewise.
25564 (arm-cpu-cdata.h): Likewise.
25565 * config/arm/arm-cpu.h: Delete.
25566 * config/arm/arm-cpu-cdata.h: Delete.
25567 * config/arm/arm-cpu-data.h: Delete.
25569 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25571 * config/arm/arm-cpus.in (cortex-a55): New.
25572 (cortex-a75): Likewise.
25573 (cortex-a75.cortex-a55): Likewise.
25574 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25576 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25577 * config/arm/arm-cpu-cdata.h: Regenerate.
25578 * config/arm/arm-cpu-data.h: Regenerate.
25579 * config/arm/arm-cpu.h: Regenerate.
25580 * config/arm/arm-tables.opt: Regenerate.
25581 * config/arm/arm-tune.md: Regenerate.
25583 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25585 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25587 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25589 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25592 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25594 PR tree-optimization/81292
25595 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25596 full_string_p, also call adjust_related_strinfos if the adjustment
25597 is simple, otherwise invalidate related strinfos.
25599 2017-07-04 Martin Liska <mliska@suse.cz>
25602 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25603 newly created variable as DECL_IGNORED_P.
25605 2017-07-04 Martin Liska <mliska@suse.cz>
25608 * ipa-inline.c (inline_small_functions):
25609 Use xstrdup_for_dump.
25611 2017-07-04 Tom de Vries <tom@codesourcery.com>
25613 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25615 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25618 * config/darwin.c (darwin_function_switched_text_sections):
25621 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25623 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25625 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25627 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25629 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25631 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25632 min_profitable_iters, and th as inclusive lower bounds.
25633 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25634 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25635 for min_profitable_iters and min_profitable_estimate.
25636 (vect_transform_loop): Treat th as an inclusive lower bound.
25637 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25639 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25642 * config/darwin.c (darwin_function_switched_text_sections):
25643 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25644 in two pieces, and suppress the use of buf.
25646 2017-07-03 Nathan Sidwell <nathan@acm.org>
25648 * hash-table.h (hash_table_mod1): Fix indentation.
25650 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25652 PR middle-end/81290
25653 * predict.c (force_edge_cold): Be more careful about propagation
25655 * profile-count.h (profile_probability::guessed,
25656 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25658 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25660 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25662 * doc/invoke.texi (rcpc architecture extension): Document it.
25664 2017-07-03 Richard Biener <rguenther@suse.de>
25666 PR tree-optimization/60510
25667 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25668 the scalar reduction PHI and use it.
25669 (vectorizable_reduction): Properly guard the single_defuse_cycle
25670 path for non-SLP reduction chains where we cannot use it.
25671 Rework reduc_def/index and vector type deduction. Rework
25672 vector operand gathering during reduction op code-gen.
25673 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25674 chains dissolve the chain and leave it to non-SLP reduction
25677 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25679 * tree-data-ref.h (dr_alignment): Declare.
25680 * tree-data-ref.c (dr_alignment): New function.
25681 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25682 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25684 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25686 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25688 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25689 and base_misalignment fields.
25690 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25691 * tree-data-ref.c: Include builtins.h.
25692 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25693 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25694 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25695 * tree-vect-data-refs.c: Include tree-cfg.h.
25696 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25697 fields instead of calculating an alignment here.
25698 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25699 innermost_loop_behavior fields.
25701 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25703 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25705 (DR_STEP_ALIGNMENT): New macro.
25706 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25707 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25708 (create_data_ref): Print it.
25709 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25710 to tell whether the step preserves vector (mis)alignment.
25711 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25712 Move the check for an integer step and generalise to all INTEGER_CST.
25713 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
25714 Print the outer step alignment.
25716 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25718 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
25719 with offset_alignment.
25720 (DR_ALIGNED_TO): Delete.
25721 (DR_OFFSET_ALIGNMENT): New macro.
25722 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
25723 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
25724 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
25725 (create_data_ref): Likewise.
25726 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25727 (vect_analyze_data_refs): Likewise.
25728 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
25729 creating dummy innermost behavior.
25731 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25733 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
25734 with a "innermost_loop_behavior *" and refeence tree.
25735 * tree-data-ref.c (dr_analyze_innermost): Likewise.
25736 (create_data_ref): Update call accordingly.
25737 * tree-predcom.c (find_looparound_phi): Likewise.
25739 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25741 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25742 fields with dr_wrt_vec_loop.
25743 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25744 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25745 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25746 (vect_dr_behavior): New function.
25747 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25748 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25749 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25750 track whether the step preserves the misalignment.
25751 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25752 Use vect_dr_behavior.
25753 (vect_setup_realignment): Update call accordingly.
25754 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25755 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25756 call to vect_create_addr_base_for_vector_ref.
25757 (vect_create_cond_for_align_checks): Likewise.
25758 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25759 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25760 (vect_recog_mask_conversion_pattern): Likewise.
25761 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25762 (new_stmt_vec_info): Remove redundant zeroing.
25764 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25766 * common/config/arm/arm-common.c (arm_be8_option): New function.
25767 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
25768 (ISA_ARMv6): Add isa_bit_be8.
25769 * config/arm/arm.h (arm_be8_option): Add prototype.
25770 (BE8_SPEC_FUNCTION): New define.
25771 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
25772 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
25773 (mlittle-endian): Similarly.
25774 (mbe8, mbe32): New options.
25775 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25776 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25778 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25780 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
25782 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25784 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
25785 (cleanup_tree_cfg_bb): Use it.
25786 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
25788 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
25790 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25793 * loop-doloop.c (add_test): Update profile.
25795 2017-07-03 Martin Liska <mliska@suse.cz>
25798 * sanopt.c (rewrite_usage_of_param): New function.
25799 (sanitize_rewrite_addressable_params): Likewise.
25800 (pass_sanopt::execute): Call rewrite_usage_of_param.
25802 2017-07-03 Richard Biener <rguenther@suse.de>
25804 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25805 back to using VIEW_CONVERT_EXPR.
25807 2017-07-03 Martin Liska <mliska@suse.cz>
25810 * doc/extend.texi: Document when a resolver function is
25811 generated for target_clones.
25813 2017-07-03 Martin Liska <mliska@suse.cz>
25815 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25816 and shadow memory of auto variables which are subject of
25817 use-after-scope sanitization.
25818 (asan_expand_mark_ifn): Add do set only when is_poison.
25820 2016-07-03 Richard Biener <rguenther@suse.de>
25822 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25824 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25825 (vectorizable_reduction): Perform reduction PHI creation when
25826 visiting a reduction PHI and adjust and simplify code generation
25827 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25828 (vect_transform_loop): Visit reduction PHIs.
25829 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25830 defs into the SLP tree.
25831 (vect_build_slp_tree): Reduction defs terminate the recursion.
25832 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25834 (vect_get_vec_defs_for_stmt_copy): Export.
25835 (vect_get_vec_defs): Likewise.
25836 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25838 (vect_get_vec_defs_for_stmt_copy): Declare.
25839 (vect_get_vec_defs): Likewise.
25841 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25843 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25844 parameter with a "loop" parameter and use it instead of the
25845 loop containing DR_STMT. Don't check simple_iv when doing
25846 BB analysis. Describe the two analysis modes in the comment.
25848 2017-07-03 Tom de Vries <tom@codesourcery.com>
25850 PR tree-optimization/69468
25851 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25852 (find_same_succ_bb): Handle ignore_edge_flags.
25854 2017-07-03 Tom de Vries <tom@codesourcery.com>
25856 PR tree-optimization/81192
25857 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25859 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25861 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25863 2017-07-03 Tom de Vries <tom@codesourcery.com>
25865 PR tree-optimization/81192
25866 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25867 BB_SAME_SUCC (bb) == NULL.
25869 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25871 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25874 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25876 * dumpfile.c: Include profile-count.h
25877 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25879 (insert_cond_bb): Update profile.
25880 * tree-cfg.h (insert_cond_bb): Update prototype.
25881 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25882 * tree-dump.c: Do not include tree-cfg.
25884 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25886 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25888 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25890 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25893 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25895 * tree-complex.c (expand_complex_div_wide): update profile.
25897 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25898 Alan Hayward <alan.hayward@arm.com>
25899 David Sherwood <david.sherwood@arm.com>
25901 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25902 (CORETYPES_H): New define.
25903 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25904 (insn-modes-inline.h, s-modes-inline-h): New rules.
25905 (generated_files): Add insn-modes-inline.h.
25906 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25907 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25908 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25909 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25910 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25911 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25912 (build/gencodes.o, build/genconditions.o): Likewise.
25913 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25914 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25915 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25916 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25917 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25918 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25919 * coretypes.h: Include everything up to real.h for generators.
25920 Include insn-modes.h first. Include wide-int-print.h after
25921 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25922 * machmode.h: Don't include insn-modes.h here.
25923 * function-tests.c: Remove includes of signop.h, machmode.h,
25924 double-int.h and wide-int.h.
25926 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25928 * optc-save-gen.awk: Likewise.
25929 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25930 * godump.c: Remove include of wide-int-print.h.
25931 * pretty-print.h: Likewise.
25932 * wide-int-print.cc: Likewise.
25933 * wide-int.cc: Likewise.
25934 * hash-map-tests.c: Remove include of signop.h.
25935 * hash-set-tests.c: Likewise.
25936 * rtl-tests.c: Likewise.
25937 * mkconfig.sh: Remove include of machmode.h.
25938 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25940 (emit_insn_modes_inline_h): ...this new function. Emit the code
25941 into an insn-modes-inline.h header file, adding appropriate
25942 include guards and end comments.
25943 (emit_insn_modes_c_header): Remove include of machmode.h.
25944 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25946 (main): Handle -i flag and call emit_insn_modes_inline_h when
25949 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25951 * tree-ssa-strlen.c (strinfo): Rename the length field to
25952 nonzero_chars. Add a full_string_p field.
25953 (compare_nonzero_chars, zero_length_string_p): New functions.
25954 (get_addr_stridx): Add an offset_out parameter.
25955 Use compare_nonzero_chars.
25956 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25957 (new_strinfo): Update after above changes to strinfo.
25958 (set_endptr_and_length): Set full_string_p.
25959 (get_string_length): Update after above changes to strinfo.
25960 (unshare_strinfo): Update call to new_strinfo.
25961 (maybe_invalidate): Likewise.
25962 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25963 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25964 as a uhwi instead of an shwi. Update after above changes to
25965 strinfo and new_strinfo.
25966 (zero_length_string): Assert that chainsi contains full strings.
25967 Use zero_length_string_p. Update call to new_strinfo.
25968 (adjust_related_strinfos): Update after above changes to strinfo.
25969 Copy full_string_p from origsi.
25970 (adjust_last_stmt): Use zero_length_string_p.
25971 (handle_builtin_strlen): Update after above changes to strinfo and
25972 new_strinfo. Install the lhs as the string length if the previous
25973 entry didn't describe a full string.
25974 (handle_builtin_strchr): Update after above changes to strinfo
25976 (handle_builtin_strcpy): Likewise.
25977 (handle_builtin_strcat): Likewise.
25978 (handle_builtin_malloc): Likewise.
25979 (handle_pointer_plus): Likewise.
25980 (handle_builtin_memcpy): Likewise. Track nonzero characters
25981 that aren't necessarily followed by a nul terminator.
25982 (handle_char_store): Likewise.
25984 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25986 PR tree-optimization/80769
25987 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25988 for malloc and calloc. Document the new invariant that all related
25989 strinfos have delayed lengths or none do.
25990 (verify_related_strinfos): Move earlier in file.
25991 (set_endptr_and_length): New function, split out from...
25992 (get_string_length): ...here. Also set the lengths of related
25994 (zero_length_string): Assert that chainsi has known (rather than
25996 (adjust_related_strinfos): Likewise.
25998 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26000 PR tree-optimization/81136
26001 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26002 assert that two references with the same misalignment have the same
26003 compile-time misalignment if those compile-time misalignments
26006 2017-07-01 Andi Kleen <ak@linux.intel.com>
26008 * print-tree.c (print_node): Print all attributes.
26010 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26012 * cfg.c (scale_bbs_frequencies): New function.
26013 * cfg.h (scale_bbs_frequencies): Declare it.
26014 * cfgloopanal.c (single_likely_exit): Cleanup.
26015 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26017 (scale_loop_profile): Likewise.
26018 (loop_version): Likewise.
26019 (create_empty_loop_on_edge): Update.
26020 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26021 scale_loop_frequencies, scale_loop_profile, loopify,
26022 loop_version): Update prototypes.
26023 * modulo-sched.c (sms_schedule): Update.
26024 * predict.c (unlikely_executed_edge_p): Also check probability.
26025 (probably_never_executed_edge_p): Fix typo.
26026 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26027 * tree-parloops.c (gen_parallel_loop): Update.
26028 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26029 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26030 * tree-ssa-loop-split.c (split_loop): Update.
26031 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26032 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26033 (vect_loop_versioning): Update.
26034 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26036 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26038 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26040 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26042 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26043 to keep profile consistent.
26045 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26047 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26048 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26049 * profile-count.h (max_safe_multiplier): Make unsigned.
26050 (profile_count::guessed_zero): New.
26052 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26054 * bb-reorder.c (fix_up_crossing_landing_pad,
26055 fix_crossing_conditional_branches): Use make_single_succ_edge
26056 to keep profile consistent.
26058 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26060 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26063 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26066 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26067 the right scopes, make sure cond_jump isn't preserved between multiple
26068 iterations. Search for fallthru edge whenever there are 3+ edges and
26069 use find_fallthru_edge for it.
26071 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26073 Patch by Alexander Monakov <amonakov@ispras.ru>
26074 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26075 probabilities consistently.
26077 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26079 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26080 emit_cmp_and_jump_insns.
26082 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26085 * tree-inline.c (expand_call_inline): Combine profile statuses.
26087 2017-06-30 Andrew Pinski <apinski@cavium.com>
26089 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26090 fold_stmt returned true.
26092 2017-06-30 Nathan Sidwell <nathan@acm.org>
26094 * ggc.h (empty_string): Delete.
26095 * cfgexpand.c (expand_asm_stmt): Use plain "".
26096 * optabs.c (expand_asm_memory_barrier): Likewise.
26097 * stringpool.c (empty_string): Delete.
26098 (digit_vector, digit_string): Delete.
26099 (ggc_alloc_string): Use plain "", don't optimize single digit
26100 strings. Use ggc_alloc_atomic.
26102 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26104 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26105 comparison set and one other set, use the cost of the non-comparison
26108 2017-06-30 Nathan Sidwell <nathan@acm.org>
26110 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26113 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26115 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26116 loops. Remove now unneeded calls to gimple_switch_set_label() that
26117 just set removed labels to NULL_TREE.
26119 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26121 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26122 (set_range_info): ...here. Only call set_range_info_raw if domain
26124 (set_nonzero_bits): Call set_range_info_raw.
26125 * tree-ssanames.h (set_range_info_raw): New.
26127 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26130 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26131 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26132 of nonimmediate_operand and <store_mask_constraint> instead of m
26133 for the input operand. For V8FI iterator, always split if input
26134 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26135 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26136 <store_mask_predicate> instead of register_operand and
26137 <store_mask_constraint> instead of v for the input operand. Make
26138 sure both operands aren't MEMs for if not <mask_applied>.
26140 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26142 * lto-wrapper.c (copy_file) Close both file descriptors before
26145 2017-06-30 Martin Liska <mliska@suse.cz>
26148 * multiple_target.c (create_dispatcher_calls): Make ifunc
26149 also for function that don't have calls or are not referenced.
26151 2017-06-30 Richard Biener <rguenther@suse.de>
26153 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26154 analyze the first scalar stmt. Move vector type computation
26155 for the BB case here from ...
26156 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26157 live operation processing in the SLP case properly.
26159 2017-06-30 Richard Biener <rguenther@suse.de>
26161 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26163 2017-06-30 Martin Liska <mliska@suse.cz>
26166 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26167 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26169 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26171 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26174 2017-06-30 Richard Biener <rguenther@suse.de>
26176 PR tree-optimization/81249
26177 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26178 condition reduction result to original scalar type.
26180 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26182 * profile-count.h (enum profile_quality): Fix typos and whitespace
26185 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26187 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26188 type for branch probabilities.
26190 2017-06-29 Julian Brown <julian@codesourcery.com>
26191 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26193 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26194 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26195 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26196 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26198 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26200 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26201 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26202 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26203 CC usage from generic code to here.
26204 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26205 CC usage into the target macros.
26207 2017-06-29 Maya Rashish <coypu@sdf.org>
26209 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26212 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26214 * arm/arm-builtins.c: Include profile-count.h
26215 * except.c (sjlj_emit_function_enter): Use
26216 profile_probability::unlikely.
26218 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26220 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26221 and tocrel_offset be pointer args rather than implicitly using
26223 (legitimate_constant_pool_address_p, rs6000_emit_move,
26224 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26225 tocrel_offset and use in toc_relative_expr_p call.
26226 (print_operand, print_operand_address): Use static tocrel_base_oac
26227 and tocrel_offset_oac.
26228 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26231 2017-06-29 Maya Rashish <coypu@sdf.org>
26233 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26235 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26237 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26238 objects, take into account only the alignment of 'op0' and 'mode1' if
26241 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26243 * ccmp.c (ccmp_tree_comparison_p): New function.
26244 (ccmp_candidate_p): Update to use above function.
26245 (get_compare_parts): New function.
26246 (expand_ccmp_next): Update to use new functions.
26247 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26249 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26250 take mode as argument.
26251 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26252 * expr.c (expand_expr_real_1): Pass mode as argument.
26254 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26256 * combine.c (combine_instructions): Print insns to dump_file, together
26259 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26261 * asan.c (asan_emit_stack_protection): Update.
26262 (create_cond_insert_point): Update.
26263 * auto-profile.c (afdo_propagate_circuit): Update.
26264 * basic-block.h (struct edge_def): Turn probability to
26265 profile_probability.
26266 (EDGE_FREQUENCY): Update.
26267 * bb-reorder.c (find_traces_1_round): Update.
26268 (better_edge_p): Update.
26269 (sanitize_hot_paths): Update.
26270 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26271 (make_single_succ_edge): Update.
26272 (check_bb_profile): Update.
26273 (dump_edge_info): Update.
26274 (update_bb_profile_for_threading): Update.
26275 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26276 probabilitycount to 0.
26277 * cfgbuild.c (compute_outgoing_frequencies): Update.
26278 * cfgcleanup.c (try_forward_edges): Update.
26279 (outgoing_edges_match): Update.
26280 (try_crossjump_to_edge): Update.
26281 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26282 (expand_gimple_tailcall): Update.
26283 (construct_init_block): Use make_single_succ_edge.
26284 (construct_exit_block): Use make_single_succ_edge.
26285 * cfghooks.c (verify_flow_info): Update.
26286 (redirect_edge_succ_nodup): Update.
26287 (split_edge): Update.
26288 (account_profile_record): Update.
26289 * cfgloopanal.c (single_likely_exit): Update.
26290 * cfgloopmanip.c (scale_loop_profile): Update.
26291 (set_zero_probability): Remove.
26292 (duplicate_loop_to_header_edge): Update.
26293 * cfgloopmanip.h (loop_version): Update prototype.
26294 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26295 (force_nonfallthru_and_redirect): Update.
26296 (update_br_prob_note): Update.
26297 (rtl_verify_edges): Update.
26298 (purge_dead_edges): Update.
26299 (rtl_lv_add_condition_to_bb): Update.
26300 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26301 * cgraphunit.c (init_lowered_empty_function): Update.
26302 (cgraph_node::expand_thunk): Update.
26303 * cilk-common.c: Include profile-count.h
26304 * dojump.c (inv): Remove.
26305 (jumpifnot): Update.
26306 (jumpifnot_1): Update.
26307 (do_jump_1): Update.
26309 (do_jump_by_parts_greater_rtx): Update.
26310 (do_compare_rtx_and_jump): Update.
26311 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26312 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26313 * dwarf2cfi.c: Include profile-count.h
26314 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26315 (sjlj_emit_dispatch_table): Likewise.
26316 * explow.c: Include profile-count.h
26317 * expmed.c (emit_store_flag_force): Update.
26318 (do_cmp_and_jump): Update.
26319 * expr.c (compare_by_pieces_d::generate): Update.
26320 (compare_by_pieces_d::finish_mode): Update.
26321 (emit_block_move_via_loop): Update.
26322 (store_expr_with_bounds): Update.
26323 (store_constructor): Update.
26324 (expand_expr_real_2): Update.
26325 (expand_expr_real_1): Update.
26326 * expr.h (try_casesi, try_tablejump): Update prototypes.
26327 * gimple-pretty-print.c (dump_probability): Update.
26328 (dump_profile): New.
26329 (dump_gimple_label): Update.
26330 (dump_gimple_bb_header): Update.
26331 * graph.c (draw_cfg_node_succ_edges): Update.
26332 * hsa-gen.c (convert_switch_statements): Update.
26333 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26334 (find_if_case_1): Update.
26335 (find_if_case_2): Update.
26336 * internal-fn.c (expand_arith_overflow_result_store): Update.
26337 (expand_addsub_overflow): Update.
26338 (expand_neg_overflow): Update.
26339 (expand_mul_overflow): Update.
26340 (expand_vector_ubsan_overflow): Update.
26341 * ipa-cp.c (good_cloning_opportunity_p): Update.
26342 * ipa-split.c (split_function): Use make_single_succ_edge.
26343 * ipa-utils.c (ipa_merge_profiles): Update.
26344 * loop-doloop.c (add_test): Update.
26345 (doloop_modify): Update.
26346 * loop-unroll.c (compare_and_jump_seq): Update.
26347 (unroll_loop_runtime_iterations): Update.
26348 * lra-constraints.c (lra_inheritance): Update.
26349 * lto-streamer-in.c (input_cfg): Update.
26350 * lto-streamer-out.c (output_cfg): Update.
26351 * mcf.c (adjust_cfg_counts): Update.
26352 * modulo-sched.c (sms_schedule): Update.
26353 * omp-expand.c (expand_omp_for_init_counts): Update.
26354 (extract_omp_for_update_vars): Update.
26355 (expand_omp_ordered_sink): Update.
26356 (expand_omp_for_ordered_loops): Update.
26357 (expand_omp_for_generic): Update.
26358 (expand_omp_for_static_nochunk): Update.
26359 (expand_omp_for_static_chunk): Update.
26360 (expand_cilk_for): Update.
26361 (expand_omp_simd): Update.
26362 (expand_omp_taskloop_for_outer): Update.
26363 (expand_omp_taskloop_for_inner): Update.
26364 * omp-simd-clone.c (simd_clone_adjust): Update.
26365 * optabs.c (expand_doubleword_shift): Update.
26366 (expand_abs): Update.
26367 (emit_cmp_and_jump_insn_1): Update.
26368 (expand_compare_and_swap_loop): Update.
26369 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26370 * predict.c (predictable_edge_p): Update.
26371 (edge_probability_reliable_p): Update.
26372 (set_even_probabilities): Update.
26373 (combine_predictions_for_insn): Update.
26374 (combine_predictions_for_bb): Update.
26375 (propagate_freq): Update.
26376 (estimate_bb_frequencies): Update.
26377 (force_edge_cold): Update.
26378 * profile-count.c (profile_count::dump): Add missing space into dump.
26379 (profile_count::debug): Add newline.
26380 (profile_count::differs_from_p): Explicitly convert to unsigned.
26381 (profile_count::stream_in): Update.
26382 (profile_probability::dump): New member function.
26383 (profile_probability::debug): New member function.
26384 (profile_probability::differs_from_p): New member function.
26385 (profile_probability::differs_lot_from_p): New member function.
26386 (profile_probability::stream_in): New member function.
26387 (profile_probability::stream_out): New member function.
26388 * profile-count.h (profile_count_quality): Rename to ...
26389 (profile_quality): ... this one.
26390 (profile_probability): New.
26391 (profile_count): Update.
26392 * profile.c (compute_branch_probabilities): Update.
26393 * recog.c (peep2_attempt): Update.
26394 * sched-ebb.c (schedule_ebbs): Update.
26395 * sched-rgn.c (find_single_block_region): Update.
26396 (compute_dom_prob_ps): Update.
26397 (schedule_region): Update.
26398 * sel-sched-ir.c (compute_succs_info): Update.
26399 * stmt.c (struct case_node): Update.
26400 (do_jump_if_equal): Update.
26401 (get_outgoing_edge_probs): Update.
26402 (conditional_probability): Update.
26403 (emit_case_dispatch_table): Update.
26404 (expand_case): Update.
26405 (expand_sjlj_dispatch_table): Update.
26406 (emit_case_nodes): Update.
26407 * targhooks.c: Update.
26408 * tracer.c (better_p): Update.
26409 (find_best_successor): Update.
26410 * trans-mem.c (expand_transaction): Update.
26411 * tree-call-cdce.c: Update.
26412 * tree-cfg.c (gimple_split_edge): Upate.
26413 (move_sese_region_to_fn): Upate.
26414 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26415 * tree-eh.c (lower_resx): Upate.
26416 (cleanup_empty_eh_move_lp): Upate.
26417 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26418 * tree-inline.c (copy_edges_for_bb): Update.
26419 (copy_cfg_body): Update.
26420 * tree-parloops.c (gen_parallel_loop): Update.
26421 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26422 (gimple_gen_time_profiler): Update.
26423 * tree-ssa-dce.c (remove_dead_stmt): Update.
26424 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26425 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26426 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26427 (unloop_loops): Update.
26428 (try_peel_loop): Update.
26429 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26430 * tree-ssa-loop-split.c (connect_loops): Update.
26431 (split_loop): Update.
26432 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26433 (hoist_guard): Update.
26434 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26435 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26436 (value_replacement): Update.
26437 * tree-ssa-reassoc.c (branch_fixup): Update.
26438 * tree-ssa-tail-merge.c (replace_block_by): Update.
26439 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26440 (create_edge_and_update_destination_phis): Update.
26441 (compute_path_counts): Update.
26442 (recompute_probabilities): Update.
26443 (update_joiner_offpath_counts): Update.
26444 (freqs_to_counts_path): Update.
26445 (duplicate_thread_path): Update.
26446 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26447 (struct switch_conv_info): Update.
26448 (gen_inbound_check): Update.
26449 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26450 (vect_do_peeling): Update.
26451 (vect_loop_versioning): Update.
26452 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26453 (optimize_mask_stores): Update.
26454 * ubsan.c (ubsan_expand_null_ifn): Update.
26455 * value-prof.c (gimple_divmod_fixed_value): Update.
26456 (gimple_divmod_fixed_value_transform): Update.
26457 (gimple_mod_pow2): Update.
26458 (gimple_mod_pow2_value_transform): Update.
26459 (gimple_mod_subtract): Update.
26460 (gimple_mod_subtract_transform): Update.
26461 (gimple_ic): Update.
26462 (gimple_stringop_fixed_value): Update.
26463 (gimple_stringops_transform): Update.
26464 * value-prof.h: Update.
26466 2017-06-29 Carl Love <cel@us.ibm.com>
26468 * config/rs6000/rs6000-c.c: Add support for built-in functions
26469 vector signed int vec_signed (vector float);
26470 vector signed long long vec_signed (vector double);
26471 vector signed int vec_signed2 (vector double, vector double);
26472 vector signed int vec_signede (vector double);
26473 vector signed int vec_signedo (vector double);
26474 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26475 instruction generator.
26476 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26477 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26478 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26480 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26481 vunsignede_v2df): Add define_expands.
26482 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26483 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26484 VEC_UNSIGNEDO): Add definitions.
26485 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26486 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26487 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26488 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26489 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26490 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26491 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26492 vec_unsignede, vec_unsignedo): Add builtin defines.
26493 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26495 * doc/extend.texi: Update the built-in documentation file for the
26496 new built-in functions.
26498 2017-06-29 Richard Biener <rguenther@suse.de>
26500 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26501 reduction chains to LOOP_VINFO_REDUCTIONS.
26502 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26503 SLP reductions after processing reduction chains.
26505 2017-06-29 Nathan Sidwell <nathan@acm.org>
26507 * builtins.c (fold_builtin_FUNCTION): Use
26508 lang_hooks.decl_printable_name.
26510 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26512 PR middle-end/81194
26513 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26514 with only one label.
26515 * stmt.c (expand_case): Assert NCASES is greater than one.
26517 2017-06-29 Richard Biener <rguenther@suse.de>
26519 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26521 (group_case_labels): Likewise.
26522 (find_taken_edge): Push sanity checking on val to workers...
26523 (find_taken_edge_cond_expr): ... here
26524 (find_taken_edge_switch_expr): ... and here, handle cases
26525 with just a default label.
26526 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26527 (group_case_labels): Likewise.
26528 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26529 group_case_labels does anything cleanup the CFG again.
26531 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26533 PR tree-optimization/81196
26534 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26535 exit condition comparing two IVs.
26537 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26539 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26540 profile to the dummy entry at the end of the list of architectures.
26541 * config/arm/arm-cpu-cdata.h: Regenerated.
26543 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26544 Michael Collison <michael.collison@arm.com>
26547 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26549 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26550 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26551 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26552 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26553 of shift when the shift amount is masked with constant equal to
26554 the size of the mode.
26555 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26558 2017-06-29 Martin Liska <mliska@suse.cz>
26560 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26562 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26565 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26568 (edge_growth_cache_entry::edge_growth_cache_entry): New
26570 (reset_edge_growth_cache): Update to use constructor.
26572 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26574 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26575 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26576 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26578 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26580 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26581 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26583 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26585 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26586 (*-linux-uclibc*): Add t-uclibc tmake_file.
26587 * config/t-musl: New.
26588 * config/t-uclibc: New.
26590 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26592 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26594 (gen_comm_data): Emit architectural setting of arch_prof.
26595 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26597 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26598 (armv8-m.base, armv8-m.main): Likewise.
26599 * arm-protos.h (arm_build_target): Add profile field.
26600 (arch_option): Likewise.
26601 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26603 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26604 arm_active_target.profile.
26606 2017-06-28 Richard Biener <rguenther@suse.de>
26608 PR middle-end/81227
26609 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26610 TYPE_OVERFLOW_WRAPS.
26611 * match.pd (negate_expr_p): Likewise.
26612 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26613 fold_build2, not fold_binary.
26615 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26617 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26618 Convert memory address to Pmode.
26619 (aarch64_print_operand): Assert MEM operands are always Pmode.
26621 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26624 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26625 Remove redundant if.
26626 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26627 * config/arm/aarch-common-protos.h
26628 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26629 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26631 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26634 * multiple_target.c (create_dispatcher_calls): Set the default
26635 clone to be static, not public.
26637 2017-06-28 Richard Biener <rguenther@suse.de>
26639 * tree-vect-loop.c (vectorizable_reduction): Move special
26640 cond reduction IV var creation ...
26641 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26642 parameter. Use STMT_VINFO_VECTYPE.
26643 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26646 2017-06-28 Martin Liska <mliska@suse.cz>
26649 * ipa-visibility.c (non_local_p): Handle visibility.
26651 2017-06-28 Martin Liska <mliska@suse.cz>
26654 * common.opt: Add IntegerRange to various options.
26655 * opt-functions.awk (integer_range_info): New function.
26656 * optc-gen.awk: Add integer_range_info to cl_options struct.
26657 * opts-common.c (decode_cmdline_option): Handle
26658 CL_ERR_INT_RANGE_ARG.
26659 (cmdline_handle_error): Likewise.
26660 * opts.c (print_filtered_help): Show valid interval in
26661 when --help is provided.
26662 * opts.h (struct cl_option): Add range_min and range_max fields.
26663 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26665 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26667 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26668 (x * C EQ/NE y * C): New transformation.
26670 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26672 * genmultilib (combination_space): Accept '+' in option names.
26674 2017-06-28 Martin Liska <mliska@suse.cz>
26677 * asan.c (instrument_derefs): Bail out inner references
26678 that are hard register variables.
26680 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26683 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26684 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26686 2017-06-28 Richard Biener <rguenther@suse.de>
26688 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26689 (vect_get_slp_defs): Adjust.
26690 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26692 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26694 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26695 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26696 (vectorizable_reduction): Adjust.
26697 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26699 (vect_get_slp_defs): Likewise.
26700 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26701 (vectorizable_bswap): Adjust.
26702 (vectorizable_call): Likewise.
26703 (vectorizable_conversion): Likewise.
26704 (vectorizable_assignment): Likewise.
26705 (vectorizable_shift): Likewise.
26706 (vectorizable_operation): Likewise.
26707 (vectorizable_store): Likewise.
26708 (vectorizable_condition): Likewise.
26709 (vectorizable_comparison): Likewise.
26711 2017-06-28 Michael Collison <michael.collison@arm.com>
26714 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
26716 (arm_gen_movmemqi): Removed unused variable 'i'.
26717 Convert 'for' loop into 'while' loop.
26718 (arm_expand_prologue): Remove last unnecessary set of insn.
26719 (thumb_pop): Remove unused variable 'pushed_words'.
26720 (thumb_exit): Remove last unnecessary set of regs_to_pop.
26722 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26724 * config/s390/predicates.md: Use s390_rel_address_ok_p.
26725 * config/s390/s390-protos.h: Add prototype of
26726 s390_rel_address_ok_p.
26727 * config/s390/s390.c (s390_got_symbol): New function.
26728 (s390_rel_address_ok_p): New function.
26729 (legitimize_pic_address): Use s390_rel_address_ok_p.
26730 (s390_load_got): Use s390_got_symbol.
26731 (s390_option_override): Issue error if
26732 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
26733 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
26735 * config/s390/s390.opt: New option mpic-data-is-text-relative.
26737 2017-06-27 Andrew Pinski <apinski@cavium.com>
26739 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
26740 (X * copysign (1.0, X)): New pattern.
26741 (X * copysign (1.0, -X)): New pattern.
26742 (copysign (-1.0, CST)): New pattern.
26744 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26746 * genmultilib (combination_space): Remove variable.
26747 Validate reuse rules against regular expression for any sequence
26748 of multilib options in any order.
26750 2017-06-27 Michael Collison <michael.collison@arm.com>
26752 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26753 call aarch64_split_simd_combine.
26754 * (aarch64_combine_internal<mode>): Delete pattern.
26755 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26756 Allow register and subreg operands.
26758 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26760 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26761 specific need, just fallback on defaults.
26762 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26764 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26765 Olivier Hainque <hainque@adacore.com>
26767 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
26769 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
26770 targets. Pick a default if no particular attempt applied.
26771 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26774 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26776 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26777 (x86_64-wrs-vxworks7): Likewise.
26779 2017-06-27 Marek Polacek <polacek@redhat.com>
26782 * ubsan.c (instrument_null): Check get_base_address's result for null.
26784 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26786 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
26788 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26790 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
26791 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
26792 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
26793 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
26794 New function types.
26795 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
26796 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
26797 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26798 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26799 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26800 BUILT_IN_FEUPDATEENV): New builtins.
26801 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26802 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26803 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26804 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26806 (builtin_structptr_types): Adjust size.
26807 * tree.c (builtin_structptr_types): Add four entries.
26809 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26810 Olivier Hainque <hainque@adacore.com>
26812 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26813 (TLS_SYM): New local macro, forcing reference to __tls__ on
26814 link command lines for VxWorks 7 RTPs, triggering initialization
26816 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26817 OS features TLS support, true for RTPs on VxWorks 7.
26818 * config/vxworks.c (vxworks_override_options): Setup emutls
26821 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26823 * predict.c (test_prediction_value_range): Use -1U instead of -1
26824 to avoid narrowing conversion warning.
26825 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26826 to avoid narrowing conversion warning.
26827 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26829 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26831 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26833 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26834 64bit configurations.
26835 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26836 (SIZE_TYPE): Likewise.
26837 * config/vxworks.c (vxworks_emutls_var_fields): Use
26838 long_unsigned_type_node instead of unsigned_type_node as the offset
26839 field type, which is "pointer" mode in emutls.c.
26841 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26844 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26846 PR middle-end/81207
26847 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26848 gimple_vuse copying separately from gimple_vdef copying.
26850 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26852 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26855 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26856 Olivier Hainque <hainque@adacore.com>
26858 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26859 for all vxworks7 targets.
26860 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26861 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26862 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26863 variations for VX6/VX7 and 32/64bits later on in ...
26864 (VXWORKS_LIB_SPEC): Leverage new macros.
26865 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26866 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26868 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26870 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26871 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26873 2017-06-26 Carl Love <cel@us.ibm.com>
26875 * config/rs6000/rs6000-c.c: Add support for built-in functions
26876 vector bool char vec_reve (vector bool char);
26877 vector signed char vec_reve (vector signed char);
26878 vector unsigned char vec_reve (vector unsigned char);
26879 vector bool int vec_reve (vector bool int);
26880 vector signed int vec_reve (vector signed int);
26881 vector unsigned int vec_reve (vector unsigned int);
26882 vector bool long long vec_reve (vector bool long long);
26883 vector signed long long vec_reve (vector signed long long);
26884 vector unsigned long long vec_reve (vector unsigned long long);
26885 vector bool short vec_reve (vector bool short);
26886 vector signed short vec_reve (vector signed short);
26887 vector double vec_reve (vector double);
26888 vector float vec_reve (vector float);
26889 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26890 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26891 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26892 (altivec_vreve): New pattern.
26893 * config/rs6000/altivec.h (vec_reve): New define.
26894 * doc/extend.texi (vec_rev): Update the built-in documentation file
26895 for the new built-in functions.
26897 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26899 PR tree-optimization/71815
26900 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26902 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26904 (slsr_process_phi): Likewise.
26905 (replace_uncond_cands_and_profitable_phis): Don't replace a
26906 multiply candidate with a stride of 1 (copy or cast).
26907 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26909 (lowest_cost_path): Likewise.
26910 (total_savings): Likewise.
26912 2017-06-26 Richard Biener <rguenther@suse.de>
26915 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26916 Use def_builtin_pure for all gather builtins.
26918 2017-06-26 Richard Biener <rguenther@suse.de>
26920 PR tree-optimization/81203
26921 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26922 non-dominating BBs.
26924 2017-06-26 Marek Polacek <polacek@redhat.com>
26927 * doc/invoke.texi: Document -Wmultistatement-macros.
26929 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26931 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26932 arm_neon_ok_no_float_abi effective target.
26934 2017-06-26 Richard Biener <rguenther@suse.de>
26936 PR tree-optimization/80928
26937 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26938 (copy_bbs): Set BB_DUPLICATED flag early.
26939 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26941 (execute_on_shrinking_pred): Likewise.
26942 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26943 BB_DUPLICATED blocks.
26944 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26945 iterate over all PHIs considering removal of *gsi.
26947 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26949 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26952 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26954 * config/rs6000/rs6000-string.c: (expand_block_clear,
26955 do_load_for_compare, select_block_compare_mode,
26956 compute_current_alignment, expand_block_compare,
26957 expand_strncmp_align_check, expand_strn_compare,
26958 expand_block_move, rs6000_output_load_multiple)
26959 Move functions related to string/block move/compare
26960 to a separate file.
26961 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26962 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26963 for this function which is now used in two files.
26964 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26965 * config.gcc: Add rs6000-string.o to extra_objs for
26966 targets powerpc*-*-* and rs6000*-*-*.
26968 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26971 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26972 32-bit, since indexed is not valid for DImode.
26973 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26974 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26975 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26976 (define_peephole2 for Altivec d-form store): Likewise.
26979 * multiple_target.c (create_dispatcher_calls): Only create the
26980 dispatcher call if the function is the default clone of a
26981 versioned function.
26983 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26985 PR middle-end/80902
26986 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26987 a call, force the call to not be a tail call.
26989 2017-06-23 Jeff Law <law@redhat.com>
26991 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26994 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26996 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26997 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26998 handling for early expansion of vector shifts (sl,sr,sra,rl).
26999 (builtin_function_type): Add vector shift right instructions
27000 to the unsigned argument list.
27002 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27004 rtl-optimizatoin/79286
27005 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27006 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27007 trap. PIC register plus a const unspec without offset can never trap.
27009 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27011 * tree.h (builtin_structptr_type): New type.
27012 (builtin_structptr_types): Declare new array.
27013 * tree.c (builtin_structptr_types): New array.
27014 (free_lang_data, build_common_tree_nodes): Use it.
27016 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27019 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27022 2017-06-22 Matt Turner <mattst88@gmail.com>
27024 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27025 Lake models to skylake case. Assume skylake for unknown
27026 models with clflushopt.
27028 2017-06-22 Jeff Law <law@redhat.com>
27030 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27031 frame sizes that do not satisfy aarch64_uimm12_shift.
27033 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27035 * profile-count.h (apply_probability,
27036 apply_scale, probability_in): Fix checks for zero.
27038 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27040 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27041 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27043 2016-06-22 Richard Biener <rguenther@suse.de>
27045 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27046 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27047 REDUC_MAX_EXPR support.
27048 (vectorizable_reduction): Likewise.
27049 (vect_create_epilog_for_reduction): Likewise.
27051 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27053 * match.pd (A / (1 << B) -> A >> B): New.
27054 * generic-match-head.c: Include optabs-tree.h.
27055 * gimple-match-head.c: Likewise.
27056 * optabs-tree.h (target_supports_op_p): New.
27057 * optabs-tree.c (target_supports_op_p): New.
27059 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27061 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27062 $gcc_cv_ld --help output.
27063 (gcc_cv_ld_demangle): Likewise.
27064 (gcc_cv_ld_eh_frame_hdr): Likewise.
27065 (gcc_cv_ld_pie): Likewise.
27066 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27067 (gcc_cv_ld_buildid): Likewise.
27068 (gcc_cv_ld_sysroot): Likewise.
27069 (ld_bndplt_support): Likewise.
27070 (ld_pushpopstate_support): Likewise.
27071 * configure: Regenerate.
27072 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27074 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27077 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27078 operands indexes to avoid gap between operands and match_dups.
27080 2017-06-21 Andrew Pinski <apinski@cavium.com>
27082 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27083 Increment Arith_shift and Arith_shift_reg by 1.
27084 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27086 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27087 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27088 (aarch64_strip_extend): Add new argument and test for it.
27089 (aarch64_cheap_mult_shift_p): New function.
27090 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27091 add a cost if it is true.
27092 Update calls to aarch64_strip_extend.
27093 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27095 2017-06-21 Andrew Pinski <apinski@cavium.com>
27097 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27099 (thunderxt88): Likewise.
27100 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27101 (thunderx_prefetch_tune): New variable.
27102 (thunderx2t99_prefetch_tune): Update for the correct values.
27103 (thunderxt88_tunings): New variable.
27104 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27105 generic_prefetch_tune.
27106 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27108 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27110 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27111 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27112 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27113 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27114 (aarch64_atomic_cas<mode>, GPI): Likewise.
27116 2017-06-21 Martin Liska <mliska@suse.cz>
27118 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27119 statements on cold and hot labels.
27120 * predict.c (tree_estimate_probability_bb): Remove the
27121 prediction from this place.
27123 2017-06-21 Martin Liska <mliska@suse.cz>
27125 PR tree-optimization/79489
27126 * gimplify.c (maybe_add_early_return_predict_stmt): New
27128 (gimplify_return_expr): Call the function.
27129 * predict.c (tree_estimate_probability_bb): Remove handling
27131 * predict.def: Update comment about early return predictor.
27132 * gimple-predict.h (is_gimple_predict): New function.
27133 * predict.def: Change default value of early return to 66.
27134 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27136 * passes.def: Put pass_strip_predict_hints to the beginning of
27139 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27141 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27142 FUNCTION_DECL declarations.
27143 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27145 (dwaf2out_decl): Likewise.
27146 * godump.c (go_early_global_decl): Skip call to the real debug hook
27147 for FUNCTION_DECL declarations.
27148 * passes.c (rest_of_decl_compilation): Skip call to the
27149 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27150 -fdump-go-spec is passed.
27152 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27154 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27155 Reorder for compactness.
27156 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27157 (def_builtin_pure, def_builtin_pure2): New functions.
27158 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27160 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27162 * match.pd (nop_convert): New predicate.
27163 ((A +- CST1) +- CST2): Allow some NOP conversions.
27165 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27168 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27169 with ctors/dtors if GOVD_SHARED is set.
27171 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27173 * config/aarch64/aarch64.md (movti_aarch64):
27174 Emit mov rather than orr.
27175 (movtf_aarch64): Likewise.
27176 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27177 Emit mov rather than orr.
27179 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27181 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27182 Swap alternatives, make integer dup more expensive.
27184 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27186 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27187 Return true for non-tls symbols.
27189 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27191 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27192 (cortex-a75): Likewise.
27193 (cortex-a75.cortex-a55): Likewise.
27194 * config/aarch64/aarch64-tune.md: Regenerate.
27195 * doc/invoke.texi (-mtune): Document new values for -mtune.
27197 2017-06-21 Tom de Vries <tom@codesourcery.com>
27199 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27200 stack_size feature.
27201 (Effective-Target Keywords, Other attributes): Suggest using
27202 dg-add-options stack_size feature to get stack limit in stack_size
27203 effective target documentation.
27205 2017-06-21 Julian Brown <julian@codesourcery.com>
27206 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27208 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27209 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27210 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27212 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27213 attribute type list for neon_multiply.
27214 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27215 attribute type list for neon_multiply.
27216 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27217 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27218 attribute type list for neon_multiply.
27219 * config/arm/types.md (crypto_pmull): Add.
27220 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27221 attribute type list.
27223 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27225 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27228 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27230 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27231 to make sure not to dereference a NULL cost_classes_ptr pointer.
27233 2017-06-20 Carl Love <cel@us.ibm.com>
27235 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27236 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27237 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27238 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27239 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27240 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27241 VMULOSW): New enum "unspec" values.
27242 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27243 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27244 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27245 altivec_vmulosw): New patterns.
27246 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27247 VMULOSW): Add definitions.
27249 2017-06-20 Julia Koval <julia.koval@intel.com>
27251 * config/i386/i386.c: Fix rounding expand for new pattern.
27252 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27254 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27256 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27257 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27259 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27261 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27264 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27266 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27267 architecture, then by implementer ID.
27268 * config/aarch64/aarch64-tune.md: Regenerate.
27270 2017-06-20 Richard Biener <rguenther@suse.de>
27272 PR middle-end/81097
27273 * fold-const.c (split_tree): Fold to type before negating.
27275 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27277 * diagnostic-show-locus.c
27278 (selftest::test_fixit_deletion_affecting_newline): New function.
27279 (selftest::diagnostic_show_locus_c_tests): Call it.
27281 2017-06-20 Andreas Schwab <schwab@suse.de>
27284 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27287 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27289 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27290 __ARM_FEATURE_COPROC according to support.
27292 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27294 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27295 Rewritten to avoid overflow for > 32-bit pointers.
27298 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27299 by removing enum keyword.
27300 (ubsan_type_descriptor): Likewise. Formatting fix.
27303 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27304 splitter): Require TARGET_SSE2 in the condition.
27306 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27309 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27310 for doing vector set of SFmode on ISA 3.0.
27311 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27312 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27314 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27315 SFmode value into a V4SF variable that was extracted from another
27316 V4SF variable without converting the element to double precision
27317 and back to single precision vector format.
27318 (vsx_insert_extract_v4sf_p9_2): Likewise.
27320 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27322 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27323 in UWHI to avoid undefined overflow.
27326 * ubsan.h (enum ubsan_encode_value_phase): New.
27327 (ubsan_encode_value): Change second argument to
27328 enum ubsan_encode_value_phase with default value of
27329 UBSAN_ENCODE_VALUE_GENERIC.
27330 * ubsan.c (ubsan_encode_value): Change second argument to
27331 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27332 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27333 create_tmp_var_raw instead of create_tmp_var and use a
27335 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27336 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27337 ubsan_encode_value callers.
27340 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27341 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27342 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27344 2017-06-19 Richard Biener <rguenther@suse.de>
27346 PR middle-end/81118
27347 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27348 estimates if we changed anything.
27350 2017-06-19 Richard Biener <rguenther@suse.de>
27352 PR tree-optimization/80887
27353 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27354 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27355 simplified lookups, then reset mprts_hook.
27356 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27358 (try_to_simplify): Likewise.
27360 2017-06-19 Martin Liska <mliska@suse.cz>
27363 * gimplify.c (gimplify_switch_expr):
27364 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27366 2017-06-19 Martin Liska <mliska@suse.cz>
27368 * doc/install.texi: Document that PGO runs in 4 stages.
27370 2017-06-19 Martin Liska <mliska@suse.cz>
27373 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27374 to dispatcher function name.
27375 * multiple_target.c (replace_function_decl): New function.
27376 (create_dispatcher_calls): Redirect both edges and references.
27378 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27380 * profile-count.c (profile_count::dump): Dump quality.
27381 (profile_count::differs_from_p): Update for unsigned val.
27382 * profile-count.h (profile_count_quality): New enum.
27383 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27385 2017-06-19 Richard Biener <rguenther@suse.de>
27387 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27388 struct function as arg.
27389 (estimate_numbers_of_iterations): Export overload with loop arg.
27390 (free_numbers_of_iterations_estimates_loop): Use an overload of
27391 free_numbers_of_iterations_estimates instead.
27392 * tree-cfg.c (remove_bb): Adjust.
27393 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27394 * tree-parloops.c (gen_parallel_loop): Likewise.
27395 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27397 (tree_unroll_loops_completely): Likewise.
27398 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27399 Use an overload instead and export.
27400 (estimated_loop_iterations): Adjust.
27401 (max_loop_iterations): Likewise.
27402 (likely_max_loop_iterations): Likewise.
27403 (estimate_numbers_of_iterations): Take struct function as arg
27405 (loop_exits_before_overflow): Adjust.
27406 (free_numbers_of_iterations_estimates_loop): Use an overload.
27407 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27408 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27410 2017-06-19 Richard Biener <rguenther@suse.de>
27413 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27414 RANGE_EXPR conservatively.
27416 2017-06-16 Carl Love <cel@us.ibm.com>
27418 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27419 definitions for vec_float, vec_float2, vec_floato,
27420 vec_floate built-ins.
27421 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27422 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27424 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27425 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27426 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27427 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27428 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27429 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27430 vec_floato): Add builtin defines.
27431 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27432 Update the built-in documentation file for the new built-in
27435 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27437 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27438 (mthumb): Mark as the negative of -marm.
27440 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27442 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27444 (ARM Options, -mtune): Document that this accepts the same
27445 extension options as -mcpu.
27446 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27448 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27450 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27451 permitted extensions.
27453 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27455 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27456 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27457 (armv8-m.main): Add option +nodsp.
27458 * config/arm/arm-cpu-cdata.h: Regenerated.
27460 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27462 * config/arm/t-fuchsia: New file.
27463 * config.gcc (arm*-*-fuchsia*): Use it.
27465 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27467 * config/arm/t-symbian: Rewrite for new option infrastructure.
27469 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27471 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27472 (MULTILIB_REQUIRED): Likewise.
27474 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27476 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27477 (MULTILIB_RESUE): Likewise.
27478 (MULTILIB_MATCHES): Likewise.
27479 (MULTLIB_REQUIRED): Likewise.
27481 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27483 * config/arm/t-rtems: Rewrite for new option framework.
27485 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27487 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27488 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27489 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27490 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27491 * config/arm/t-multilib: ... here.
27492 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27493 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27494 armv7-a and armv8*-a when A-profile libraries have not been built.
27495 * config/arm/t-rmprofile: Rewrite.
27497 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27499 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27500 with a backslash. Remove the backslash after substituting unescaped
27502 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27504 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27506 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27507 the driver to pass through the default setting of -mfloat-abi.
27508 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27510 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27511 (all_feat_combs): New rule.
27512 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27514 * config/arm/t-aprofile: Rewrite.
27516 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27518 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27519 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27520 fpu is not specified by the user/command-line.
27521 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27522 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27523 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27524 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27525 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27526 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27528 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27530 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27531 * config/arm/t-arm-elf: Rewritten.
27533 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27535 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27536 have some floating-point instructions.
27537 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27538 (TARGET_MAYBE_HARD_FLOAT): New macro.
27539 * config/arm/arm-builtins.c (arm_init_builtins): Use
27540 TARGET_MAYBE_HARD_FLOAT.
27541 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27543 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27545 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27546 (configargs.h): Include it.
27547 (arm_print_hint_for_fpu_option): New function.
27548 (arm_parse_fpu_option): New function.
27549 (candidate_extension): New class.
27550 (arm_canon_for_multilib): New function.
27551 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27552 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27553 (ARCH_CANONICAL_SPECS): New macro.
27554 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27556 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27558 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27559 are set after handling multilib fragments. Set target_cpu_default2
27562 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27564 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27566 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27567 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27569 (check_arch): Likewise.
27570 * config/arm/arm.c (arm_configure_build_target): Handle
27571 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27572 options in the default.
27574 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27576 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27577 when an option is an alias of another.
27578 * config/arm/parsecpu.awk (optalias): New parser token.
27579 (gen_comm_data): Mark non-alias options as such. Emit entries
27580 for extension aliases.
27581 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27582 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27583 (armv6kz, armv6zk, armv6t2): Likewise.
27584 (armv7): Make vfpv3-d16 an alias.
27585 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27587 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27588 Sort in canonical order.
27589 (armv8-a): Sort in canonical order.
27590 (armv8.1-a, armv8.2-a): Likewise.
27591 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27593 (cortex-a9): Sort in canonical order.
27594 * config/arm/arm.c (selftests.h): Include it.
27595 (arm_test_cpu_arch_data): New function.
27596 (arm_run_self_tests): New function.
27597 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27598 (targetm): Move declaration to the end of the file.
27599 * arm-cpu-cdata.h: Regenerated.
27601 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27603 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27604 call to target_mode_check describing the type of option passed.
27605 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27606 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27607 arm_parse_cpu_option_name to match parameters against list of
27609 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27610 arm_arch_core_flags data structure.
27611 * config/arm/arm-cpu_cdata.h: Regenerated.
27613 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27615 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27617 (arm_print_hint_for_cpu_option): Likewise.
27618 (arm_print_hint_for_arch_option): Likewise.
27619 (arm_parse_cpu_option_name): Likewise.
27620 (arm_parse_arch_option_name): Likewise.
27621 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27622 of entries in the all_fpus list.
27623 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27624 (arm_parse_cpu_option_name): Declare.
27625 (arm_parse_arch_option_name): Declare.
27626 (arm_parse_option_features): Declare.
27627 (arm_intialize_isa): Declare.
27628 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27630 (gen_comm_data): ... here. Make definitions non-static.
27631 * config/arm/arm-cpu-data.h: Regenerated.
27632 * config/arm/arm-cpu-cdata.h: Regenerated.
27634 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27636 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27637 (cpu_arch_extension): New structure.
27638 (cpu_arch_option, arch_option, cpu_option): New structures.
27639 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27640 architecture types.
27641 (gen_data): Generate new format data tables.
27642 * config/arm/arm.c (cpu_tune): New structure.
27643 (cpu_option, processors): Delete.
27644 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27645 (arm_print_hint_for_cpu_option): ... this and ...
27646 (arm_print_hint_for_arch_option): ... this.
27647 (arm_parse_arch_cpu_name): Delete. Replace with ...
27648 (arm_parse_cpu_option_name): ... this and ...
27649 (arm_parse_arch_option_name): ... this.
27650 (arm_unrecognized_feature): Change type of target parameter to
27652 (arm_parse_arch_cpu_features): Delete. Replace with ...
27653 (arm_parse_option_features): ... this.
27654 (arm_configure_build_target): Rework to use new configuration data
27656 (arm_print_tune_info): Rework for new configuration data tables.
27657 * config/arm/arm-cpu-data.h: Regenerated.
27658 * config/arm/arm-cpu.h: Regenerated.
27660 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27662 * Makefile.in (OBJS): Move sbitmap.o from here ...
27663 (OBJS-libcommon): ... to here.
27665 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27667 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27668 (ISA_ALL_CRYPTO): New macro.
27669 (ISA_ALL_SIMD): New macro
27670 (ISA_ALL_FP): New macro.
27671 * config/arm/arm.c (fpu_bitlist): Update initializer.
27672 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27674 (arm9e): Add fpu. Add option for nofp
27675 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27676 (arm926ej-s, arm1026ej-s): Likewise.
27677 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27678 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27679 neon-fp16, neon-vfpv4, nofp and nosimd.
27680 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27681 (cortex-a8): Add fpu. Add option for nofp.
27682 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27683 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27684 (cortex-r4f): Add fpu.
27685 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27686 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27688 (cortex-r8): Likewise.
27689 (cortex-m4): Add fpu. Add option for nofp.
27690 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27691 (cortex-a17.cortex-a7): Likewise.
27692 (cortex-a32): Add fpu. Add options for crypto and nofp.
27693 (cortex-a35, cortex-a53): Likewise.
27694 (cortex-a57): Add fpu. Add option for crypto.
27695 (cortex-a72, cortex-a73): Likewise.
27696 (exynos-m1): Likewise.
27697 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27698 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27699 (cortex-m33): Add fpu. Add option for nofp.
27700 * config/arm/arm-cpu-cdata.h: Regenerated
27701 * config/arm/arm-cpu-data.h: Regenerated.
27703 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27705 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27706 (armv5te, armv5tej): Likewise.
27707 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27708 (armv7): Add options fp and vfpv3-d16.
27709 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27710 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27712 (armv7ve): Likewise.
27713 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
27714 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
27715 (armv8-a): Add nocrypto option.
27716 (armv8.1-a, armv8.2-a): Likewise.
27717 (armv8-m.main): add options fp, fp.dp and nofp.
27719 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27721 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
27723 (armv8-a+crc): Delete.
27724 (armv8.1-a): Add options simd, crypto and nofp.
27725 (armv8.2-a): Add options fp16, simd, crypto and nofp.
27726 (armv8.2-a+fp16): Delete.
27727 (armv8-m.main): Add option dsp.
27728 (armv8-m.main+dsp): Delete.
27729 (cortex-a8): Add fpu. Add nofp option.
27730 (cortex-a9): Add fpu. Add nofp and nosimd options.
27731 * config/arm/parsecpu.awk (gen_data): Generate option tables and
27732 link to main cpu and architecture data structures.
27733 (gen_comm_data): Only put isa attributes from the main architecture
27735 (option): New statement for architecture and CPU entries.
27736 * arm.c (struct cpu_option): New structure.
27737 (struct processors): Add entry for options.
27738 (arm_unrecognized_feature): New function.
27739 (arm_parse_arch_cpu_name): Ignore any characters after the first
27741 (arm_parse_arch_cpu_feature): New function.
27742 (arm_configure_build_target): Separate out any CPU and architecture
27743 features and parse separately. Don't error out if -mfpu=auto is
27744 used with only an architecture string.
27745 (arm_print_asm_arch_directives): New function.
27746 (arm_file_start): Call it.
27747 * config/arm/arm-cpu-cdata.h: Regenerated.
27748 * config/arm/arm-cpu-data.h: Likewise.
27749 * config/arm/arm-tables.opt: Likewise.
27751 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27753 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27754 assembler when it is not -mfpu=auto.
27756 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27758 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27759 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27760 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27761 (ASM_CPU_SPEC): Rewrite.
27762 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27763 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27764 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27765 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
27766 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
27767 copied string is NUL-terminated. Also strip any characters prefixed
27769 (arm_rewrite_selected_arch): New function.
27770 (arm_rewrite_march): New function.
27772 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27774 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27775 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27776 (march, mcpu, mtune): Convert to string-based options.
27777 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27778 (arm_parse_arch_cpu_name): New function.
27779 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
27780 identify selected architecture or CPU.
27781 (arm_option_save): New function.
27782 (TARGET_OPTION_SAVE): Redefine.
27783 (arm_option_restore): Restore string options.
27784 (arm_option_print): Print string options.
27786 2017-06-16 Martin Sebor <msebor@redhat.com>
27788 PR tree-optimization/80933
27789 PR tree-optimization/80934
27790 * builtins.c (fold_builtin_3): Do not handle bcmp here.
27791 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
27792 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
27793 (gimple_fold_builtin): Call them.
27795 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27797 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27798 as unlikely; update profile.
27800 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27802 * predict.c (force_edge_cold): Handle declaring edges impossible
27805 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27807 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27809 (try_unroll_loop_completely): Fix reporting.
27811 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27813 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27815 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27818 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27819 if given a non-constant argument for an intrinsic which requires a
27822 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27824 * profile.c (compare_freqs): New function.
27825 (branch_prob): Sort edge list.
27826 (find_spanning_tree): Assume that the list is priority sorted.
27828 2017-06-16 Richard Biener <rguenther@suse.de>
27830 PR tree-optimization/81090
27831 * passes.def (pass_record_bounds): Remove.
27832 * tree-pass.h (make_pass_record_bounds): Likewise.
27833 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27834 make_pass_record_bounds): Likewise.
27835 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27836 not free niter estimates at the beginning but at the end.
27837 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27839 2017-06-16 Richard Biener <rguenther@suse.de>
27841 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27842 initializer to workaround ICE in host GCC 4.8.
27844 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27846 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27848 (clone_inlined_nodes): Update.
27850 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27852 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27853 prefetch settings, and enable prefetching by default at -O3.
27855 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27857 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27858 Set flag_prefetch_loop_arrays according to tuning data.
27860 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27862 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27863 New tune structure.
27864 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27865 [Unrelated to main purpose of the patch] Place the pointer field last
27866 to enable type checking errors when tune structure are wrongly merged.
27867 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27868 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27869 (thunderx2t99_prefetch_tune): New tune constants.
27870 (tune_params *_tunings): Update all tunings (no functional change).
27871 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27872 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27873 from tunings structures.
27875 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27878 * ubsan.c (instrument_null): Add T argument, use it instead
27879 of computing it based on IS_LHS.
27880 (instrument_object_size): Likewise.
27881 (pass_ubsan::execute): Adjust instrument_null and
27882 instrument_object_size callers to pass gimple_get_lhs or
27883 gimple_assign_rhs1 result to it. Use instrument_null instead of
27884 calling get_base_address and instrument_mem_ref. Handle
27885 aggregate call arguments for object-size sanitization.
27887 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27889 PR tree-optimization/81089
27890 * tree-vrp.c (is_masked_range_test): Validate operands of
27893 2017-06-15 Martin Sebor <msebor@redhat.com>
27896 * dumpfile.c (dump_register): Avoid calling memset to initialize
27897 a class with a default ctor.
27898 * gcc.c (struct compiler): Remove const qualification.
27899 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27900 * hash-table.h: Ditto.
27901 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27903 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27904 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27906 * params.h (struct param_info): Make struct members non-const.
27907 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27908 with default initialization.
27909 * vec.h (vec_copy_construct, vec_default_construct): New helper
27911 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27912 with vec_copy_construct.
27913 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27914 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27915 * doc/invoke.texi (-Wclass-memaccess): Document.
27917 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27919 * emit-rtl.h (is_leaf): Update comment about local
27920 register allocator.
27922 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27925 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27926 for a variable to have a section before checking if the section has a
27928 Set section to.persistent if persistent attribute is set.
27929 Warn if .persistent attribute is used on an automatic variable.
27931 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27933 PR rtl-optimization/80474
27934 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27936 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27938 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27941 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27943 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27944 (casesi): Emit bounds checking as RTL.
27945 (casesi_internal_mips16_<mode>): Remove bounds checking.
27947 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27949 * config/xtensa/xtensa.c (xtensa_option_override): Append
27950 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27951 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27952 xtensa_doloop_hooks): Define unconditionally.
27953 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27955 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27956 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27957 for it in xtensa_option_override.
27958 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27959 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27961 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27963 * doc/cppopts.texi: Document '-' special value to -MF.
27965 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27967 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27968 (cortex_a53_fconst): Likewise.
27969 (cortex_a53_fpmul): Likewise.
27970 (cortex_a53_f_load_64): Likewise.
27971 (cortex_a53_f_load_many): Likewise.
27972 (cortex_a53_advsimd_alu): Likewise.
27973 (cortex_a53_advsimd_alu_q): Likewise.
27974 (cortex_a53_advsimd_mul): Likewise.
27975 (cortex_a53_advsimd_mul_q): Likewise.
27976 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27977 Add missing fmul, r2f_cvt and fconst cases.
27979 2017-06-14 Richard Biener <rguenther@suse.de>
27981 PR middle-end/81088
27982 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27984 (fold_binary_loc): When associating do not treat pre-existing
27985 TREE_OVERFLOW on literal constants as a reason to allow
27986 TREE_OVERFLOW on associated literal constants.
27988 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27990 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27991 (MASK_FEATURES): New macro.
27992 * config/sparc/sparc.c (sparc_option_override): Remove the special
27993 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27995 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27997 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27998 a division of 0 if non-call exceptions are enabled.
28000 2017-06-14 Andrew Pinski <apinski@cavium.com>
28001 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28004 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28005 Improve vector initialization code gen for only variable case.
28007 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28009 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28011 2017-06-14 Richard Biener <rguenther@suse.de>
28013 PR tree-optimization/81083
28014 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28017 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28019 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28020 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28021 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28022 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28023 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28024 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28026 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28028 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28029 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28031 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28033 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28035 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28037 * config/rs6000/t-rtems: Don't handle SPE.
28039 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28041 * config/rs6000/t-linux: Don't handle SPE.
28043 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28045 * config/rs6000/eabispe.h: Delete file.
28047 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28049 * config/rs6000/t-spe: Delete file.
28051 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28053 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28054 (rs6000_legitimate_offset_address_p): Return false for anything in
28055 V2SImode or V2SFmode.
28057 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28059 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28060 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28061 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28063 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28064 (rs6000_legitimate_offset_address_p): Ditto.
28065 (rs6000_emit_move): Ditto.
28066 (rs6000_init_builtins): Remove V4HI_type_node.
28068 2017-06-13 Martin Liska <mliska@suse.cz>
28071 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28072 (gate_asan): Likewise.
28073 * asan.h (asan_no_sanitize_address_p): Remove the function.
28074 (sanitize_flags_p): New function.
28075 * builtins.def: Fix coding style.
28076 * common.opt: Use renamed enum value.
28077 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28078 * doc/extend.texi: Document no_sanitize attribute.
28079 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28080 to SANITIZE_UNDEFINED_NONDEFAULT.
28081 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28082 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28083 Use sanitize_flags_p.
28084 * gimplify.c (gimplify_function_tree): Likewise.
28085 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28086 * opts.c (parse_no_sanitize_attribute): New function.
28087 (common_handle_option): Use renamed enum value.
28088 * opts.h (parse_no_sanitize_attribute): Declare.
28089 * tree.c (sanitize_flags_p): New function.
28090 * tree.h: Declared here.
28091 * tsan.c: Use sanitize_flags_p.
28092 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28093 (instrument_mem_ref): Likewise.
28094 (instrument_bool_enum_load): Likewise.
28095 (do_ubsan_in_current_function): Remove the function.
28096 (pass_ubsan::execute): Use sanitize_flags_p.
28097 * ubsan.h: Remove do_ubsan_in_current_function
28098 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28099 (dump_function_to_file): Use it here.
28101 2017-06-13 Martin Jambor <mjambor@suse.cz>
28103 PR tree-optimization/80803
28104 PR tree-optimization/81063
28105 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28106 (propagate_subaccesses_across_link): Enqueue subtree whenever
28107 necessary instead of relying on the caller.
28109 2017-06-13 Martin Jambor <mjambor@suse.cz>
28111 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28112 that have a first_link.
28113 (sort_and_splice_var_accesses): Do not check first_link before
28115 (subtree_mark_written_and_enqueue): Likewise.
28116 (propagate_all_subaccesses): Likewise and do not stop at first
28117 parent with a first_link.
28119 2017-06-13 Martin Jambor <mjambor@suse.cz>
28121 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28124 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28126 * match.pd: New pattern.
28128 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28130 * tree-vrp.c (is_masked_range_test): New function.
28131 (register_edge_assert_for): Determine ranges for
28134 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28136 PR tree-optimization/67328
28137 * fold-const.c (maskable_range_p): New function.
28138 (build_range_check): Generate bittests if possible.
28140 2017-06-13 Martin Liska <mliska@suse.cz>
28142 * gimple-pretty-print.c (dump_probability): Add new argument.
28143 (dump_edge_probability): Dump both probability and count.
28144 (dump_gimple_label): Likewise.
28145 (dump_gimple_bb_header): Likewise.
28147 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28150 * config/avr/avr-devices.c: Fix indentation.
28151 * config/avr/gen-avr-mmcu-specs.c: Dito.
28153 2017-06-13 Richard Biener <rguenther@suse.de>
28155 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28156 instead get vector type from stmt_info.
28157 (vectorizable_reduction): Adjust. Remove dead code.
28159 2017-06-13 Richard Biener <rguenther@suse.de>
28161 PR middle-end/81065
28162 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28163 case of C * (x * C2 + C3).
28164 (fold_addr_of_array_ref_difference): Properly fold index difference.
28166 2017-06-12 David S. Miller <davem@davemloft.net>
28169 * config/sparc/sparc.md (return expander): Emit frame blockage if
28170 function uses alloca.
28172 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28174 * combine.c (make_field_assignment): Check len rather than the mode
28175 precision when calling force_to_mode.
28177 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28179 Support multilibs and devices that see flash in RAM address range.
28182 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28183 (avr_mcu_t) <flash_pm_offset>: New field.
28184 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28185 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28186 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28187 (AVR_TINY_PM_OFFSET): Remove macro.
28188 * config/avr/avr.opt (-mshort-calls): New option.
28189 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28190 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28191 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28192 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28193 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28194 instead of avr_arch->have_jmp_call.
28195 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28196 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28197 avr_arch->flash_pm_offset to define.
28198 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28199 new field flash_pm_offset. Add entry for avrxmega3.
28200 (avr_texinfo): Add entry for avrxmega3.
28201 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28202 attiny212, attiny214,
28203 attiny412, attiny414, attiny416, attiny417,
28204 attiny814, attiny816, attiny817,
28205 attiny1614, attiny1616, attiny1617,
28206 attiny3214, attiny3216, attiny3217.
28207 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28208 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28209 (avr_print_operand_address) [AVR_TINY]: Same.
28210 (avr_asm_init_sections) <readonly_data_section>: Only patch
28211 callback if avr_arch->flash_pm_offset = 0.
28212 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28213 for rodata if avr_arch->flash_pm_offset != 0.
28214 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28215 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28216 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28217 (m_options): Append opt_rcall.
28218 (m_dirnames): Append dir_rcall.
28219 * config/avr/t-multilib: Regenerate.
28221 * configure.ac [target=avr]: Check whether avrxmega3 default
28222 linker description file works as needed.
28223 * configure: Regenerate.
28224 * doc/avr-mmcu.texi: Regenerate.
28225 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28226 <__AVR_ARCH__>: Document avrxmega3 and 103.
28227 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28228 <__AVR_SHORT_CALLS__>: Document it.
28229 <__AVR_PM_BASE_ADDRESS__>: Document it.
28230 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28231 (AVR Variable Attributes) <progmem>: Document this is
28232 not needed for avrxmega3.
28233 (AVR Named Address Spaces) <__flash>: Dito.
28235 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28237 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28239 2017-06-12 Doug Rupp <rupp@adacore.com>
28241 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28242 Append vxworks-stdint.h to the tm_file list.
28243 * config/vxworks-stdint.h: New file.
28245 2017-06-12 Martin Liska <mliska@suse.cz>
28247 PR tree-optimization/81041
28248 * tree-profile.c (gimple_gen_ic_func_profiler):
28249 Create an extra BB in profile-generate
28250 (gimple_gen_time_profiler): Likewise.
28252 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28254 PR tree-optimization/81003
28255 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28256 (update_range_test): Use it instead of force_gimple_operand_gsi.
28258 2017-06-12 Richard Biener <rguenther@suse.de>
28260 PR tree-optimization/81053
28261 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28262 with backedge value not defined in loop. Simplify def stmt
28265 2017-06-11 Tom de Vries <tom@codesourcery.com>
28268 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28270 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28271 nvptx_cannot_force_const_mem.
28273 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28275 * opts.c (finish_options): Move test for flag_split_stack after
28276 it has been initialized.
28278 2017-06-11 Jason Merrill <jason@redhat.com>
28280 * tree.h (id_equal): New.
28281 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28282 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28283 instead of strcmp of IDENTIFIER_POINTER.
28285 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28287 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28288 (mark_all_inlined_calls_cdtor): Fix formating.
28289 (inline_transform): Rescale profile before inlining.
28291 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28293 * cgraph.h (cgraph_edge::clone): Update prototype.
28294 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28295 (cgraph_node::create_clone): Update.
28296 (cgraph_node::create_version_clone): Update.
28297 * tree-inline.c (copy_bb): Update.
28298 (expand_call_inline): Update.
28300 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28302 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28303 factored out from ...
28304 (rs6000_emit_prologue): ... here.
28306 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28308 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28309 factored out from ...
28310 (rs6000_emit_prologue): ... here.
28312 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28314 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28316 (handle_missing_profiles): Fix computation of tp_first_run.
28317 (counts_to_freqs): Do not touch freqs when count is 0.
28319 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28321 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28324 2017-06-10 Tom de Vries <tom@codesourcery.com>
28326 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28327 attributes): Document signal effective target.
28329 2017-06-10 Tom de Vries <tom@codesourcery.com>
28331 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28332 Document effective target stack_size.
28334 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28336 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28337 to the edit_context if they can be auto-applied.
28339 2017-06-9 Ian Lance Taylor <iant@golang.org>
28341 * opts.c (finish_options): If -fsplit-stack, disable implicit
28342 -forder-blocks-and-partition.
28343 * doc/invoke.texi (Optimize Options): Document that when using
28344 -fsplit-stack -forder-blocks-and-partition is not implicitly
28347 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28349 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28350 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28351 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28352 * builtins.def (abort, trap, unreachable): Declare cold.
28353 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28354 * tree-core.h (ECF_COLD): New.
28355 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28356 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28358 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28360 * predict.c (unlikely_executed_stmt_p): Cleanup.
28362 2017-06-09 Richard Biener <rguenther@suse.de>
28364 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28365 model if the ref is always written to.
28367 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28369 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28371 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28373 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28376 2017-06-09 Tom de Vries <tom@codesourcery.com>
28379 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28380 "target cannot support label values" when encountering LABEL_REF.
28382 2017-06-09 Martin Liska <mliska@suse.cz>
28384 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28385 (gimple_gen_ic_func_profiler): Emit direct comparison
28386 of __gcov_indirect_call_callee with NULL.
28387 (gimple_gen_time_profiler): Change probability from
28388 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28390 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28392 * profile.c (edge_gcov_counts): Turn to pointer.
28393 (compute_branch_probabilities, compute_branch_probabilities): Update.
28394 (branch_prob): Do not clear edge_gcov_count.
28395 * profile.h (edge_gcov_counts): Turn to pointer.
28396 (edge_gcov_count): Update.
28398 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28400 * gimple.h (gimple_check_failed): Mark cold.
28402 2017-06-09 Richard Biener <rguenther@suse.de>
28404 PR tree-optimization/66623
28405 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28406 refactor check_reduction into two parts, properly computing
28407 whether we have to check reduction validity for outer loop
28410 2017-06-09 Richard Biener <rguenther@suse.de>
28412 PR tree-optimization/79483
28413 * graphite-scop-detection.c (order): New global.
28414 (get_order): Compute bb to order mapping that satisfies code
28415 generation constraints.
28416 (cmp_pbbs): New helper.
28417 (build_scops): Start domwalk at entry block, sort generated
28420 2017-06-09 Richard Biener <rguenther@suse.de>
28422 PR middle-end/81007
28423 * ipa-polymorphic-call.c
28424 (ipa_polymorphic_call_context::restrict_to_inner_class):
28425 Skip FIELD_DECLs with error_mark_node type.
28426 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28429 2017-06-09 Martin Liska <mliska@suse.cz>
28431 * predict.c (struct branch_predictor): New struct.
28432 (test_prediction_value_range): New test.
28433 (predict_c_tests): New function.
28434 * selftest-run-tests.c (selftest::run_tests): Run the function.
28435 * selftest.h: Declare new tests.
28437 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28440 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28441 gen_add3_insn did not fail.
28442 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28443 r0, construct that number in a temporary reg and add that reg to r0.
28444 If asked to put the result in r0 as well, fail.
28446 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28448 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28449 for early expansion of vec_eqv.
28451 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28453 PR middle-end/81005
28454 * ubsan.c (instrument_null): Avoid pointless code temporary.
28455 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28457 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28461 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28464 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28467 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28469 * predict.c (unlikely_executed_edge_p): Move ahead.
28470 (probably_never_executed_edge_p): Use it.
28472 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28474 PR middle-end/79988
28475 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28476 gimple_call_builtin_p call.
28478 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28480 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28481 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28482 rtl_check_failed_type2, rtl_check_failed_code1,
28483 rtl_check_failed_code2, rtl_check_failed_code_mode,
28484 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28485 rtvec_check_failed_bounds, rtl_check_failed_flag,
28486 _fatal_insn_not_found, _fatal_insn): Likewise.
28487 * tree.h (tree_contains_struct_check_failed,
28488 tree_check_failed, tree_not_check_failed,
28489 tree_class_check_failed, tree_range_check_failed,
28490 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28491 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28492 tree_operand_check_failed, omp_clause_check_failed,
28493 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28496 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28498 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28499 flag_branch_probabilities.
28500 * ipa-inline.c (edge_badness): Likewise.
28501 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28502 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28503 * predict.c (maybe_hot_frequency_p): Likewise.
28504 (probably_never_executed): Likewise.
28505 * sched-ebb.c (schedule_ebbs): Likewise.
28506 * sched-rgn.c (find_single_block_region): Likewise.
28507 * tracer.c (tail_duplicate): Likewise.
28509 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28511 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28512 longer requires x_flag_profile_use.
28514 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28516 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28517 instead of flag_reorder_blocks_and_partition.
28518 * dbxout.c (dbxout_function_end): Likewise.
28519 * dwarf2out.c (gen_subprogram_die): Likewise.
28520 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28521 * hw-doloop.c (reorg_loops): Likewise.
28522 * varasm.c (assemble_start_function,
28523 assemble_end_function): Likewise.
28524 (decide_function_section): Do not check for
28525 flag_reorder_blocks_and_partition.
28527 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28529 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28531 (chkp_get_hard_register_fake_addr_expr): Ditto.
28532 (chkp_build_addr_expr): Add check for hard reg case.
28533 (chkp_parse_array_and_component_ref): Ditto.
28534 (chkp_find_bounds_1): Ditto.
28535 (chkp_process_stmt): Don't generate bounds store for
28538 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28540 * predict.c (maybe_hot_bb_p): Do not check profile status.
28541 (maybe_hot_edge_p): Likewise.
28542 (probably_never_executed): Check for zero counts even if profile
28544 (unlikely_executed_edge_p): New function.
28545 (unlikely_executed_stmt_p): New function.
28546 (unlikely_executed_bb_p): New function.
28547 (set_even_probabilities): Use unlikely predicates.
28548 (combine_predictions_for_bb): Likewise.
28549 (predict_paths_for_bb): Likewise.
28550 (predict_paths_leading_to_edge): Likewise.
28551 (determine_unlikely_bbs): New function.
28552 (estimate_bb_frequencies): Use it.
28553 (compute_function_frequency): Use zero counts even if profile is
28555 * profile-count.h: Fix typo.
28557 2017-08-08 Julia Koval <julia.koval@intel.com>
28559 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28560 _mm512_mask_cvtsepi16_storeu_epi8,
28561 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28562 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28563 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28564 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28565 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28566 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28567 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28568 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28569 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28570 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28571 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28572 __builtin_ia32_pmovuswb256mem_mask,
28573 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28574 __builtin_ia32_pmovwb512mem_mask): New builtins.
28576 2017-08-08 Julia Koval <julia.koval@intel.com>
28578 PR target/73350,80862
28579 * config/i386/subst.md (round): Fix round pattern.
28580 * config/i386/i386.c (ix86_erase_embedded_rounding):
28581 Fix erasing rounding for the fixed pattern.
28583 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28585 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28587 2017-06-08 Martin Liska <mliska@suse.cz>
28589 PR gcov-profile/80911
28590 * gcov.c (block_info::block_info): New constructor.
28592 2017-06-07 Carl Love <cel@us.ibm.com>
28594 * config/rs6000/rs6000-c: The return type of the following
28595 built-in functions was implemented as int not long long. Fix sign
28596 of return value for the unsigned version of vec_mulo and vec_mule.
28597 vector unsigned long long vec_bperm (vector unsigned long long,
28598 vector unsigned char)
28599 vector signed long long vec_mule (vector signed int,
28601 vector unsigned long long vec_mule (vector unsigned int,
28602 vector unsigned int)
28603 vector signed long long vec_mulo (vector signed int,
28605 vector unsigned long long vec_mulo (vector unsigned int,
28606 vector unsigned int)
28607 * doc/extend.texi: Fix the documentation for the built-in
28610 2017-06-07 Carl Love <cel@us.ibm.com>
28613 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28616 2017-06-07 Carl Love <cel@us.ibm.com>
28618 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28619 support, Generate doublehv for signed int/float for BE case only.
28621 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28623 * doc/invoke.texi (mcx16): Rewrite.
28625 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28627 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28628 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28629 *mov<mode>_softfloat, and an anonymous splitter): Use
28630 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28632 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28634 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28636 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28637 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28638 (rs6000_debug_reg_global): Adjust.
28639 (rs6000_init_hard_regno_mode_ok): Adjust.
28640 (rs6000_dbx_register_number): Adjust.
28641 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28642 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28643 Remove SPE_ACC and SPEFSCR.
28644 (REG_ALLOC_ORDER): Ditto.
28645 (FRAME_POINTER_REGNUM): Change to 111.
28646 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28647 (REG_CLASS_NAMES): Ditto.
28648 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28649 (REGISTER_NAMES): Ditto.
28650 (ADDITIONAL_REG_NAMES): Ditto.
28651 (rs6000_reg_names): Ditto.
28652 * config/rs6000/rs6000.md: Renumber some register number
28655 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28657 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28659 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28660 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28662 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28663 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28664 Delete the SPE high registers.
28665 (REG_ALLOC_ORDER): Ditto.
28666 (enum reg_class): Remove SPE_HIGH_REGS.
28667 (REG_CLASS_NAMES): Ditto.
28668 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28669 (REGISTER_NAMES): Ditto.
28670 (rs6000_reg_names): Ditto.
28671 * doc/tm.texi.in: Remove SPE as example.
28672 * doc/tm.texi: Regenerate.
28674 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28676 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28677 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28678 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28679 * config/rs6000/rs6000.md (type): Remove "brinc".
28681 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28683 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28684 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28685 * config/rs6000/linuxspe.h: Delete file.
28686 * config/rs6000/rs6000.md: Don't include spe.md.
28687 * config/rs6000/spe.h: Delete file.
28688 * config/rs6000/spe.md: Delete file.
28689 * config/rs6000/t-rs6000: Remove spe.md.
28691 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28693 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28694 (reg_or_none500mem_operand): Delete.
28695 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28696 instead of reg_or_none500mem_operand.
28698 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28700 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28701 handling of SPE flags.
28702 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28704 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28706 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28708 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28709 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28710 paired_divv2sf3): Similar.
28711 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28712 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
28713 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
28715 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
28716 Rename the paired_* instruction patterns.
28717 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
28719 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
28720 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
28721 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
28722 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
28723 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
28724 PAIRED_VECTOR_MODE.
28725 (struct machine_function): Delete field spe_insn_chain_scanned_p.
28726 (spe_func_has_64bit_regs_p): Delete.
28727 (spe_expand_predicate_builtin): Delete.
28728 (spe_expand_evsel_builtin): Delete.
28729 (TARGET_DWARF_REGISTER_SPAN): Do not define.
28730 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
28731 (invalid_e500_subreg): Delete.
28732 (rs6000_legitimize_address): Always force_reg op2 as well, for
28733 paired single memory accesses.
28734 (rs6000_member_type_forces_blk): Delete.
28735 (rs6000_spe_function_arg): Delete.
28736 (rs6000_expand_unop_builtin): Delete SPE handling.
28737 (rs6000_expand_binop_builtin): Ditto.
28738 (spe_expand_stv_builtin): Delete.
28739 (bdesc_2arg_spe): Delete.
28740 (spe_expand_builtin): Delete.
28741 (spe_expand_predicate_builtin): Delete.
28742 (spe_expand_evsel_builtin): Delete.
28743 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28744 (spe_init_builtins): Delete.
28745 (spe_func_has_64bit_regs_p): Delete.
28746 (savres_routine_name): Delete "info" parameter. Adjust callers.
28747 (rs6000_emit_stack_reset): Ditto.
28748 (rs6000_dwarf_register_span): Delete.
28749 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28750 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28751 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28753 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28755 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28756 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28757 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28758 mulv2sf3, divv2sf3): Delete expanders.
28760 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28762 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28764 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28766 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
28767 * config/rs6000/rs6000.c: Ditto.
28769 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28771 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
28772 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
28773 comparison_operator.
28775 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28777 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28778 * config/rs6000/rs6000.opt: Ditto.
28779 * config/rs6000/t-rtems: Ditto.
28781 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28783 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
28784 TARGET_E500_SINGLE by 0, simplify.
28785 * config/rs6000/rs6000.c: Ditto.
28786 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
28787 (spe_build_register_parallel): Delete.
28788 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
28789 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
28790 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
28791 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
28792 (E500_CONVERT): Delete.
28793 * config/rs6000/spe.md: Remove many patterns and all define_constants.
28795 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28797 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28798 * config/rs6000/dfp.md: Ditto.
28799 (negdd2, *negdd2_fpr): Merge.
28800 (absdd2, *absdd2_fpr): Merge.
28801 (negtd2, *negtd2_fpr): Merge.
28802 (abstd2, *abstd2_fpr): Merge.
28803 * config/rs6000/e500.h: Delete file.
28804 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28805 TARGET_FPRS by 1 and simplify.
28806 * config/rs6000/rs6000-c.c: Ditto.
28807 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28808 TARGET_DF_SPE by 0.
28809 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28811 * config/rs6000/rs6000.md: Ditto.
28812 (floatdidf2, *floatdidf2_fpr): Merge.
28813 (move_from_CR_gt_bit): Delete.
28814 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28815 (E500_CR_IOR_COMPARE): Delete.
28816 (All patterns that require !TARGET_FPRS): Delete.
28817 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28819 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28821 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28823 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28825 * graphds.c (add_edge): Intitialize edge's attached data.
28826 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28827 pointer parameter. Call pointed function on each edge during
28828 graph traversing. Skip traversing the edge when the function
28830 (graphds_dfs, graphds_scc): Ditto.
28831 (for_each_edge): New parameter. Pass the new parameter to callback
28833 * graphds.h (skip_edge_callback): New function pointer type.
28834 (graphds_dfs, graphds_scc): New function pointer parameter.
28835 (graphds_edge_callback, for_each_edge): New parameter.
28837 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28839 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28840 out code checking if runtime alias check is possible to below ...
28841 Call the new function.
28842 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28843 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28845 2017-06-07 Marek Polacek <polacek@redhat.com>
28848 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28849 TYPE_OVERFLOW_WRAPS check.
28851 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28853 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28854 if versioning is required.
28855 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28856 peeling with the check for versioning.
28858 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28860 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28861 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28862 Set true to new parameter if new ssa variable is defined.
28863 (vect_gen_vector_loop_niters): Refactor. Set range information
28864 for the new vector loop bound variable.
28865 (vect_do_peeling): Ditto.
28867 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28869 * tree-affine.c (ssa.h): Include header file.
28870 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28871 has wrapping overflow behavior.
28873 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28875 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28877 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28879 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28880 (tree_to_aff_combination): ... here.
28882 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28884 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28885 reg_pressure model function.
28886 (ivopts_global_cost_for_size): Delete.
28887 (determine_set_costs, iv_ca_recount_cost): Call new model function
28888 ivopts_estimate_reg_pressure.
28890 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28892 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28893 expensive than udiv. Remove floating point cases from mod.
28895 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28897 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28898 Increase idiv cost.
28900 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28902 * config/aarch64/aarch64.md
28903 (copysignsf3): Fix mask generation.
28905 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28907 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28909 (class dump_manager): Add register_dumps method.
28910 * dumpfile.c: Include langhooks.h.
28911 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28912 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28913 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28914 (dump_manager::dump_register): Start with 512 entries instead of 32.
28915 (dump_manager::register_dumps): New method.
28916 * toplev.c (general_init): Instead of invoking register_dumps
28917 langhook, invoke register_dumps method on the dump manager.
28918 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28921 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28923 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28924 Say that # requires an associated define_split to exist, and that
28925 the define_split must be suitable for use after register allocation.
28927 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28929 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28930 (compute_outgoing_frequencies): Also initialize zero counts.
28931 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28932 around loops; preserve more of profile when nothing changes.
28934 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28936 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28938 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28939 * config/arm/arm-cpu-cdata.h: Regenerate.
28940 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28941 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28942 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28943 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28944 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28946 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28948 * config/arm/t-rmprofile: Likewise.
28949 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28951 2017-06-06 David S. Miller <davem@davemloft.net>
28954 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28955 blockage if function uses alloca.
28957 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28959 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28960 New "uid" fields to hold pretty-print IDs of group and ref.
28961 Memory references are now identified as <group_id>:<ref_id>
28962 instead of using [random] addresses.
28963 (dump_mem_details): Simplify, no functional change.
28964 (dump_mem_ref): Simplify and make output more concise.
28965 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28966 (find_or_create_group): Initialize group uid.
28967 (record_ref): Initialize ref uid. Improve debug output.
28968 (prune_group_by_reuse, should_issue_prefetch_p,)
28969 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28970 (mark_nontemporal_store, determine_loop_nest_reuse):
28971 Improve debug output.
28973 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28975 * dbgcnt.def (prefetch): New debug counter.
28976 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28977 (schedule_prefetches): Stop issueing prefetches if debug counter
28980 2017-06-06 Tom de Vries <tom@codesourcery.com>
28982 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28983 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28985 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28987 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28988 Use aarch64_reg_or_zero predicate for operand 4.
28989 (aarch64_compare_and_swap<mode> define_insn_and_split):
28990 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28991 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28993 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28995 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28996 (arm_compute_save_core_reg_mask): This.
28997 (thumb1_compute_save_reg_mask): Rename into ...
28998 (thumb1_compute_save_core_reg_mask): This.
28999 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29000 (arm_compute_frame_layout): Likewise.
29002 2017-06-06 Richard Biener <rguenther@suse.de>
29004 PR tree-optimization/80974
29005 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29006 keep or clear leaders SSA info.
29008 2017-06-06 Tom de Vries <tom@codesourcery.com>
29010 * config/nvptx/nvptx.c (split_mode_p): New function.
29011 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29013 2017-06-06 Tom de Vries <tom@codesourcery.com>
29015 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29017 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29020 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29023 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29025 * shrink-wrap.c (handle_simple_exit): Update profile.
29026 (try_shrink_wrapping): Upate profile.
29028 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29030 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29031 (tree_guess_outgoing_edge_probabilities): New.
29032 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29033 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29035 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29037 * ipa-split.c (split_function): Initialize return bb profile.
29039 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29041 * profile.c (compute_branch_probabilities): Also initialize
29042 EXIT_BLOCK profile.
29044 2017-06-06 Richard Biener <rguenther@suse.de>
29046 PR tree-optimization/80928
29047 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29048 (vect_analyze_loop_operations): Properly guard analysis for
29050 (vect_transform_loop): Likewise.
29051 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29052 (vect_model_induction_cost): Do not cost for pure SLP.
29053 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29054 of induction in inner loop vectorization.
29055 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29056 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29057 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29059 (vect_analyze_slp_cost_1): Cost induction.
29060 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29061 (vect_get_slp_vect_defs): Likewise.
29062 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29063 (vect_transform_stmt): Handle SLP reductions.
29064 * tree-vectorizer.h (vectorizable_induction): Adjust.
29066 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29068 * config/rs6000/rs6000.c (make_resolver_func): Update
29069 init_lowered_empty_function call.
29071 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29073 * doc/invoke.texi: Document the -fprofile-abs-path option.
29074 * common.opt (fprofile-abs-path): New option.
29075 * gcov-io.h (gcov_write_filename): Declare.
29076 * gcov-io.c (gcov_write_filename): New function.
29077 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29078 * profile.c (output_location): Likewise.
29080 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29082 * shring-wrap.c: Revert accidental commit.
29084 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29086 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29088 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29090 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29092 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29093 profile in callgraph edge.
29094 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29095 (apply_scale): Likewise.
29096 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29097 Also scale profile when inlining function with zero profile.
29098 (initialize_cfun): Update exit block profile even when it is zero.
29099 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29100 when profile is read.
29102 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29104 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29105 (CLONE_*): New constants to define the processors we can generate
29106 code for with the target_clone attribute.
29107 (rs6000_clone_map): New array to identify which clone processors
29108 the current program is running on.
29109 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29110 target_clone attribute.
29111 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29112 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29113 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29114 (cpu_expand_builtin): Add support for target_clone attribute.
29115 (rs6000_valid_attribute_p): Allow "default" attribute.
29116 (get_decl_name): New debug function to simplify printing the
29117 current function name in debugging statements.
29118 (rs6000_clone_priority): New functions to support the target_clone
29119 attribute, and be able to generate code to switch between ISA 2.05
29120 through ISA 3.0 (power6 through power9).
29121 (rs6000_compare_version_priority): Likewise.
29122 (rs6000_get_function_versions_dispatcher): Likewise.
29123 (make_resolver_func): Likewise.
29124 (add_condition_to_bb): Likewise.
29125 (dispatch_function_versions): Likewise.
29126 (rs6000_generate_version_dispatcher_body): Likewise.
29127 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29128 (fusion_gpr_load_p): Fix a spacing issue.
29129 * doc/extend.texi (Common Function Attributes): Document that the
29130 PowerPC supports the target_clone attribute.
29132 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29134 * config/arm/arm.h: explain F symbol found in description of ARM
29135 register allocation in its legend.
29137 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29139 * config/mips/frame-header-opt.c: Include profile-count.h.
29140 * config/riscv/riscv.c: Include profile-count.h
29142 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29144 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29146 (sm_set_flag_if_changed): Add bbs field.
29147 (execute_sm_if_changed_flag_set): Pass BBS.
29148 (execute_sm): Update.
29150 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29152 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29155 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29157 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29158 (peephole2): New peephole2 to emit the above.
29159 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29161 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29163 * config/aarch64/aarch64.c (define_peephole2 above
29164 *sub_<shift>_<mode>): New peephole.
29166 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29168 * config/i386/i386.c (make_resolver_func): Update.
29169 * Makefile.in: Add profile-count.h and profile-count.o
29170 * auto-profile.c (afdo_indirect_call): Update to new API.
29171 (afdo_set_bb_count): Update.
29172 (afdo_propagate_edge): Update.
29173 (afdo_propagate_circuit): Update.
29174 (afdo_calculate_branch_prob): Update.
29175 (afdo_annotate_cfg): Update.
29176 * basic-block.h: Include profile-count.h
29177 (struct edge_def): Turn count to profile_count.
29178 (struct basic_block_def): Likewie.
29179 (REG_BR_PROB_BASE): Move to profile-count.h
29180 (RDIV): Move to profile-count.h
29181 * bb-reorder.c (max_entry_count): Turn to profile_count.
29182 (find_traces): Update.
29183 (rotate_loop):Update.
29184 (connect_traces):Update.
29185 (sanitize_hot_paths):Update.
29186 * bt-load.c (migrate_btr_defs): Update.
29187 * cfg.c (RDIV): Remove.
29188 (init_flow): Use alloc_block.
29189 (alloc_block): Uninitialize count.
29190 (unchecked_make_edge): Uninitialize count.
29191 (check_bb_profile): Update.
29192 (dump_edge_info): Update.
29193 (dump_bb_info): Update.
29194 (update_bb_profile_for_threading): Update.
29195 (scale_bbs_frequencies_int): Update.
29196 (scale_bbs_frequencies_gcov_type): Update.
29197 (scale_bbs_frequencies_profile_count): New.
29198 * cfg.h (update_bb_profile_for_threading): Update.
29199 (scale_bbs_frequencies_profile_count): Declare.
29200 * cfgbuild.c (compute_outgoing_frequencies): Update.
29201 (find_many_sub_basic_blocks): Update.
29202 * cfgcleanup.c (try_forward_edges): Update.
29203 (try_crossjump_to_edge): Update.
29204 * cfgexpand.c (expand_gimple_tailcall): Update.
29205 (construct_exit_block): Update.
29206 * cfghooks.c (verify_flow_info): Update.
29207 (dump_bb_for_graph): Update.
29208 (split_edge): Update.
29209 (make_forwarder_block): Update.
29210 (duplicate_block): Update.
29211 (account_profile_record): Update.
29212 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29213 (get_estimated_loop_iterations): Update.
29214 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29215 (single_likely_exit): Update.
29216 * cfgloopmanip.c (scale_loop_profile): Update.
29218 (set_zero_probability): Update.
29219 (lv_adjust_loop_entry_edge): Update.
29220 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29221 (purge_dead_edges): Update.
29222 (rtl_account_profile_record): Update.
29223 * cgraph.c (cgraph_node::create): Uninitialize count.
29224 (symbol_table::create_edge): Uninitialize count.
29225 (cgraph_update_edges_for_call_stmt_node): Update.
29226 (cgraph_edge::dump_edge_flags): Update.
29227 (cgraph_node::dump): Update.
29228 (cgraph_edge::maybe_hot_p): Update.
29229 * cgraph.h: Include profile-count.h
29230 (create_clone), create_edge, create_indirect_edge): Update.
29231 (cgraph_node): Turn count to profile_count.
29232 (cgraph_edge0: Likewise.
29233 (make_speculative, clone): Update.
29234 (create_edge): Update.
29235 (init_lowered_empty_function): Update.
29236 * cgraphclones.c (cgraph_edge::clone): Update.
29237 (duplicate_thunk_for_node): Update.
29238 (cgraph_node::create_clone): Update.
29239 * cgraphunit.c (cgraph_node::analyze): Update.
29240 (cgraph_node::expand_thunk): Update.
29241 * final.c (dump_basic_block_info): Update.
29242 * gimple-streamer-in.c (input_bb): Update.
29243 * gimple-streamer-out.c (output_bb): Update.
29244 * graphite.c (print_global_statistics): Update.
29245 (print_graphite_scop_statistics): Update.
29246 * hsa-brig.c: Include basic-block.h.
29247 * hsa-dump.c: Include basic-block.h.
29248 * hsa-gen.c (T sum_slice): Update.
29249 (convert_switch_statements):Update.
29250 * hsa-regalloc.c: Include basic-block.h.
29251 * ipa-chkp.c (chkp_produce_thunks): Update.
29252 * ipa-cp.c (struct caller_statistics): Update.
29253 (init_caller_stats): Update.
29254 (gather_caller_stats): Update.
29255 (ipcp_cloning_candidate_p): Update.
29256 (good_cloning_opportunity_p): Update.
29257 (get_info_about_necessary_edges): Update.
29258 (dump_profile_updates): Update.
29259 (update_profiling_info): Update.
29260 (update_specialized_profile): Update.
29261 (perhaps_add_new_callers): Update.
29262 (decide_about_value): Update.
29263 (ipa_cp_c_finalize): Update.
29264 * ipa-devirt.c (struct odr_type_warn_count): Update.
29265 (struct decl_warn_count): Update.
29266 (struct final_warning_record): Update.
29267 (possible_polymorphic_call_targets): Update.
29268 (ipa_devirt): Update.
29269 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29270 * ipa-icf.c (sem_function::merge): Update.
29271 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29272 * ipa-inline.c (compute_uninlined_call_time): Update.
29273 (compute_inlined_call_time): Update.
29274 (want_inline_small_function_p): Update.
29275 (want_inline_self_recursive_call_p): Update.
29276 (edge_badness): Update.
29277 (lookup_recursive_calls): Update.
29278 (recursive_inlining): Update.
29279 (inline_small_functions): Update.
29280 (dump_overall_stats): Update.
29281 (dump_inline_stats): Update.
29282 * ipa-profile.c (ipa_profile_generate_summary): Update.
29283 (ipa_propagate_frequency): Update.
29284 (ipa_profile): Update.
29285 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29286 * ipa-utils.c (ipa_merge_profiles): Update.
29287 * loop-doloop.c (doloop_modify): Update.
29288 * loop-unroll.c (report_unroll): Update.
29289 (unroll_loop_runtime_iterations): Update.
29290 * lto-cgraph.c (lto_output_edge): Update.
29291 (lto_output_node): Update.
29292 (input_node): Update.
29293 (input_edge): Update.
29294 (merge_profile_summaries): Update.
29295 * lto-streamer-in.c (input_cfg): Update.
29296 * lto-streamer-out.c (output_cfg): Update.
29297 * mcf.c (create_fixup_graph): Update.
29298 (adjust_cfg_counts): Update.
29299 (sum_edge_counts): Update.
29300 * modulo-sched.c (sms_schedule): Update.
29301 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29302 * predict.c (maybe_hot_count_p): Update.
29303 (probably_never_executed): Update.
29304 (dump_prediction): Update.
29305 (combine_predictions_for_bb): Update.
29306 (propagate_freq): Update.
29307 (handle_missing_profiles): Update.
29308 (counts_to_freqs): Update.
29309 (rebuild_frequencies): Update.
29310 (force_edge_cold): Update.
29311 * predict.h: Include profile-count.h
29312 (maybe_hot_count_p, counts_to_freqs): UPdate.
29313 * print-rtl-function.c: Do not include cfg.h
29314 * print-rtl.c: Include basic-block.h
29315 * profile-count.c: New file.
29316 * profile-count.h: New file.
29317 * profile.c (is_edge_inconsistent): Update.
29318 (correct_negative_edge_counts): Update.
29319 (is_inconsistent): Update.
29320 (set_bb_counts): Update.
29321 (read_profile_edge_counts): Update.
29322 (compute_frequency_overlap): Update.
29323 (compute_branch_probabilities): Update; Initialize and deinitialize
29325 (branch_prob): Update.
29326 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29327 (edge_gcov_count): New.
29328 (bb_gcov_count): New.
29329 * shrink-wrap.c (try_shrink_wrapping): Update.
29330 * tracer.c (better_p): Update.
29331 * trans-mem.c (expand_transaction): Update.
29332 (ipa_tm_insert_irr_call): Update.
29333 (ipa_tm_insert_gettmclone_call): Update.
29334 * tree-call-cdce.c: Update.
29335 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29336 (gimple_duplicate_sese_tail): Update.
29337 (gimple_account_profile_record): Update.
29338 (execute_fixup_cfg): Update.
29339 * tree-inline.c (copy_bb): Update.
29340 (copy_edges_for_bb): Update.
29341 (initialize_cfun): Update.
29342 (freqs_to_counts): Update.
29343 (copy_cfg_body): Update.
29344 (expand_call_inline): Update.
29345 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29346 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29347 (try_unroll_loop_completely): Update.
29348 (try_peel_loop): Update.
29349 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29350 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29351 * tree-ssa-loop-split.c (connect_loops): Update.
29352 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29353 * tree-ssa-reassoc.c (branch_fixup): Update.
29354 * tree-ssa-tail-merge.c (replace_block_by): Update.
29355 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29356 (compute_path_counts): Update.
29357 (update_profile): Update.
29358 (recompute_probabilities): Update.
29359 (update_joiner_offpath_counts): Update.
29360 (estimated_freqs_path): Update.
29361 (freqs_to_counts_path): Update.
29362 (clear_counts_path): Update.
29363 (ssa_fix_duplicate_block_edges): Update.
29364 (duplicate_thread_path): Update.
29365 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29366 (struct switch_conv_info): Update.
29367 * tree-tailcall.c (decrease_profile): Update.
29368 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29369 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29370 * value-prof.c (check_counter): Update.
29371 (gimple_divmod_fixed_value): Update.
29372 (gimple_mod_pow2): Update.
29373 (gimple_mod_subtract): Update.
29374 (gimple_ic_transform): Update.
29375 (gimple_stringop_fixed_value): Update.
29376 * value-prof.h (gimple_ic): Update.
29378 2017-06-02 Carl Love <cel@us.ibm.com>
29380 * config/rs6000/rs6000-c: Add support for built-in functions
29381 vector double vec_doublee (vector signed int);
29382 vector double vec_doublee (vector unsigned int);
29383 vector double vec_doublee (vector float);
29384 vector double vec_doubleh (vector signed int);
29385 vector double vec_doubleh (vector unsigned int);
29386 vector double vec_doubleh (vector float);
29387 vector double vec_doublel (vector signed int);
29388 vector double vec_doublel (vector unsigned int);
29389 vector double vec_doublel (vector float);
29390 vector double vec_doubleo (vector signed int);
29391 vector double vec_doubleo (vector unsigned int);
29392 vector double vec_doubleo (vector float);.
29393 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29394 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29396 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29397 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29398 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29400 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29401 vec_doublel, vec_doubleh.
29402 * doc/extend.texi: Update the built-in documentation file for the
29403 new built-in functions.
29405 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29408 * ipa-inline-analysis.c (free_growth_caches): Set
29409 edge_removal_hook_holder to NULL after removing it.
29411 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29413 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29414 comparision with zero.
29416 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29417 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29418 for early expansion of vec_min and vec_max builtins.
29419 (builtin_function_type): Add min/max unsigned variants to those
29420 identified as having unsigned arguments.
29422 2017-06-02 Olivier Hainque <hainque@adacore.com>
29424 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29426 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29428 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29429 Use VALL_F16 iterator rather than VALL.
29431 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29433 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29434 Emit CBNZ inside loop when doing a strong exchange and comparing
29435 against zero. Generate the CC flags after the loop.
29437 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29439 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29440 (dl_section_ref): New.
29441 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29442 On AIX, append an expression to subtract the size of the
29443 section length to dl_section_ref.
29445 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29447 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29448 for early expansion of vector absolute builtins.
29450 2017-06-02 Richard Biener <rguenther@suse.de>
29452 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29453 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29455 2017-06-02 Richard Biener <rguenther@suse.de>
29457 PR tree-optimization/80948
29458 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29459 stmt order as well.
29461 2017-06-02 Richard Biener <rguenther@suse.de>
29463 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29465 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29466 for inductions as relevant.
29468 2017-06-02 Richard Biener <rguenther@suse.de>
29470 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29471 (vectorizable_induction): ... this. Remove dead code.
29473 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29475 * builtins. (expand_builtin_alloca): Remove second parameter and
29476 infer its value from the first parameter instead.
29477 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29479 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29481 PR rtl-optimization/80903
29482 * loop-doloop.c (add_test): Unshare sequence.
29484 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29486 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29488 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29490 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29492 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29493 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29494 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29495 (xlogue_layout::get_stub_rtx): Make static.
29496 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29497 (xlogue_layout::compute_stub_managed_regs): Rename to...
29498 (xlogue_layout::count_stub_managed_regs): ...this.
29499 (xlogue_layout::is_stub_managed_reg): New function.
29500 (xlogue_layout::m_stub_names): Rename to...
29501 (xlogue_layout::s_stub_names): ...this, make static.
29502 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29503 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29504 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29505 xlogue_layout::s_stub_names): Instantiate statics.
29506 (stub_managed_regs): Remove.
29507 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29508 (disable_call_ms2sysv_xlogues): Rename to...
29509 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29510 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29512 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29513 change after reload_completed.
29514 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29516 (ix86_expand_prologue): Likewise.
29517 (ix86_expand_epilogue): Likewise.
29518 (ix86_expand_split_stack_prologue): Likewise.
29519 (ix86_compute_frame_layout): Remove frame parameter ...
29520 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29521 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29523 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29524 (ix86_frame): Move from here ...
29525 * config/i386/i386.h (ix86_frame): ... to here.
29526 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29527 complete ix86_frame data structure instead. Remove some_ld_name.
29529 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29531 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29532 symbols that hold a DECL_VALUE_EXPR.
29534 2017-06-01 Martin Jambor <mjambor@suse.cz>
29536 PR tree-optimization/80898
29537 * tree-sra.c (process_subtree_disqualification): Removed.
29538 (disqualify_candidate): Do not acll
29539 process_subtree_disqualification.
29540 (subtree_mark_written_and_enqueue): New function.
29541 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29542 RHS has been disqualified and re-queue LHS if necessary. Apart
29543 from that, ignore disqualified RHS.
29545 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29547 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29548 address fetch for z10 or later.
29550 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29552 * config/arc/arc.md (tst_movb): Add guard when splitting.
29554 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29556 * config/arc/arc.c (arc_can_eliminate): Test against
29557 arc_frame_pointer_needed.
29559 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29561 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29562 to prevent store reordering.
29563 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29564 (type): Add block type.
29565 (stack_tie): Define special instruction to be used in
29568 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29570 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29571 constraint. It is not valid for the pattern.
29572 (noncommutative_binary_comparison): Likewise.
29574 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29576 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29579 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29581 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29582 be used by the reg-alloc.
29584 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29586 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29587 reg-alloc when having mul64 or mul32x16 instructions.
29588 (mulsidi3): Likewise.
29589 (umulsidi3): Likewise.
29590 (mulsi32x16): New pattern.
29591 (mulsi64): Likewise.
29592 (mulsidi64): Likewise.
29593 (umulsidi64): Likewise.
29594 (MUL32x16_REG): Define.
29595 (mul64_600): Use MUL32x16_REG.
29596 (mac64_600): Likewise.
29597 (umul64_600): Likewise.
29598 (umac64_600): Likewise.
29600 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29602 * config/arc/arc.md (mulsi3_700): Make it commutative.
29604 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29606 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29608 (*sign_extendsidi2_insn): Likewise for movstosw.
29610 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29612 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29613 the type of the input discriminant value. Convert the
29614 discriminant value of signedness vary.
29616 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29618 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29619 Add to -Wall section.
29621 2017-06-01 Richard Biener <rguenther@suse.de>
29623 PR middle-end/66313
29624 * fold-const.c (fold_plusminus_mult_expr): If the factored
29625 factor may be zero use a wrapping type for the inner operation.
29626 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29627 and handle moved defs.
29628 (process_assignment): Properly guard the unary op case. Return a
29629 tri-state indicating that moving the stmt before the call may allow
29630 to continue. Pass through to_move.
29631 (find_tail_calls): Handle moving unrelated defs before
29634 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29637 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29638 splitter result in the canonical way.
29640 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29642 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29643 also for 32bit target. Update insn attributes.
29644 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29646 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29648 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29649 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29650 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29651 (_mm_maskz_min_ss): New intrinsics.
29653 2017-05-31 Martin Liska <mliska@suse.cz>
29655 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29656 Change comment style to one we normally use.
29657 (vectorizable_reduction): Likewise.
29658 (vectorizable_induction): Likewise.
29659 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29660 (vectorizable_call): Likewise.
29661 (vectorizable_simd_clone_call): Likewise.
29662 (vectorizable_conversion): Likewise.
29663 (vectorizable_assignment): Likewise.
29664 (vectorizable_shift): Likewise.
29665 (vectorizable_operation): Likewise.
29666 (vectorizable_store): Likewise.
29667 (vectorizable_load): Likewise.
29668 * tree-vectorizer.h: Likewise.
29670 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29672 * passes.c (emergency_dump_function): New.
29673 * tree-pass.h (emergency_dump_function): Declare.
29674 * plugin.c (plugins_internal_error_function): Remove.
29675 * plugin.h (plugins_internal_error_function): Remove declaration.
29676 * toplev.c (internal_error_function): New static function. Use it...
29677 (general_init): ...here.
29679 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29681 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29682 (arc_rtx_costs): Add costs for new patterns.
29683 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29684 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29686 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29688 * tree-ssa-strlen.c (get_next_strinfo): New function.
29689 (get_stridx_plus_constant): Use it.
29690 (zero_length_string): Likewise.
29691 (adjust_related_strinfos): Likewise.
29692 (adjust_last_stmt): Likewise.
29694 2017-05-31 Richard Biener <rguenther@suse.de>
29697 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29698 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29700 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29702 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29703 loop_vinfo argument and use of dependence distance vectors.
29704 Check instead whether the two references differ only in their
29705 initial value and assume that they have the same alignment if the
29706 difference is a multiple of the vector alignment.
29707 (vect_analyze_data_refs_alignment): Update call accordingly.
29709 2017-05-31 Martin Liska <mliska@suse.cz>
29712 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
29714 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29716 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
29717 (create_intersect_range_checks): Move from ...
29718 * tree-data-ref.c (create_intersect_range_checks_index)
29719 (create_intersect_range_checks): ... to here.
29720 (create_runtime_alias_checks): New function factored from ...
29721 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
29722 here. Call above function.
29723 * tree-data-ref.h (create_runtime_alias_checks): New function.
29725 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29727 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
29728 segment length for dr_b and compute it in wide_int.
29730 2017-05-31 Richard Biener <rguenther@suse.de>
29732 PR tree-optimization/80906
29733 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
29734 and pass through iv_map.
29735 (copy_bb_and_scalar_dependences): Adjust.
29736 (translate_pending_phi_nodes): Likewise.
29737 (copy_loop_close_phi_args): Handle code-generating IVs instead
29740 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29742 * diagnostic-color.c (color_dict): Add "type-diff".
29743 (parse_gcc_colors): Update comment.
29744 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29745 -fdiagnostics-show-template-tree and -fno-elide-type.
29746 (GCC_COLORS): Add type-diff to example.
29748 (-fdiagnostics-show-template-tree): New.
29749 (-fno-elide-type): New.
29750 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29751 the pp_format_decoder callback. Call any m_format_postprocessor's
29753 (pretty_printer::pretty_printer): Initialize
29754 m_format_postprocessor.
29755 (pretty_printer::~pretty_printer): Delete any
29756 m_format_postprocessor.
29757 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29758 (class format_postprocessor): New class.
29759 (struct pretty_printer::format_decoder): Document the new parameters.
29760 (struct pretty_printer::m_format_postprocessor): New field.
29761 * tree-diagnostic.c (default_tree_printer): Update for new
29762 bool and const char ** params.
29763 * tree-diagnostic.h (default_tree_printer): Likewise.
29765 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
29767 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
29768 (lwa_operand): Delete rs6000_gen_cell_microcode test.
29769 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29770 rs6000_gen_cell_microcode code.
29771 (rs6000_final_prescan_insn): Delete.
29772 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
29773 "warn-cell-microcode" entries.
29774 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29775 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29776 throughout. Change cc_reg_not_micro_cr0_operand to
29777 cc_reg_not_cr0_operand throughout.
29778 (*extendhi<mode>2_noload): Delete.
29779 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
29780 (mwarn-cell-microcode): Delete.
29781 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
29782 -mgen-cell-microcode and -mwarn-cell-microcode.
29784 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
29787 * config/i386/constraints.md (Yd): New constraint.
29789 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
29790 and (?Yd, r) alternatives. Update insn attributes.
29791 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
29792 and (?*Yd, r) alternatives. Update insn attributes.
29793 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
29795 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
29797 * gimplify.c (gimplify_modify_expr): Don't create a
29798 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29801 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29803 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29805 2017-05-30 Richard Biener <rguenther@suse.de>
29807 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29808 and reduc_def fields.
29809 (STMT_VINFO_REDUC_TYPE): New define.
29810 (STMT_VINFO_REDUC_DEF): Likewise.
29811 (vect_force_simple_reduction): Adjust prototype.
29812 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29813 (vect_is_simple_reduction): Remove check_reduction argument.
29814 (vect_force_simple_reduction): Adjust and set
29815 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29816 (vectorizable_reduction): Do not re-do reduction analysis
29817 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29818 * tree-parloops.c (gather_scalar_reductions): Adjust.
29820 2017-05-30 Richard Biener <rguenther@suse.de>
29822 PR middle-end/80901
29823 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29826 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29828 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29829 Introduce unknown_misalignment parameter and remove vf.
29830 (vect_peeling_hash_get_lowest_cost):
29831 Pass unknown_misalignment parameter.
29832 (vect_enhance_data_refs_alignment):
29833 Fix unsupportable data ref treatment.
29835 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29837 * tree-vect-data-refs.c (vect_get_data_access_cost):
29838 Workaround for SLP handling.
29839 (vect_enhance_data_refs_alignment):
29840 Compute costs for doing no peeling at all, compare to the best
29841 peeling costs so far and avoid peeling if cheaper.
29843 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29845 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29846 Return peeling info and set costs to zero for unlimited cost
29848 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29849 with unknown misalignment. Compute and costs for unknown
29850 misalignment, compare them to the costs for known misalignment
29851 and choose the cheapest for peeling.
29853 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29855 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29856 (vect_get_peeling_costs_all_drs): Create function.
29857 (vect_peeling_hash_get_lowest_cost):
29858 Use vect_get_peeling_costs_all_drs.
29859 (vect_peeling_supportable): Create function.
29860 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29862 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29864 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29865 DR_HAS_NEGATIVE_STEP.
29866 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29867 (vect_enhance_data_refs_alignment): Use.
29868 (vect_duplicate_ssa_name_ptr_info): Use.
29869 * tree-vectorizer.h (dr_misalignment): Use.
29870 (known_alignment_for_access_p): Use.
29872 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29875 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29877 (has_section_name): New function.
29879 2017-05-30 Martin Liska <mliska@suse.cz>
29882 * auto-profile.c (get_function_decl_from_block): Fix
29885 2017-05-30 Richard Biener <rguenther@suse.de>
29887 PR middle-end/80876
29888 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29890 2017-05-30 Martin Liska <mliska@suse.cz>
29892 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29893 * dumpfile.h (struct dump_file_info): Remove ctors.
29895 2017-05-30 Martin Liska <mliska@suse.cz>
29897 * predict.def: Fix GNU coding style.
29899 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29901 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29902 Mark 'to' argument with ATTRIBUTE_UNUSED.
29904 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29906 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29907 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29908 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29911 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29913 * doc/install.texi (Options specification): Restore entry of
29914 --enable-sjlj-exceptions.
29916 2017-05-27 Michael Eager <eager@eagercon.com>
29919 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29921 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29923 * config/microblaze/microblaze.h
29924 (FIXED_REGISTERS): Update in macro.
29925 (CALL_USED_REGISTERS): Update in macro.
29927 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29929 * doc/install.texi: Add links to macOS binary distributions.
29931 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29935 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29937 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29939 2017-05-26 Martin Liska <mliska@suse.cz>
29941 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29943 2017-05-26 Martin Liska <mliska@suse.cz>
29945 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29946 always leading ';; '.
29947 (dump_bb_info): Likewise.
29948 (brief_dump_cfg): Likewise.
29949 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29950 * dumpfile.c: Remove usage of TDF_VERBOSE.
29951 * dumpfile.h (enum dump_kind): Likewise.
29952 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29953 * print-tree.c (debug_verbose): Remove.
29954 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29955 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29956 * tree-diagnostic.c (default_tree_printer): Replace
29957 TDF_DIAGNOSTIC with TDF_SLIM.
29959 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29961 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29962 in parameter loop, rather than loop_vinfo.
29963 (create_intersect_range_checks): Ditto.
29964 (vect_create_cond_for_alias_checks): Update call to above functions.
29966 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29968 PR tree-optimization/80815
29969 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29970 for merging runtime alias checks. Handle negative DR_STEPs.
29972 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29974 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29976 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29977 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29978 out code pruning runtime alias checks.
29979 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29980 factored out from above.
29981 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29983 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29985 (prune_runtime_alias_test_list): New decalaration.
29987 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29989 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29990 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29991 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29992 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29993 (operator==, comp_dr_with_seg_len_pair): Ditto.
29994 (vect_prune_runtime_alias_test_list): Ditto.
29996 2017-05-26 Martin Liska <mliska@suse.cz>
29999 * params.def: Bound partial-inlining-entry-probability param.
30001 2017-05-26 Marek Polacek <polacek@redhat.com>
30004 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30007 2017-05-26 Richard Biener <rguenther@suse.de>
30009 PR tree-optimization/80842
30010 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30013 2017-05-26 Richard Biener <rguenther@suse.de>
30015 PR tree-optimization/80844
30016 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30018 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30020 * doc/md.texi (Machine Constraints): Update x86 family
30021 machine constraints section to match 'config/i386/constraints.md'.
30023 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30025 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30027 2017-05-25 Nathan Sidwell <nathan@acm.org>
30029 * doc/invoke.texi (--enable-languages): Update documentation.
30031 2017-05-25 Martin Liska <mliska@suse.cz>
30033 * dumpfile.c: Add TDF_FOLDING.
30034 * dumpfile.h (enum dump_kind): Likewise.
30035 * genmatch.c (dt_simplify::gen_1): Use it.
30037 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30039 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30041 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30043 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30044 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30046 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30048 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30049 * match.pd (X == C): Rewrite it here.
30050 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30051 with_certain_nonzero_bits2): New predicates.
30052 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30054 2017-05-24 Nathan Sidwell <nathan@acm.org>
30056 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30059 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30062 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30064 * config/powerpcspe: New port. Files are copied from the rs6000
30065 port, with "rs6000" in filenames replaced by "powerpcspe".
30067 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30069 PR rtl-optimization/80754
30070 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30072 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30074 * config/sparc/sparc.md (length): Return the correct value for -mflat
30075 sibcalls to match output_sibcall.
30077 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30081 * config/rs6000/rs6000.c (struct machine_function): Add new field
30083 (rs6000_get_separate_components): Init that field, use it.
30084 (rs6000_components_for_bb): Use the field.
30086 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30088 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30090 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30092 PR middle-end/80823
30093 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30095 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30098 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30099 address against address_operand predicate.
30100 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30102 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30104 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30105 parameters passed indirectly.
30107 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30109 * config/i386/i386.md (*movdi_internal): Remove SSE4
30110 alternative 18 (?r, *v). Update insn attributes.
30111 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30112 Update insn attributes.
30113 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30114 Update insn attributes.
30115 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30116 alternative 1 (r, v). Remove isa attribute.
30117 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30118 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30119 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30121 2017-05-23 Tom de Vries <tom@codesourcery.com>
30123 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30126 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30128 * cgraphunit.c (symbol_table::process_new_functions): Update.
30129 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30130 (inline_generate_summary): Rename to ...
30131 (ipa_fn_summary_generate): ... this one.
30132 (inline_read_summary): Rename to ...
30133 (ipa_fn_summary_read): ... this one.
30134 (inline_write_summary): Rename to ...
30135 (ipa_fn_summary_write): ... this one.
30136 (inline_free_summary): Rename to ...
30137 (ipa_free_fn_summary): ... this one.
30138 (pass_data_local_fn_summary, pass_local_fn_summary,
30139 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30140 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30141 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30142 make_pass_ipa_fn_summary): New.
30143 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30144 inline_write_summary, inline_free_summary): Remove.
30145 (ipa_free_fn_summary) : New.
30146 * ipa-inline.c (ipa_inline): Update.
30147 (pass_ipa_inline): Do not generate summaries.
30148 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30150 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30151 and add pass_ipa_fn_summary.
30152 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30154 (make_pass_inline_parameters): Remove.
30156 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30158 * omp-low.c (struct omp_context): Remove "default_kind" member.
30161 * omp-offload.c (execute_oacc_device_lower): Remove the
30162 parallelism dimensions function attributes for unparallelized
30163 OpenACC kernels constructs.
30165 2017-05-23 Martin Liska <mliska@suse.cz>
30167 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30169 (cgraph_edge::make_speculative): Likewise.
30170 (cgraph_edge::resolve_speculation): Likewise.
30171 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30172 (cgraph_node::dump): Likewise.
30173 * cgraph.h: Likewise.
30174 * cgraphunit.c (analyze_functions): Likewise.
30175 (symbol_table::compile): Likewise.
30176 * ipa-cp.c (print_all_lattices): Likewise.
30177 (determine_versionability): Likewise.
30178 (initialize_node_lattices): Likewise.
30179 (ipcp_verify_propagated_values): Likewise.
30180 (estimate_local_effects): Likewise.
30181 (update_profiling_info): Likewise.
30182 (create_specialized_node): Likewise.
30183 (perhaps_add_new_callers): Likewise.
30184 (decide_about_value): Likewise.
30185 (decide_whether_version_node): Likewise.
30186 (identify_dead_nodes): Likewise.
30187 (ipcp_store_bits_results): Likewise.
30188 * ipa-devirt.c (dump_targets): Likewise.
30189 (ipa_devirt): Likewise.
30190 * ipa-icf.c (sem_item::dump): Likewise.
30191 (sem_function::equals): Likewise.
30192 (sem_variable::equals): Likewise.
30193 (sem_item_optimizer::read_section): Likewise.
30194 (sem_item_optimizer::execute): Likewise.
30195 (congruence_class::dump): Likewise.
30196 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30197 (dump_inline_summary): Likewise.
30198 (estimate_node_size_and_time): Likewise.
30199 (inline_analyze_function): Likewise.
30200 * ipa-inline-transform.c (inline_call): Likewise.
30201 * ipa-inline.c (report_inline_failed_reason): Likewise.
30202 (want_early_inline_function_p): Likewise.
30203 (edge_badness): Likewise.
30204 (update_edge_key): Likewise.
30205 (inline_small_functions): Likewise.
30206 * ipa-profile.c (ipa_profile): Likewise.
30207 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30208 (ipa_make_edge_direct_to_target): Likewise.
30209 (remove_described_reference): Likewise.
30210 (ipa_impossible_devirt_target): Likewise.
30211 (propagate_controlled_uses): Likewise.
30212 (ipa_print_node_params): Likewise.
30213 (ipcp_transform_function): Likewise.
30214 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30215 (propagate_pure_const): Likewise.
30216 * ipa-reference.c (generate_summary): Likewise.
30217 (read_write_all_from_decl): Likewise.
30218 (propagate): Likewise.
30219 (ipa_reference_read_optimization_summary): Likewise.
30220 * ipa-utils.c (ipa_merge_profiles): Likewise.
30221 * ipa.c (walk_polymorphic_call_targets): Likewise.
30222 (symbol_table::remove_unreachable_nodes): Likewise.
30223 (ipa_single_use): Likewise.
30224 * passes.c (execute_todo): Likewise.
30225 * predict.c (drop_profile): Likewise.
30226 * symtab.c (symtab_node::get_dump_name): New function.
30227 (symtab_node::dump_name): Likewise.
30228 (symtab_node::dump_asm_name): Likewise.
30229 (symtab_node::dump_references): Likewise.
30230 (symtab_node::dump_referring): Likewise.
30231 (symtab_node::dump_base): Likewise.
30232 (symtab_node::debug_symtab): Likewise.
30233 * tree-sra.c (convert_callers_for_node): Likewise.
30234 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30235 * value-prof.c (init_node_map): Likewise.
30237 2017-05-23 Martin Liska <mliska@suse.cz>
30239 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30240 and symtab_node::debug_symtab to symbol_table::debug.
30241 * cgraphunit.c (analyze_functions): Use the renamed function.
30242 (symbol_table::compile): Likewise.
30243 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30244 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30245 * passes.c (execute_todo): Likewise.
30246 * symtab.c (symbol_table::dump): New function.
30247 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30249 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30251 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30252 that nonconst implies exec.
30254 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30256 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30257 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30258 (inline_edge_summary_vec): Turn into ...
30259 (ipa_call_summaries): ... this one.
30260 (redirect_to_unreachable, edge_set_predicate,
30261 evaluate_properties_for_edge, inline_summary_alloc,
30262 reset_ipa_call_summary, reset_inline_summary,
30263 inline_summary_t::duplicate): Update.
30264 (inline_edge_duplication_hook): Turn to ...
30265 (ipa_call_summary_t::duplicate): ... this one.
30266 (inline_edge_removal_hook): Turn to ...
30267 (ipa_call_summary_t::remove): ... this one.
30268 (dump_inline_edge_summary): Turn to ...
30269 (dump_ipa_call_summary): ... this one.
30270 (estimate_function_body_sizes): Update.
30271 (inline_update_callee_summaries): Update.
30272 (remap_edge_change_prob): Update.
30273 (remap_edge_summaries): Update.
30274 (inline_merge_summary): Update.
30275 (do_estimate_edge_time): Update.
30276 (inline_generate_summary): Update.
30277 (inline_read_section): Update.
30278 (inline_read_summary): Update.
30279 (inline_free_summary): Update.
30280 * ipa-inline.c (can_inline_edge_p): Update.
30281 (compute_inlined_call_time): Update.
30282 (want_inline_small_function_p): Update.
30283 (edge_badness): Update.
30284 (early_inliner): Update.
30285 * ipa-inline.h (inline_edge_summary): Turn to ...
30286 (ipa_call_summary): ... this one.
30287 (ipa_call_summary_t): New class.
30288 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30289 (ipa_call_summaries): New.
30290 (inline_edge_summary): Remove.
30291 (estimate_edge_growth): Update.
30292 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30293 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30294 * ipa-split.c (execute_split_functions): Update.
30295 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30297 2017-05-23 Tom de Vries <tom@codesourcery.com>
30299 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30300 attributes): Document rdrand effective target.
30302 2017-05-23 Tom de Vries <tom@codesourcery.com>
30304 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30305 attributes): Sort alphabetically.
30307 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30309 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30311 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30314 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30315 V2DF/V2DI splat into two separate patterns, one that handles
30316 registers, and the other that only handles memory. Drop support
30317 for splatting from a GPR on ISA 2.07 and then splitting the
30318 splat into direct move and splat.
30319 (vsx_splat_<mode>_reg): Likewise.
30320 (vsx_splat_<mode>_mem): Likewise.
30322 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30324 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30326 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30328 PR middle-end/80809
30329 * omp-low.c (finish_taskreg_remap): New function.
30330 (finish_taskreg_scan): If unit size of ctx->record_type
30331 is non-constant, unshare the size expression and replace
30332 decls in it with possible outer var refs.
30334 PR middle-end/80809
30335 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30336 GOVD_SHARED rather than GOVD_PRIVATE with it.
30337 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30338 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30340 PR middle-end/80853
30341 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30342 as last argument to build_outer_var_ref for pointer bases of array
30343 section reductions.
30345 2017-05-19 Martin Sebor <msebor@redhat.com>
30347 * print-tree.c (print_node): Print DECL_READ_P flag.
30349 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30351 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30352 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30353 * cgraph.c: Likewise.
30354 * cgraphunit.c: Likewise.
30355 * gengtype.c: Likewise.
30356 * ipa-cp.c: Likewise.
30357 * ipa-devirt.c: Likewise.
30358 * ipa-icf.c: Likewise.
30359 * ipa-predicate.c: Likewise.
30360 * ipa-profile.c: Likewise.
30361 * ipa-prop.c: Likewise.
30362 * ipa-split.c: Likewise.
30364 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30365 edge_predicate_pool, dump_inline_hints,
30366 inline_summary::account_size_time, redirect_to_unreachable,
30367 edge_set_predicate, set_hint_predicate,
30368 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30369 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30370 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30371 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30372 ipa_call_summary_t::remove, initialize_growth_caches,
30373 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30374 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30375 mark_modified, unmodified_parm_1, unmodified_parm,
30376 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30377 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30378 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30379 will_be_nonconstant_predicate, record_modified_bb_info,
30380 get_minimal_bb, record_modified, param_change_prob,
30381 phi_result_unknown_predicate, predicate_for_phi_result,
30382 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30383 estimate_function_body_sizes, compute_inline_parameters,
30384 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30385 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30386 inline_update_callee_summaries, remap_edge_change_prob,
30387 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30388 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30389 inline_analyze_function, inline_summary_t::insert,
30390 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30391 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30392 inline_free_summary): Move to ipa-fnsummary.h
30393 (predicate_t): Remove.
30394 * ipa-fnsummary.c: New file.
30395 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30396 (enum inline_hints_vals, inline_hints, agg_position_info,
30397 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30398 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30399 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30400 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30401 inline_read_summary, inline_write_summary, inline_free_summary,
30402 inline_analyze_function, initialize_inline_failed,
30403 inline_merge_summary, inline_update_overall_summary,
30404 compute_inline_parameters): Move to ipa-fnsummary.h
30405 * ipa-fnsummary.h: New file.
30406 * ipa-inline-transform.h: Include ipa-inline.h.
30407 * ipa-inline.c: LIkewise.
30409 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30411 * ipa-inline.c (edge_badness): Use inlined_time instead of
30412 inline_summaries->get.
30414 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30416 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30418 2017-05-22 Nathan Sidwell <nathan@acm.org>
30420 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30421 (fdump-lang): Document 'raw' option.
30422 * dumpfile.h (TDI_tu): Delete.
30423 * dumpfile.c (dump_files): Remove translation-unit.
30424 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30426 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30428 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30429 command option from $(AWK) call.
30430 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30431 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30432 [FORMAT]: Remove handling of variable.
30433 * config/avr/t-multilib: Regenerate.
30435 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30437 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30439 (dump_inline_summary): Do not print self_time.
30440 (estimate_function_body_sizes): Do not set self_time.
30441 (compute_inline_parameters): Likewise.
30442 (inline_read_section, inline_write_summary): Do not stream self_time.
30443 * ipa-inline.h (inline_summary): Drop self_time.
30445 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30447 * ipa-inline-analysis.c (account_size_time): Rename to ...
30448 (inline_summary::account_size_time): ... this one.
30449 (reset_ipa_call_summary): Turn to ...
30450 (ipa_call_summary::reset): ... this one.
30451 (reset_inline_summary): Turn to ...
30452 (inline_summary::reset): ... this one.
30453 (inline_summary_t::remove): Update.
30454 (inline_summary_t::duplicate): Update.
30455 (ipa_call_summary_t::remove): Update.
30456 (dump_inline_summary): Update.
30457 (estimate_function_body_sizes): Update.
30458 (compute_inline_parameters): Update.
30459 (estimate_node_size_and_time): Update.
30460 (inline_merge_summary): Update.
30461 (inline_update_overall_summary): Update.
30462 (inline_read_section): Update.
30463 (inline_write_summary): Update.
30464 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30465 add account_size_time and reset member functions.
30466 (ipa_call_summary): Add reset function.
30467 * ipa-predicate.h (predicate::operator &): Constify.
30469 2017-05-22 Richard Biener <rguenther@suse.de>
30471 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30473 2017-05-19 Jason Merrill <jason@redhat.com>
30475 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30477 2017-05-19 Marek Polacek <polacek@redhat.com>
30480 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30481 TYPE_OVERFLOW_WRAPS checks.
30483 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30485 * tree-core.h (enum omp_clause_default_kind): Add
30486 "OMP_CLAUSE_DEFAULT_PRESENT".
30487 * tree-pretty-print.c (dump_omp_clause): Handle it.
30488 * gimplify.c (enum gimplify_omp_var_data): Add
30489 "GOVD_MAP_FORCE_PRESENT".
30490 (gimplify_adjust_omp_clauses_1): Map it to
30491 "GOMP_MAP_FORCE_PRESENT".
30492 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30494 * gimplify.c (oacc_default_clause): Clarify.
30496 2017-05-19 Nathan Sidwell <nathan@acm.org>
30498 LANG_HOOK_REGISTER_DUMPS
30499 * toplev.c (general_init): Call register dump lang hook.
30500 * doc/invoke.texi: Document -fdump-lang option family.
30501 * dumpfile.c (dump_files): Remove class dump here.
30502 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30503 * dumpfile.h (tree_dump_index): Remove TDI_class.
30504 * langhooks-def.h (lhd_register_dumps): Declare.
30505 (LANG_HOOKS_REGISTER_DUMPS): Define.
30506 (LANG_HOOKS_INITIALIZER): Add it.
30507 * langhooks.c (lhd_register_dumps): Define.
30508 * langhooks.h (struct lang_hooks): Add register_dumps.
30510 2017-05-19 Nathan Sidwell <nathan@acm.org>
30512 * context.h (context::set_passes): New.
30513 * context.c (context::context): Do not create pass manager.
30514 * toplev.c (general_init): Create pass manager here.
30516 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30518 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30519 use this splitter if two add or or instructions would also work for
30520 the constant we want to generate.
30522 2017-05-19 Richard Biener <rguenther@suse.de>
30525 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30526 predicate evaluation.
30528 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30530 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30532 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30535 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30537 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30538 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30539 (inline_edge_summary_vec): Turn into ...
30540 (ipa_call_summaries): ... this one.
30541 (redirect_to_unreachable, edge_set_predicate,
30542 evaluate_properties_for_edge, inline_summary_alloc,
30543 reset_ipa_call_summary, reset_inline_summary,
30544 inline_summary_t::duplicate): Update.
30545 (inline_edge_duplication_hook): Turn to ...
30546 (ipa_call_summary_t::duplicate): ... this one.
30547 (inline_edge_removal_hook): Turn to ...
30548 (ipa_call_summary_t::remove): ... this one.
30549 (dump_inline_edge_summary): Turn to ...
30550 (dump_ipa_call_summary): ... this one.
30551 (estimate_function_body_sizes): Update.
30552 (inline_update_callee_summaries): Update.
30553 (remap_edge_change_prob): Update.
30554 (remap_edge_summaries): Update.
30555 (inline_merge_summary): Update.
30556 (do_estimate_edge_time): Update.
30557 (inline_generate_summary): Update.
30558 (inline_read_section): Update.
30559 (inline_read_summary): Update.
30560 (inline_free_summary): Update.
30561 * ipa-inline.c (can_inline_edge_p): Update.
30562 (compute_inlined_call_time): Update.
30563 (want_inline_small_function_p): Update.
30564 (edge_badness): Update.
30565 (early_inliner): Update.
30566 * ipa-inline.h (inline_edge_summary): Turn to ...
30567 (ipa_call_summary): ... this one.
30568 (ipa_call_summary_t): New class.
30569 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30570 (ipa_call_summaries): New.
30571 (inline_edge_summary): Remove.
30572 (estimate_edge_growth): Update.
30573 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30574 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30575 * ipa-split.c (execute_split_functions): Update.
30576 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30578 2017-05-19 Richard Biener <rguenther@suse.de>
30580 PR middle-end/80764
30581 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30583 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30585 * config/rs6000/rs6000.c (struct machine_function): Add field
30586 fpr_is_wrapped_separately.
30587 (rs6000_get_separate_components): Use 64 components. Handle the
30588 new FPR components.
30589 (rs6000_components_for_bb): Handle the FPR components.
30590 (rs6000_emit_prologue_components): Handle the FPR components.
30591 (rs6000_emit_epilogue_components): Handle the FPR components.
30592 (rs6000_set_handled_components): Handle the FPR components.
30593 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30594 that are already separately shrink-wrapped.
30595 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30596 that are already separately shrink-wrapped.
30598 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30601 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30604 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30605 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30606 cases where the register allocator uses a move and an offsettable
30607 memory operation to/from a FPR register on ISA 2.06/2.07.
30608 (define_peephole2 for Altivec d-form store): Likewise.
30610 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30613 * config/i386/mmx.md (*mov<mode>_internal): Enable
30614 alternatives 11, 12, 13 and 14 also for 32bit targets.
30615 Remove alternatives 15, 16, 17 and 18.
30616 * config/i386/sse.md (vec_concatv2di): Change
30617 alternative (!x, *y) to (x, ?!*Yn).
30619 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30621 * dumpfile.h (enum dump_kind): Remove stray comma.
30623 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30625 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30626 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30627 predicate::num_conditions
30628 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30629 (CHANGED): turn into predicate::changed.
30630 (agg_position_info): Move to ipa-predicate.h
30631 (add_condition, predicate::add_clause, predicate::operator &=,
30632 predicate::or_with, predicate::evaluate, predicate::probability,
30633 dump_condition, dump_clause, predicate::dump,
30634 predicate::remap_after_duplication, predicate::remap_after_inlining,
30635 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30636 (evaluate_conditions_for_known_args): Update.
30637 (set_cond_stmt_execution_predicate): Update.
30638 * ipa-inline.h: Include ipa-predicate.h
30639 (condition, inline_param_summary, conditions, agg_position_info,
30640 predicate): Move to ipa-predicate.h
30641 * ipa-predicate.c: New file.
30642 * ipa-predicate.h: New file.
30644 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30646 * final.c (leaf_function_p): Check we are not in a sequence.
30648 2017-05-18 Martin Liska <mliska@suse.cz>
30650 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30651 * dumpfile.c (dump_register): Use new enum dump_kind.
30652 (get_dump_file_name): Likewise.
30653 (dump_enable_all): Likewise.
30654 (dump_switch_p_1): Likewise.
30655 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30656 * dumpfile.h (enum dump_kind): New enum type.
30657 (struct dump_file_info): Create constructor and
30658 format fields and comments.
30659 * passes.c (pass_manager::register_one_dump_file):
30661 * statistics.c (statistics_early_init): Likewise.
30662 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30663 TDF_TREE with TDF_SLIM.
30664 (gather_memory_references_ref): Likewise.
30666 2017-05-18 Martin Liska <mliska@suse.cz>
30668 * vec.h (struct vnull): Use it.
30670 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30672 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30673 (true_predicate, false_predicate, true_predicate_p,
30674 false_predicate_p): Remove.
30675 (single_cond_predicate, not_inlined_predicate): Turn to member function
30677 (add_condition): Update.
30678 (add_clause): Turn to...
30679 (predicate::add_clause): ... this one; update; allow passing NULL
30681 (and_predicates): Turn to ...
30682 (predicate::operator &=): ... this one.
30683 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30684 (or_predicates): Turn to ...
30685 (predicate::or_with): ... this one.
30686 (evaluate_predicate): Turn to ...
30687 (predicate::evaluate): ... this one.
30688 (predicate_probability): Turn to ...
30689 (predicate::probability): ... this one.
30690 (dump_condition): Update.
30691 (dump_predicate): Turn to ...
30692 (predicate::dump): ... this one.
30693 (account_size_time): Update.
30694 (edge_set_predicate): Update.
30695 (set_hint_predicate): UPdate.
30696 (evaluate_conditions_for_known_args): Update.
30697 (evaluate_properties_for_edge): Update.
30698 (remap_predicate_after_duplication): Turn to...
30699 (predicate::remap_after_duplication): ... this one.
30700 (remap_hint_predicate_after_duplication): Update.
30701 (inline_summary_t::duplicate): UPdate.
30702 (dump_inline_edge_summary): Update.
30703 (dump_inline_summary): Update.
30704 (set_cond_stmt_execution_predicate): Update.
30705 (set_switch_stmt_execution_predicate): Update.
30706 (compute_bb_predicates): Update.
30707 (will_be_nonconstant_expr_predicate): Update.
30708 (will_be_nonconstant_predicate): Update.
30709 (phi_result_unknown_predicate): Update.
30710 (predicate_for_phi_result): Update.
30711 (array_index_predicate): Update.
30712 (estimate_function_body_sizes): Update.
30713 (estimate_node_size_and_time): Update.
30714 (estimate_ipcp_clone_size_and_time): Update.
30715 (remap_predicate): Rename to ...
30716 (predicate::remap_after_inlining): ... this one.
30717 (remap_hint_predicate): Update.
30718 (inline_merge_summary): Update.
30719 (inline_update_overall_summary): Update.
30720 (estimate_size_after_inlining): Update.
30721 (read_predicate): Rename to ...
30722 (predicate::stream_in): ... this one.
30723 (read_inline_edge_summary): Update.
30724 (write_predicate): Rename to ...
30725 (predicate::stream_out): ... this one.
30726 (write_inline_edge_summary): Update.
30727 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
30728 (clause_t): Turn to uint32_t
30729 (predicate): Turn to class; implement constructor and operators
30731 (size_time_entry): Update.
30732 (inline_summary): Update.
30733 (inline_edge_summary): Update.
30735 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
30737 * fold-const.c (fold_binary_loc): Move transformation...
30738 * match.pd (C - X CMP X): ... here.
30740 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30742 * config/sparc/sparc.c (sparc_option_override): Set function
30743 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30744 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30746 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30748 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30750 2017-05-18 Marek Polacek <polacek@redhat.com>
30753 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30754 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30755 gimple_assign_load_p.
30757 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30759 PR middle-end/80692
30760 * real.c (do_compare): Give decimal_do_compare preference over
30761 comparing just the signs.
30763 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30765 * doc/md.texi (Canonicalization of Instructions): Describe the
30766 canonical form of instructions that inherently set a condition
30769 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
30771 PR middle-end/80775
30772 * tree-cfg.c: Move deletion of unreachable case statements to after
30773 the merging of consecutive case labels.
30775 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30777 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30778 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
30779 restoring of callee-saved registers.
30781 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
30783 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
30784 * config/visium/visium.c (single_set_and_flags): Likewise.
30785 * config/visium/visium.md (Substitutions): Likewise.
30787 2017-05-17 Martin Liska <mliska@suse.cz>
30789 * cfg.c: Introduce dump_flags_t type and
30790 use it instead of int type.
30792 * cfghooks.c: Likewise.
30793 * cfghooks.h (struct cfg_hooks): Likewise.
30794 * cfgrtl.c: Likewise.
30795 * cfgrtl.h: Likewise.
30796 * cgraph.c (cgraph_node::get_body): Likewise.
30797 * coretypes.h: Likewise.
30798 * domwalk.c: Likewise.
30799 * domwalk.h: Likewise.
30800 * dumpfile.c (struct dump_option_value_info): Likewise.
30801 (dump_enable_all): Likewise.
30802 (dump_switch_p_1): Likewise.
30803 (opt_info_switch_p): Likewise.
30804 * dumpfile.h (enum tree_dump_index): Likewise.
30805 (struct dump_file_info): Likewise.
30806 * genemit.c: Likewise.
30807 * generic-match-head.c: Likewise.
30808 * gengtype.c (open_base_files): Likewise.
30809 * gimple-pretty-print.c: Likewise.
30810 * gimple-pretty-print.h: Likewise.
30811 * graph.c (print_graph_cfg): Likewise.
30812 * graphite-scop-detection.c (dot_all_sese): Likewise.
30813 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30814 * loop-unroll.c (report_unroll): Likewise.
30815 * passes.c (pass_manager::register_one_dump_file): Likewise.
30816 * print-tree.c: Likewise.
30817 * statistics.c: Likewise.
30818 * tree-cfg.c: Likewise.
30819 * tree-cfg.h: Likewise.
30820 * tree-dfa.c: Likewise.
30821 * tree-dfa.h: Likewise.
30822 * tree-dump.c (dump_function): Likewise.
30823 * tree-dump.h (struct dump_info): Likewise.
30824 * tree-pretty-print.c: Likewise.
30825 * tree-pretty-print.h: Likewise.
30826 * tree-ssa-live.c: Likewise.
30827 * tree-ssa-live.h: Likewise.
30828 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30829 * tree-vect-loop.c: Likewise.
30830 * tree-vect-slp.c: Likewise.
30832 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30833 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30835 PR tree-optimization/80457
30836 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30837 of all arguments to a statement as scalar_to_vec operations.
30838 (vectorizable_call): Adjust call to vect_model_simple_cost for
30840 (vectorizable_conversion): Likewise.
30841 (vectorizable_assignment): Likewise.
30842 (vectorizable_shift): Likewise.
30843 (vectorizable_operation): Likewise.
30844 (vectorizable_comparison): Likewise.
30845 (vect_is_simple_cond): Record the def types for operands.
30846 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30847 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30848 for statement argument count.
30850 2017-05-16 Carl Love <cel@us.ibm.com>
30852 * config/rs6000/rs6000-c: Add support for built-in functions
30853 vector unsigned long long vec_bperm (vector unsigned long long,
30854 vector unsigned char)
30855 vector signed long long vec_mule (vector signed int,
30857 vector unsigned long long vec_mule (vector unsigned int,
30858 vector unsigned int)
30859 vector signed long long vec_mulo (vector signed int,
30861 vector unsigned long long vec_mulo (vector unsigned int,
30862 vector unsigned int)
30863 vector signed char vec_sldw (vector signed char,
30864 vector signed char,
30866 vector unsigned char vec_sldw (vector unsigned char,
30867 vector unsigned char,
30869 vector signed short vec_sldw (vector signed short,
30870 vector signed short,
30872 vector unsigned short vec_sldw (vector unsigned short,
30873 vector unsigned short,
30875 vector signed int vec_sldw (vector signed int,
30878 vector unsigned int vec_sldw (vector unsigned int,
30879 vector unsigned int,
30881 vector signed long long vec_sldw (vector signed long long,
30882 vector signed long long,
30884 vector unsigned long long vec_sldw (vector unsigned long long,
30885 vector unsigned long long,
30887 * config/rs6000/rs6000-c: Add support for built-in functions
30888 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30889 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30890 * doc/extend.texi: Update the built-in documentation for the
30891 new built-in functions.
30893 2017-05-16 Marek Polacek <polacek@redhat.com>
30897 * tree.c (save_expr): Don't fold the expression.
30899 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30901 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30902 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30903 and (?*y,m). Update insn attributes.
30905 2017-05-16 Martin Liska <mliska@suse.cz>
30907 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30908 flags argument of print_gimple_stmt, print_gimple_expr,
30909 print_generic_stmt and print_generic_expr.
30910 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30911 * coretypes.h: Likewise.
30912 * except.c (dump_eh_tree): Likewise.
30913 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30914 * gimple-pretty-print.h: Likewise.
30915 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30916 (backprop::push_to_worklist): Likewise.
30917 (backprop::pop_from_worklist): Likewise.
30918 (backprop::process_use): Likewise.
30919 (backprop::intersect_uses): Likewise.
30920 (note_replacement): Likewise.
30921 * gimple-ssa-store-merging.c
30922 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30923 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30924 (pass_store_merging::execute): Likewise.
30925 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30926 (ssa_base_cand_dump_callback): Likewise.
30927 (dump_incr_vec): Likewise.
30928 (replace_refs): Likewise.
30929 (replace_mult_candidate): Likewise.
30930 (create_add_on_incoming_edge): Likewise.
30931 (create_phi_basis): Likewise.
30932 (insert_initializers): Likewise.
30933 (all_phi_incrs_profitable): Likewise.
30934 (introduce_cast_before_cand): Likewise.
30935 (replace_one_candidate): Likewise.
30936 * gimplify.c (gimplify_expr): Likewise.
30937 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30938 (set_rename): Likewise.
30939 (rename_uses): Likewise.
30940 (copy_loop_phi_nodes): Likewise.
30941 (add_close_phis_to_merge_points): Likewise.
30942 (copy_loop_close_phi_args): Likewise.
30943 (copy_cond_phi_args): Likewise.
30944 (graphite_copy_stmts_from_block): Likewise.
30945 (translate_pending_phi_nodes): Likewise.
30946 * graphite-poly.c (print_pdr): Likewise.
30947 (dump_gbb_cases): Likewise.
30948 (dump_gbb_conditions): Likewise.
30949 (print_scop_params): Likewise.
30950 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30951 (build_cross_bb_scalars_use): Likewise.
30952 (gather_bbs::before_dom_children): Likewise.
30953 * hsa-dump.c (dump_hsa_immed): Likewise.
30954 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30955 (get_replacement_map): Likewise.
30956 * ipa-inline-analysis.c (dump_condition): Likewise.
30957 (estimate_function_body_sizes): Likewise.
30958 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30959 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30960 * ipa-prop.c (ipa_dump_param): Likewise.
30961 (ipa_print_node_jump_functions_for_edge): Likewise.
30962 (ipa_modify_call_arguments): Likewise.
30963 (ipa_modify_expr): Likewise.
30964 (ipa_dump_param_adjustments): Likewise.
30965 (ipa_dump_agg_replacement_values): Likewise.
30966 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30967 * ipa-pure-const.c (check_stmt): Likewise.
30968 (pass_nothrow::execute): Likewise.
30969 * ipa-split.c (execute_split_functions): Likewise.
30970 * omp-offload.c (dump_oacc_loop_part): Likewise.
30971 (dump_oacc_loop): Likewise.
30972 * trans-mem.c (tm_log_emit): Likewise.
30973 (tm_memopt_accumulate_memops): Likewise.
30974 (dump_tm_memopt_set): Likewise.
30975 (dump_tm_memopt_transform): Likewise.
30976 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30977 (print_loop): Likewise.
30978 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30979 (chkp_gather_checks_info): Likewise.
30980 (chkp_get_check_result): Likewise.
30981 (chkp_remove_check_if_pass): Likewise.
30982 (chkp_use_outer_bounds_if_possible): Likewise.
30983 (chkp_reduce_bounds_lifetime): Likewise.
30984 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30985 (chkp_mark_completed_bounds): Likewise.
30986 (chkp_register_incomplete_bounds): Likewise.
30987 (chkp_mark_invalid_bounds): Likewise.
30988 (chkp_maybe_copy_and_register_bounds): Likewise.
30989 (chkp_build_returned_bound): Likewise.
30990 (chkp_get_bound_for_parm): Likewise.
30991 (chkp_build_bndldx): Likewise.
30992 (chkp_get_bounds_by_definition): Likewise.
30993 (chkp_generate_extern_var_bounds): Likewise.
30994 (chkp_get_bounds_for_decl_addr): Likewise.
30995 * tree-chrec.c (chrec_apply): Likewise.
30996 * tree-data-ref.c (dump_data_reference): Likewise.
30997 (dump_subscript): Likewise.
30998 (dump_data_dependence_relation): Likewise.
30999 (analyze_overlapping_iterations): Likewise.
31000 * tree-inline.c (expand_call_inline): Likewise.
31001 (tree_function_versioning): Likewise.
31002 * tree-into-ssa.c (dump_defs_stack): Likewise.
31003 (dump_currdefs): Likewise.
31004 (dump_names_replaced_by): Likewise.
31005 (dump_update_ssa): Likewise.
31006 (update_ssa): Likewise.
31007 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31008 * tree-parloops.c (build_new_reduction): Likewise.
31009 (try_create_reduction_list): Likewise.
31010 (ref_conflicts_with_region): Likewise.
31011 (oacc_entry_exit_ok_1): Likewise.
31012 (oacc_entry_exit_single_gang): Likewise.
31013 * tree-pretty-print.h: Likewise.
31014 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31015 (get_scalar_evolution): Likewise.
31016 (add_to_evolution): Likewise.
31017 (get_loop_exit_condition): Likewise.
31018 (analyze_evolution_in_loop): Likewise.
31019 (analyze_initial_condition): Likewise.
31020 (analyze_scalar_evolution): Likewise.
31021 (instantiate_scev): Likewise.
31022 (number_of_latch_executions): Likewise.
31023 (gather_chrec_stats): Likewise.
31024 (final_value_replacement_loop): Likewise.
31025 (scev_const_prop): Likewise.
31026 * tree-sra.c (dump_access): Likewise.
31027 (disqualify_candidate): Likewise.
31028 (create_access): Likewise.
31029 (reject): Likewise.
31030 (maybe_add_sra_candidate): Likewise.
31031 (create_access_replacement): Likewise.
31032 (analyze_access_subtree): Likewise.
31033 (analyze_all_variable_accesses): Likewise.
31034 (sra_modify_assign): Likewise.
31035 (initialize_constant_pool_replacements): Likewise.
31036 (find_param_candidates): Likewise.
31037 (decide_one_param_reduction): Likewise.
31038 (replace_removed_params_ssa_names): Likewise.
31039 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31040 * tree-ssa-copy.c (dump_copy_of): Likewise.
31041 (copy_prop_visit_cond_stmt): Likewise.
31042 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31043 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31044 (record_equivalences_from_stmt): Likewise.
31045 * tree-ssa-dse.c (compute_trims): Likewise.
31046 (delete_dead_call): Likewise.
31047 (delete_dead_assignment): Likewise.
31048 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31049 (forward_propagate_into_cond): Likewise.
31050 (pass_forwprop::execute): Likewise.
31051 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31052 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31054 (move_computations_worker): Likewise.
31055 (execute_sm): Likewise.
31056 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31057 (remove_exits_and_undefined_stmts): Likewise.
31058 (remove_redundant_iv_tests): Likewise.
31059 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31060 (adjust_iv_update_pos): Likewise.
31061 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31062 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31063 (value_replacement): Likewise.
31064 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31065 * tree-ssa-pre.c (print_pre_expr): Likewise.
31066 (get_representative_for): Likewise.
31067 (create_expression_by_pieces): Likewise.
31068 (insert_into_preds_of_block): Likewise.
31069 (eliminate_insert): Likewise.
31070 (eliminate_dom_walker::before_dom_children): Likewise.
31071 (eliminate): Likewise.
31072 (remove_dead_inserted_code): Likewise.
31073 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31074 * tree-ssa-reassoc.c (get_rank): Likewise.
31075 (eliminate_duplicate_pair): Likewise.
31076 (eliminate_plus_minus_pair): Likewise.
31077 (eliminate_not_pairs): Likewise.
31078 (undistribute_ops_list): Likewise.
31079 (eliminate_redundant_comparison): Likewise.
31080 (update_range_test): Likewise.
31081 (optimize_range_tests_var_bound): Likewise.
31082 (optimize_vec_cond_expr): Likewise.
31083 (rewrite_expr_tree): Likewise.
31084 (rewrite_expr_tree_parallel): Likewise.
31085 (linearize_expr): Likewise.
31086 (break_up_subtract): Likewise.
31087 (linearize_expr_tree): Likewise.
31088 (attempt_builtin_powi): Likewise.
31089 (attempt_builtin_copysign): Likewise.
31090 (transform_stmt_to_copy): Likewise.
31091 (transform_stmt_to_multiply): Likewise.
31092 (dump_ops_vector): Likewise.
31093 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31094 (print_scc): Likewise.
31095 (set_ssa_val_to): Likewise.
31096 (visit_reference_op_store): Likewise.
31097 (visit_use): Likewise.
31098 (sccvn_dom_walker::before_dom_children): Likewise.
31099 (run_scc_vn): Likewise.
31100 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31102 (expr_hash_elt::print): Likewise.
31103 (const_and_copies::pop_to_marker): Likewise.
31104 (const_and_copies::record_const_or_copy_raw): Likewise.
31105 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31106 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31107 (dump_predicates): Likewise.
31108 (find_uninit_use): Likewise.
31109 (warn_uninitialized_phi): Likewise.
31110 (pass_late_warn_uninitialized::execute): Likewise.
31111 * tree-ssa.c (verify_vssa): Likewise.
31112 (verify_ssa): Likewise.
31113 (maybe_optimize_var): Likewise.
31114 * tree-vrp.c (dump_value_range): Likewise.
31115 (dump_all_value_ranges): Likewise.
31116 (dump_asserts_for): Likewise.
31117 (register_edge_assert_for_2): Likewise.
31118 (vrp_visit_cond_stmt): Likewise.
31119 (vrp_visit_switch_stmt): Likewise.
31120 (vrp_visit_stmt): Likewise.
31121 (vrp_visit_phi_node): Likewise.
31122 (simplify_cond_using_ranges_1): Likewise.
31123 (fold_predicate_in): Likewise.
31124 (evrp_dom_walker::before_dom_children): Likewise.
31125 (evrp_dom_walker::push_value_range): Likewise.
31126 (evrp_dom_walker::pop_value_range): Likewise.
31127 (execute_early_vrp): Likewise.
31129 2017-05-16 Richard Biener <rguenther@suse.de>
31131 * dwarf2out.c (loc_list_from_tree_1): Do not create
31132 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31134 2017-05-16 Richard Biener <rguenther@suse.de>
31136 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31138 (note_variable_value_in_expr): If we resolved the decl ref
31139 do not push to the stack.
31141 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31143 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31144 operations in fast-math mode.
31145 (vaddq_f16): Likewise.
31146 (vmul_f16): Likewise.
31147 (vmulq_f16): Likewise.
31148 (vsub_f16): Likewise.
31149 (vsubq_f16): Likewise.
31150 * config/arm/neon.md (add<mode>3): New.
31152 (fma:<VH:mode>3): New. Also remove outdated comment.
31155 2017-05-16 Martin Liska <mliska@suse.cz>
31159 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31160 (odr_types_equivalent_p): Likewise.
31162 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31164 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31166 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31169 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31170 non-interunit SSE move alternatives with '?'.
31171 (zero-extendsidi peephole2): New peephole to skip intermediate
31172 general register in SSE zero-extend sequence.
31174 2017-05-15 Jeff Law <law@redhat.com>
31176 * reorg.c (relax_delay_slots): Create a new variable to hold
31177 the temporary target rather than clobbering TARGET_LABEL.
31179 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31180 missing argument to extract_bit_field call.
31181 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31183 2017-05-15 Martin Liska <mliska@suse.cz>
31186 * gcc.c (process_command): Do not allow empty argument of -o option.
31188 2017-05-15 Renlin Li <renlin.li@arm.com>
31190 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31191 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31192 * config/aarch64/constraints.md (Usf): Add long call check.
31193 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31194 (call_value): Likewise.
31195 (sibcall): Likewise.
31196 (sibcall_value): Likewise.
31198 (call_value_insn): New.
31199 (sibcall_insn): Update rtx pattern.
31200 (sibcall_value_insn): Likewise.
31201 (call_internal): Remove.
31202 (call_value_internal): Likewise.
31203 (sibcall_internal): Likewise.
31204 (sibcall_value_internal): Likewise.
31205 (call_reg): Likewise.
31206 (call_symbol): Likewise.
31207 (call_value_reg): Likewise.
31208 (call_value_symbol): Likewise.
31210 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31213 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31215 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31217 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31218 compatible with CCGOCmode and with CCZmode.
31220 2017-05-14 Martin Sebor <msebor@redhat.com>
31222 PR middle-end/77671
31223 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31224 (gimple_fold_builtin_snprintf): Same.
31225 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31226 (gimple_fold_builtin_snprintf): Same.
31227 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31228 of character types.
31229 (is_call_safe): New function.
31230 (try_substitute_return_value): Call it.
31231 (try_simplify_call): New function.
31232 (pass_sprintf_length::handle_gimple_call): Call it.
31234 2017-05-14 Martin Sebor <msebor@redhat.com>
31236 PR middle-end/80669
31237 * builtins.c (expand_builtin_stpncpy): Simplify.
31239 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31241 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31242 * config/i386/i386.h
31243 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31244 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31245 (struct machine_function): Add new members call_ms2sysv,
31246 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31247 (struct machine_frame_state): New fields sp_realigned and
31248 sp_realigned_offset.
31249 * config/i386/i386.c
31250 (enum xlogue_stub): New enum.
31251 (enum xlogue_stub_sets): New enum.
31252 (class xlogue_layout): New class.
31253 (struct ix86_frame): New fields stack_realign_allocate_offset,
31254 stack_realign_offset and outlined_save_offset. Modify comments to
31255 detail stack layout when using out-of-line stubs.
31256 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31257 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31258 -mcall-ms2sysv-xlogues.
31259 (stub_managed_regs): New static variable.
31260 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31261 registers managed by out-of-line stub.
31262 (disable_call_ms2sysv_xlogues): New function.
31263 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31264 m->call_ms2sysv when appropriate and compute frame layout for
31266 (sp_valid_at, fp_valid_at): New inline functions.
31267 (choose_basereg): New function.
31268 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31270 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31271 Use align parameter of choose_baseaddr to generated aligned SSE movs
31273 (pro_epilogue_adjust_stack): Modify to track
31274 machine_frame_state::sp_realigned.
31275 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31276 (ix86_nsaved_sseregs): Likewise.
31277 (ix86_emit_save_regs): Likewise.
31278 (ix86_emit_save_regs_using_mov): Likewise.
31279 (ix86_emit_save_sse_regs_using_mov): Likewise.
31280 (get_scratch_register_on_entry): Likewise.
31281 (gen_frame_set): New function.
31282 (gen_frame_load): Likewise.
31283 (gen_frame_store): Likewise.
31284 (emit_outlined_ms2sysv_save): Likewise.
31285 (emit_outlined_ms2sysv_restore): Likewise.
31286 (ix86_expand_prologue): Modify stack re-alignment code and call
31287 emit_outlined_ms2sysv_save when appropriate.
31288 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31289 parameter rtx_insn *insn, which allows the function to be used to only
31290 generate the notes.
31291 (ix86_expand_epilogue): Modify validity checks of frame and stack
31292 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31293 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31294 * config/i386/predicates.md
31295 (save_multiple): New predicate.
31296 (restore_multiple): Likewise.
31297 * config/i386/sse.md
31298 (save_multiple<mode>): New pattern.
31299 (save_multiple_realign<mode>): Likewise.
31300 (restore_multiple<mode>): Likewise.
31301 (restore_multiple_and_return<mode>): Likewise.
31302 (restore_multiple_leave_return<mode>): Likewise.
31303 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31305 2017-05-14 Julia Koval <julia.koval@intel.com>
31307 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31308 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31309 (__builtin_ia32_xsetbv): New builtins.
31310 * config/i386/i386.c (ix86_expand_special_args_builtin):
31312 (ix86_expand_builtin): Special expand for new intrinsics.
31313 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31314 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31315 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31317 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31319 * cfganal.c (inverted_post_order_compute): Change argument type
31321 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31322 * df-core.c (rest_of_handle_df_initialize): Adjust.
31323 (rest_of_handle_df_finish): Likewise.
31324 (df_analyze_1): Likewise.
31325 (df_analyze): Likewise.
31326 (loop_inverted_post_order_compute): Change argument to be a vec *.
31327 (df_analyze_loop): Adjust.
31328 (df_get_n_blocks): Likewise.
31329 (df_get_postorder): Likewise.
31330 * df.h (struct df_d): Change field to be a vec.
31331 * lcm.c (compute_laterin): Adjust.
31332 (compute_available): Likewise.
31333 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31334 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31335 * tree-ssa-pre.c (compute_antic): Likewise.
31337 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31339 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31340 (depth_first_search::depth_first_search): Change structure init
31341 function to this constructor.
31342 (depth_first_search::add_bb): Rename function to this member.
31343 (depth_first_search::execute): Likewise.
31344 (flow_dfs_compute_reverse_finish): Adjust.
31346 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31348 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31349 (longest_simple_path): Likewise.
31350 * shrink-wrap.c (spread_components): Likewise.
31351 (disqualify_problematic_components): Likewise.
31352 (emit_common_heads_for_components): Likewise.
31353 (emit_common_tails_for_components): Likewise.
31354 (insert_prologue_epilogue_for_components): Likewise.
31356 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31358 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31361 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31363 * df-core.c (df_set_blocks): Start using auto_bitmap.
31364 (df_compact_blocks): Likewise.
31365 * df-problems.c (df_rd_confluence_n): Likewise.
31366 * df-scan.c (df_insn_rescan_all): Likewise.
31367 (df_process_deferred_rescans): Likewise.
31368 (df_update_entry_block_defs): Likewise.
31369 (df_update_exit_block_uses): Likewise.
31370 (df_entry_block_bitmap_verify): Likewise.
31371 (df_exit_block_bitmap_verify): Likewise.
31372 (df_scan_verify): Likewise.
31373 * lra-constraints.c (lra_constraints): Likewise.
31374 (undo_optional_reloads): Likewise.
31375 (lra_undo_inheritance): Likewise.
31376 * lra-remat.c (calculate_gen_cands): Likewise.
31377 (do_remat): Likewise.
31378 * lra-spills.c (assign_spill_hard_regs): Likewise.
31379 (spill_pseudos): Likewise.
31380 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31381 (bitmap_set_subtract_values): Likewise.
31383 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31385 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31386 management with auto_bitmap.
31387 (fix_inter_tick): Likewise.
31388 (fix_recovery_deps): Likewise.
31389 * ira.c (add_store_equivs): Likewise.
31390 (find_moveable_pseudos): Likewise.
31391 (split_live_ranges_for_shrink_wrap): Likewise.
31392 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31393 (rtx_reuse_manager::seen_def_p): Likewise.
31394 (rtx_reuse_manager::set_seen_def): Likewise.
31395 * print-rtl.h (class rtx_reuse_manager): Likewise.
31397 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31399 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31401 (migrate_btr_def): Likewise.
31402 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31403 * df-core.c (loop_post_order_compute): Likewise.
31404 (loop_inverted_post_order_compute): Likewise.
31405 * hsa-common.h: Likewise.
31406 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31407 * init-regs.c (initialize_uninitialized_regs): Likewise.
31408 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31409 (inline_small_functions): Likewise.
31410 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31411 * ira.c (combine_and_move_insns): Likewise.
31412 (build_insn_chain): Likewise.
31413 * loop-invariant.c (find_invariants): Likewise.
31414 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31415 * predict.c (tree_predict_by_opcode): Likewise.
31416 (predict_paths_leading_to): Likewise.
31417 (predict_paths_leading_to_edge): Likewise.
31418 (estimate_loops_at_level): Likewise.
31419 (estimate_loops): Likewise.
31420 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31421 (spread_components): Likewise.
31422 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31423 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31424 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31425 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31426 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31427 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31428 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31429 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31430 (mark_threaded_blocks): Likewise.
31431 (thread_through_all_blocks): Likewise.
31432 * tree-ssa.c (verify_ssa): Likewise.
31433 (execute_update_addresses_taken): Likewise.
31434 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31436 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31438 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31440 (post_order_compute): Likewise.
31441 (inverted_post_order_compute): Likewise.
31442 (pre_and_rev_post_order_compute_fn): Likewise.
31444 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31446 * genrecog.c (int_set::int_set): Explicitly construct our
31447 auto_vec base class.
31448 * vec.h (auto_vec::auto_vec): New constructor.
31450 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31452 * bitmap.h (class auto_bitmap): New constructor taking
31453 bitmap_obstack * argument.
31455 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31457 * bitmap.h (class auto_bitmap): Change type of m_bits to
31458 bitmap_head, and adjust ctor / dtor and member operators.
31460 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31462 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31463 when returned register mode doesn't match original mode.
31465 2017-05-12 Jeff Law <law@redhat.com>
31466 Jakub Jelinek <jakub@redhat.com>
31468 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31469 we look for cc setter after the compare-elim changes.
31470 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31471 within the vector to match what compare-elim now expects.
31472 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31473 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31475 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31476 after the compare-elim changes.
31477 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31478 the vector to match what compare-elim now expects.
31479 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31480 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31481 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31482 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31483 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31485 * config/visium/visium.c (single_set_and_flags): Fix where
31486 we look for cc setter after the compare-elim changes.
31487 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31488 with the vector to match what compare-elim now expects.
31489 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31490 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31491 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31492 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31493 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31494 (neg<mode>2_insn_set_overflow): Likewise.
31496 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31498 PR middle-end/79794
31499 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31500 maybe_expand_insn call, set ops[0].target. If still set after call,
31501 set alt_rtl. Add extra arg to recursive calls.
31502 (extract_bit_field): Add alt_rtl argument. Pass to
31504 * expmed.h (extract_bit_field): Fix prototype.
31505 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31506 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31507 to extract_bit_field_calls.
31508 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31509 Pass alt_rtl to extract_bit_field calls.
31510 * calls.c (store_unaligned_arguments_into_psuedos)
31511 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31512 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31514 * optabs.h (struct expand_operand): Add target bitfield.
31516 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31518 * compare-elim.c (try_eliminate_compare): Canonicalize
31519 operation with embedded compare to
31520 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31521 (set (reg) (operation)].
31523 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31525 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31528 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31529 cost of adding a carry flag for ADC instruction.
31530 [case MINUS]: Ignore the cost of subtracting a carry flag
31531 for SBB instruction.
31533 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31535 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31537 * config/rs6000/bmiintrin.h: New file.
31538 * config/rs6000/bmi2intrin.h: New file.
31539 * config/rs6000/x86intrin.h: New file.
31541 2017-05-12 Jeff Law <law@redhat.com>
31543 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31546 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31548 PR middle-end/80707
31549 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31551 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31553 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31554 early expansion of vector divide builtins.
31555 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31556 builtins identified as having unsigned arguments.
31558 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31560 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31561 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31562 expansion of vector logical operations (and, andc, or, xor,
31565 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31567 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31568 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31570 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31572 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31573 early GIMPLE expansion of vector multiplies.
31575 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31577 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31578 TARGET_HAVE_MOVT conditional.
31579 (movt splitter): Likewise.
31581 2017-05-12 Richard Biener <rguenther@suse.de>
31583 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31584 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31585 Fold all stmts not inplace.
31587 2017-05-12 Richard Biener <rguenther@suse.de>
31589 PR tree-optimization/80713
31590 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31591 inserted_exprs bit for not removed stmts.
31593 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31595 PR middle-end/69921
31596 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31597 parallelized" attribute for parallelized OpenACC kernels.
31598 * omp-offload.c (execute_oacc_device_lower): Use it.
31600 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31601 Set "oacc kernels" attribute.
31602 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31603 parameter. Adjust all users.
31604 (oacc_fn_attrib_kernels_p): Remove function.
31605 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31606 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31607 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31608 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31609 assert "oacc kernels" attribute is set.
31611 2017-05-11 Carl Love <cel@us.ibm.com>
31613 * config/rs6000/rs6000-c: Add support for built-in functions
31614 vector unsigned char vec_popcnt (vector signed char)
31615 vector unsigned char vec_popcnt (vector unsigned char)
31616 vector unsigned short vec_popcnt (vector signed short)
31617 vector unsigned short vec_popcnt (vector unsigned short)
31618 vector unsigned int vec_popcnt (vector signed int)
31619 vector unsigned int vec_popcnt (vector unsigned int)
31620 vector unsigned long long vec_popcnt (vector signed long long)
31621 vector unsigned long long vec_popcnt (vector unsigned long long)
31622 vector signed long long vec_slo (vector signed long long,
31623 vector signed char)
31624 vector signed long long vec_slo (vector signed long long,
31625 vector unsigned char)
31626 vector unsigned long long vec_slo (vector unsigned long long,
31627 vector signed char)
31628 vector unsigned long long vec_slo (vector unsigned long long,
31629 vector unsigned char)
31630 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31631 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31632 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31633 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31634 * doc/extend.texi: Update the built-in documentation file for the
31635 new built-in functions.
31637 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31639 * attribs.h (sorted_attr_string): Move machine independent
31640 functions for target clone support from the i386 port to common
31641 code. Rename ix86_function_versions to common_function_versions.
31642 Rename make_name to make_unique_name.
31643 (common_function_versions): Likewise.
31644 (make_unique_name): Likewise.
31645 (make_dispatcher_decl): Likewise.
31646 (is_function_default_version): Likewise.
31647 * attribs.c (attr_strcmp): Likewise.
31648 (sorted_attr_string): Likewise.
31649 (common_function_versions): Likewise.
31650 (make_unique_name): Likewise.
31651 (make_dispatcher_decl): Likewise.
31652 (is_function_default_version): Likewise.
31653 * config/i386/i386.c (attr_strcmp): Likewise.
31654 (sorted_attr_string): Likewise.
31655 (ix86_function_versions): Likewise.
31656 (make_name): Likewise.
31657 (make_dispatcher_decl): Likewise.
31658 (is_function_default_version): Likewise.
31659 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31661 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31664 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31665 Account for direct move costs for vec_construct of integer
31668 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31671 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31672 (UNSPEC_STX_ATOMIC): Ditto.
31673 (loaddi_via_sse): New insn.
31674 (storedi_via_sse): Ditto.
31675 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31676 Update corresponding peephole2 patterns.
31677 (atomic_storedi_fpu): Ditto.
31679 2017-05-11 Julia Koval <julia.koval@intel.com>
31681 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31682 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31684 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31685 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31686 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31688 2017-05-11 Nathan Sidwell <nathan@acm.org>
31690 * graphite-poly.c: Include dumpfile.h.
31692 * dumpfle.h (dump_function): Declare here ...
31693 * tree-dump.h (dump_function): ... not here.
31694 * dumpfile.c: #include tree-cfg.h.
31695 (dump_function): Move here from ...
31696 * tree-dump.c (dump_function): ... here.
31697 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31698 * graphite-poly.c: Don't include tree-dump.h.
31699 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31700 * print-tree.c: Likewise.
31701 * stor-layout.c: Likewise.
31702 * tree-nested.c: Likewise.
31704 * dumpfile.c (dump_start): Use TDF_FLAGS.
31705 (dump_enable_all): Fix TDF_KIND check thinko.
31707 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31709 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31710 array entries to represent two legal parameterizations of the
31711 overloaded __builtin_cmpb function, as represented by the
31712 P6_OV_BUILTIN_CMPB constant.
31713 (altivec_resolve_overloaded_builtin): Add special case handling
31714 for the __builtin_cmpb function, as represented by the
31715 P6_OV_BUILTIN_CMPB constant.
31716 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
31717 (BU_P6_64BIT_2): New macro.
31718 (BU_P6_OVERLOAD_2): New macro
31719 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
31720 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
31721 (CMPB): Add overload support to represent both 32-bit and 64-bit
31722 compare-bytes function.
31723 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
31724 support for TARGET_CMPB.
31725 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
31726 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31727 documentation of the __builtin_cmpb overloaded built-in function.
31729 2017-05-11 Richard Biener <rguenther@suse.de>
31731 PR tree-optimization/80705
31732 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
31733 bases are not vectorizable.
31735 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31737 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
31738 when counting register pressure.
31740 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31742 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31743 (iv_ca_more_deps): Renamed to ...
31744 (iv_ca_compare_deps): ... this.
31745 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31747 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31749 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31751 (determine_group_iv_costs): ... here.
31752 (find_inv_vars_cb): Record inv var if it's not recorded before.
31754 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31756 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31757 (get_shiftadd_cost): Ditto.
31759 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31761 * tree-ssa-address.c: Include header file.
31762 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31764 (add_to_parts): Refactor.
31765 (addr_to_parts): New parameter. Update use of move_hint_to_base.
31766 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
31769 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31771 PR tree-optimization/53090
31772 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31774 (extract_cond_operands): Detect condition with IV on both sides
31775 and return COMP_IV_EXPR_2.
31776 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31777 (rewrite_use_compare): Simplify by removing call to function
31778 extract_cond_operands.
31780 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31782 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
31783 (extract_cond_operands): Detect condition comparing against non-
31784 invariant bound and return appropriate enum value.
31785 (find_interesting_uses_cond): Update use of extract_cond_operands.
31786 Handle its return value accordingly.
31787 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
31789 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31791 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
31792 nonlinear iv_use computation in loop invariant sensitive way.
31794 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31796 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
31797 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31799 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31801 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31802 (dump_cand): Support iv_cand.inv_exprs.
31803 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31805 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31808 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31810 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31812 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31813 as local function. Include necessary header files.
31814 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31816 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31818 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31820 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31822 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31823 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31824 RSHIFT_EXPR and BIT_NOT_EXPR.
31826 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31828 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31829 (adjust_setup_cost): New parameter supporting round up adjustment.
31830 (struct address_cost_data): Delete.
31831 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31832 (split_address_cost, ptr_difference_cost): Delete.
31833 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31834 (struct ainc_cost_data): New struct.
31835 (get_address_cost_ainc): New function.
31836 (get_address_cost, get_computation_cost): Reimplement.
31837 (determine_group_iv_cost_address): Record inv_expr for all uses of
31839 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31840 (iv_ca_has_deps): Reimplemented to ...
31841 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31843 (iv_ca_extend): Call iv_ca_more_deps.
31845 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31847 * tree-ssa-address.c (struct mem_address): Move to header file.
31848 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31849 * tree-ssa-address.h (struct mem_address): Move from C file.
31850 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31852 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31854 * tree-affine.h (aff_combination_type): New interface.
31855 (aff_combination_zero_p): Remove static.
31856 (aff_combination_const_p): New interface.
31857 (aff_combination_singleton_var_p): New interfaces.
31859 2017-05-11 Richard Biener <rguenther@suse.de>
31861 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31862 Skip unreachable blocks and destinations.
31863 (eliminate): Move stmt removal and fixup ...
31864 (fini_eliminate): ... here. Skip inserted exprs.
31865 (pass_pre::execute): Move fini_pre after fini_eliminate.
31866 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31867 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31868 PRE to get rid of dead code that has invalid SSA form and
31869 split critical edges again.
31871 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31873 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31875 2017-05-11 Richard Biener <rguenther@suse.de>
31877 * passes.c (execute_function_todo): Verify loops if they are
31878 said to be up-to-date.
31879 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31880 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31882 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31885 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31886 handle calling assemble_external ourself.
31889 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31890 modes with zero size. Enhance comment.
31892 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31894 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31895 built-ins for vec_xl and vec_xst with short and char pointer
31898 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31900 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31901 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31902 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31903 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31904 (_mm_maskz_min_round_ss): New intrinsics.
31905 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31906 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31907 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31908 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31909 (__builtin_ia32_minss_mask_round): New builtins.
31910 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31911 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31912 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31914 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31915 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31917 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31920 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31922 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31923 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31924 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31925 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31926 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31927 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31928 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31929 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31930 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31931 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31932 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31933 (__builtin_ia32_mulss_mask_round): New builtins.
31934 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31935 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31936 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31938 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31939 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31941 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31944 2017-05-10 Julia Koval <julia.koval@intel.com>
31946 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31947 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31948 (_mm256_setr_m128i): New intrinsics.
31950 2017-05-10 Julia Koval <julia.koval@intel.com>
31952 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31953 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31954 (_mm_maskz_rcp14_ss): New intrinsics.
31955 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31956 (__builtin_ia32_rcp14ss_mask): New builtins.
31957 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31959 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31961 PR tree-optimization/51513
31962 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31963 (assert_unreachable_fallthru_edge_p): Use it.
31964 (group_case_labels_stmt): Likewise.
31965 * tree-cfg.h: Prototype it.
31966 * stmt.c: Include cfghooks.h and tree-cfg.h.
31967 (emit_case_dispatch_table) <gap_label>: New local variable.
31968 Use it to fill dispatch table gaps.
31969 Test for default_label before updating probabilities.
31970 (expand_case) <default_label>: Remove unneeded initialization.
31971 Test for unreachable default case statement and remove its edge.
31972 Set default_label accordingly.
31973 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31975 2017-05-10 Carl Love <cel@us.ibm.com>
31977 * config/rs6000/rs6000-c: Add support for built-in functions
31978 vector signed char vec_neg (vector signed char)
31979 vector signed short int vec_neg (vector short int)
31980 vector signed int vec_neg (vector signed int)
31981 vector signed long long vec_neg (vector signed long long)
31982 vector float vec_neg (vector float)
31983 vector double vec_neg (vector double)
31984 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31986 * config/rs6000/altivec.h: Add define for vec_neg
31987 * doc/extend.texi: Update the built-in documentation for the
31988 new built-in functions.
31990 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31992 PR tree-optimization/77644
31993 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31995 2017-05-10 Nathan Sidwell <nathan@acm.org>
31997 * dumpfile.h (TDI_lang_all): New.
31998 (TDF_KIND): New. Renumber others
31999 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32001 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32003 (get_dump_file_name): Adjust suffix generation.
32004 (dump_enable_all): Use TDF_KIND.
32005 * doc/invoke.texi (-fdump-lang-all): Document.
32007 * dumpfile.h: Tabify.
32009 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32012 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32013 Move member access before delete.
32015 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32017 * tree-inline.c (expand_call_inline): Split block at stmt
32020 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32023 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32024 are now unused after splitting mov{sf,sd}_hardfloat.
32025 (f32_lr2): Likewise.
32026 (f32_lm): Likewise.
32027 (f32_lm2): Likewise.
32028 (f32_li): Likewise.
32029 (f32_li2): Likewise.
32030 (f32_lv): Likewise.
32031 (f32_sr): Likewise.
32032 (f32_sr2): Likewise.
32033 (f32_sm): Likewise.
32034 (f32_sm2): Likewise.
32035 (f32_si): Likewise.
32036 (f32_si2): Likewise.
32037 (f32_sv): Likewise.
32038 (f32_dm): Likewise.
32039 (f32_vsx): Likewise.
32040 (f32_av): Likewise.
32041 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32042 For movsf, order stores so the VSX stores occur before the GPR
32043 store which encourages the register allocator to use a traditional
32044 FPR instead of a GPR. For movsd, order the stores so that the GPR
32045 store comes before the VSX stores to allow the power6 to work.
32046 This is due to the power6 not having a 32-bit integer store
32047 instruction from a FPR.
32048 (movsf_hardfloat): Likewise.
32049 (movsd_hardfloat): Likewise.
32051 2017-05-09 Martin Sebor <msebor@redhat.com>
32053 PR translation/80280
32054 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32057 PR translation/80280
32058 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32059 data member added in r247778.
32060 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32062 2017-05-09 Nathan Sidwell <nathan@acm.org>
32064 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32066 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32069 2017-05-09 Marek Polacek <polacek@redhat.com>
32071 * doc/invoke.texi: Fix typo.
32073 2017-05-09 Richard Biener <rguenther@suse.de>
32075 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32076 (vrp_val_is_min): Likewise.
32077 (set_value_range_to_value): Likewise.
32078 (set_value_range_to_nonnegative): Likewise.
32079 (gimple_assign_nonzero_p): Likewise.
32080 (gimple_stmt_nonzero_p): Likewise.
32081 (vrp_int_const_binop): Likewise. Remove unreachable case.
32082 (adjust_range_with_scev): Adjust comments.
32083 (compare_range_with_value): Likewise.
32084 (extract_range_from_phi_node): Likewise.
32085 (test_for_singularity): Likewise.
32087 2017-05-09 Richard Biener <rguenther@suse.de>
32089 * tree-vrp.c (get_single_symbol): Add assert that we don't
32090 get overflowed constants as invariant part.
32091 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32092 checks. Use wi::cmp instead of recursing for integer constants.
32093 (compare_values): Just ignore whether we assumed undefined
32094 overflow instead of failing the compare.
32095 (extract_range_for_var_from_comparison_expr): Add comment before the
32096 TREE_NO_WARNING sets.
32097 (test_for_singularity): Likewise.
32098 (extract_range_from_comparison): Do not disable optimization
32099 when we assumed undefined overflow.
32100 (extract_range_basic): Remove init of unused var.
32102 2017-05-09 Richard Biener <rguenther@suse.de>
32104 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32105 (extract_range_from_multiplicative_op_1): Adjust.
32106 (extract_range_from_binary_expr_1): Use int_const_binop.
32108 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32111 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32112 rs6000_store_data_bypass_p in seven define_bypass directives and
32113 in several comments.
32114 * config/rs6000/rs6000-protos.h: Add prototype for
32115 rs6000_store_data_bypass_p function.
32116 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32117 function implements slightly different (rs6000-specific) semantics
32118 than store_data_bypass_p, returning false rather than aborting
32119 with assertion error when arguments do not satisfy the
32120 requirements of store data bypass.
32121 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32122 rs6000_store_data_bypass_p.
32124 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32126 * config/xtensa/xtensa-protos.h
32127 (xtensa_initial_elimination_offset): New declaration.
32128 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32129 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32130 macro definition, add case for FRAME_POINTER_REGNUM when
32131 FRAME_GROWS_DOWNWARD.
32132 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32133 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32134 xtensa_initial_elimination_offset.
32136 2017-05-08 Nathan Sidwell <nathan@acm.org>
32138 * doc/invoke.texi: Alphabetize -fdump options.
32140 2017-05-08 Martin Sebor <msebor@redhat.com>
32142 PR translation/80280
32143 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32145 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32147 * target.def (compute_frame_layout): New optional target hook.
32148 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32149 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32150 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32152 * reload1.c (verify_initial_elim_offsets): Likewise.
32153 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32154 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32155 (arm_get_frame_offsets): Split up into this ...
32156 (arm_compute_frame_layout): ... and this function.
32158 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32160 * config/aarch64/constraints.md (Usa): New constraint.
32161 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32163 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32165 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32166 with_multilib_list after it has been checked.
32168 2017-05-08 Richard Biener <rguenther@suse.de>
32170 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32171 (bitmap_set_subtract_values): Likewise.
32173 2017-05-08 Richard Biener <rguenther@suse.de>
32175 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32176 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32178 (gimple_stmt_nonzero_warnv_p): Rename to ...
32179 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32181 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32182 (extract_range_basic): Adjust, do not disable propagation on
32183 strict overflow sensitive simplification.
32184 (vrp_visit_cond_stmt): Likewise.
32186 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32188 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32189 body size unconditionally.
32191 2017-05-07 Jeff Law <law@redhat.com>
32194 2017-05-06 Jeff Law <law@redhat.com>
32195 PR tree-optimization/78496
32196 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32199 PR tree-optimization/78496
32200 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32201 (simplify_stmt_using_ranges): Call it.
32202 (vrp_dom_walker::before_dom_children): Extract equivalences
32203 from an ASSERT_EXPR with an equality comparison against a
32206 2017-05-06 Jeff Law <law@redhat.com>
32208 PR tree-optimization/78496
32209 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32212 PR tree-optimization/78496
32213 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32214 (simplify_stmt_using_ranges): Call it.
32215 (vrp_dom_walker::before_dom_children): Extract equivalences
32216 from an ASSERT_EXPR with an equality comparison against a
32219 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32221 * lra-constraints.c (lra_copy_reg_equiv): New function.
32222 (split_reg): Use it to copy equivalence information from the
32223 original register to the spill register.
32225 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32227 PR rtl-optimization/75964
32228 * simplify-rtx.c (simplify_const_relational_operation): Remove
32229 invalid handling of comparisons of integer ABS.
32231 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32233 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32234 initialize to zero.
32235 (init_regs): Remove declaration.
32236 (function_arg_advance_32): Initialize error_p as boolean variable.
32238 2017-05-05 Nathan Sidwell <nathan@acm.org>
32240 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32241 lines. Use for (;;).
32243 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32245 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32246 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32247 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32248 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32249 VF=2 that require versioning.
32251 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32253 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32256 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32258 * diagnostic.h (diagnostic_override_option_index): Convert from
32259 macro to inline function.
32261 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32263 * diagnostic.c (last_module_changed_p): New function.
32264 (set_last_module): New function.
32265 (diagnostic_report_current_module): Convert macro usage to
32266 the above functions.
32267 * diagnostic.h (diagnostic_context::last_module): Strengthen
32268 from const line_map * to const line_map_ordinary *.
32269 (diagnostic_last_module_changed): Delete macro.
32270 (diagnostic_set_last_module): Delete macro.
32272 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32274 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32275 with diagnostic_report_diagnostic.
32276 (diagnostic_n_impl_richloc): Likewise.
32277 * diagnostic.h (report_diagnostic): Delete macro.
32278 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32279 with diagnostic_report_diagnostic.
32280 * substring-locations.c (format_warning_va): Likewise.
32282 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32284 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32285 save/restor of format_spec. Move option-printing code to...
32286 (print_option_information): ...this new function, and
32287 reimplement by simply printing to the pretty_printer,
32288 rather than appending to the format string.
32290 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32292 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32293 handling logic into...
32294 (update_effective_level_from_pragmas): ...this new function.
32296 2017-05-04 Andrew Waterman <andrew@sifive.com>
32298 * config/riscv/riscv.opt (mstrict-align): New option.
32299 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32300 (SLOW_UNALIGNED_ACCESS): Define.
32301 (riscv_slow_unaligned_access): Declare.
32302 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32304 (riscv_slow_unaligned_access): New variable.
32305 (rocket_tune_info): Set slow_unaligned_access to true.
32306 (optimize_size_tune_info): Set slow_unaligned_access to false.
32307 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32308 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32309 (riscv_option_override): Set riscv_slow_unaligned_access.
32310 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32312 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32314 * config/riscv/riscv.md: Unify indentation.
32316 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32321 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32323 (extendsi<mode>2): Add support for doing sign extension via
32324 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32325 don't have ISA 3.0 instructions.
32326 (extendsi<mode>2 splitter): Likewise.
32327 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32328 generate the normal insns since SImode can now go in vector
32329 registers. Disallow the special UNSPECs needed for previous
32330 machines to hide SImode being used. Add new insns
32331 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32332 (fix_trunc<mode>si2_stfiwx): Likewise.
32333 (fix_trunc<mode>si2_internal): Likewise.
32334 (fixuns_trunc<mode>si2): Likewise.
32335 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32336 (fctiw<u>z_<mode>_smallint): Likewise.
32337 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32338 of floating point to 32-bit integer from doing a direct move to
32339 the GPR registers to do a store.
32340 (fctiwz_<mode>): Break long line.
32342 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32344 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32345 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32346 (addr_list, addr_offset_valid_p): New.
32347 (split_address_groups): Check offset validity with above function.
32348 (gt-tree-ssa-loop-ivopts.h): Include header file.
32350 2017-05-05 Nathan Sidwell <nathan@acm.org>
32352 * config.gcc (arm*-*-*): Add missing 'fi'.
32354 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32356 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32357 included in -fopt-info does not matter.
32358 * doc/optinfo.texi (-fopt-info): Fix description of default
32359 behavour. Explicitly say order of options included in -fopt-info
32362 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32364 * config.gcc: Allow combinations of aprofile and rmprofile values for
32365 --with-multilib-list.
32366 * config/arm/t-multilib: New file.
32367 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32368 variables. Remove setting of ISA and floating-point ABI in
32369 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32370 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32371 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32372 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32374 * config/arm/t-rmprofile: Likewise except for the matches changes.
32375 * doc/install.texi (--with-multilib-list): Document the combination of
32376 aprofile and rmprofile values and warn about pitfalls in doing that.
32378 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32380 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32381 (movdi_aarch64): Likewise.
32383 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32385 PR tree-optimization/80632
32386 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32388 (build_arrays): Initialize it for virtual phis.
32389 (fix_phi_nodes): Use it for virtual phis.
32391 PR tree-optimization/80558
32392 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32393 [x, y] op z into [x op, y op z] for op & or | if conditions
32396 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32397 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32400 * config/arm/arm.md (use_literal_pool): Remove.
32401 (64-bit immediate split): No longer takes cost into consideration
32402 if arm_disable_literal_pool is enabled.
32403 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32404 used when arm_disable_literal_pool is enabled.
32405 (arm_max_const_double_inline_cost): Remove use of
32406 arm_disable_literal_pool.
32407 (push_minipool_fix): Add assert.
32408 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32409 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32410 (no_literal_pool_sf_immediate): New.
32412 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32414 PR tree-optimization/80613
32415 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32416 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32418 2017-05-05 Richard Biener <rguenther@suse.de>
32420 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32422 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32424 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32425 of this flag from insn conditions due to removal from r247495.
32427 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32429 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32431 (arm_early_store_addr_dep_ptr): Likewise.
32432 * config/arm/aarch-common-protos.h
32433 (arm_early_load_addr_dep_ptr): Add prototype.
32434 (arm_early_store_addr_dep_ptr): Likewise.
32435 * config/arm/cortex-a53.md: Add new bypasses.
32437 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32439 * tree.c (next_type_uid): Change type to unsigned.
32440 (type_hash_canon): Decrement back next_type_uid if
32441 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32442 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32445 2017-05-04 Martin Sebor <msebor@redhat.com>
32447 * builtins.c: Fix a trivial typo in a comment.
32449 PR middle-end/79234
32450 * builtins.c (check_sizes): Adjust to handle reading past the end.
32451 Avoid printing excessive upper bound of ranges. Use %E to print
32452 tree nodes instead of converting them to %wu.
32453 (expand_builtin_memchr): New function.
32454 (compute_dest_size): Rename...
32455 (compute_objsize): ...to this.
32456 (expand_builtin_memcpy): Adjust.
32457 (expand_builtin_mempcpy): Adjust.
32458 (expand_builtin_strcat): Adjust.
32459 (expand_builtin_strcpy): Adjust.
32460 (check_strncat_sizes): Adjust.
32461 (expand_builtin_strncat): Adjust.
32462 (expand_builtin_strncpy): Adjust and simplify.
32463 (expand_builtin_memset): Adjust.
32464 (expand_builtin_bzero): Adjust.
32465 (expand_builtin_memcmp): Adjust.
32466 (expand_builtin): Handle memcmp.
32467 (maybe_emit_chk_warning): Check strncat just once.
32469 2017-05-04 Martin Sebor <msebor@redhat.com>
32471 PR preprocessor/79214
32472 PR middle-end/79222
32473 PR middle-end/79223
32474 * builtins.c (check_sizes): Add inlining context and issue
32475 warnings even when -Wno-system-headers is set.
32476 (check_strncat_sizes): Same.
32477 (expand_builtin_strncat): Same.
32478 (expand_builtin_memmove): New function.
32479 (expand_builtin_stpncpy): Same.
32480 (expand_builtin): Handle memmove and stpncpy.
32482 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32484 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32485 which is not used any more.
32487 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32489 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32491 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32493 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32494 (cortexa53_tunings): Likewise.
32495 (cortexa57_tunings): Likewise.
32496 (cortexa72_tunings): Likewise.
32497 (cortexa73_tunings): Likewise.
32499 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32501 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32502 Set loop alignment to 8.
32504 2017-05-04 Martin Sebor <msebor@redhat.com>
32506 PR translation/80280
32507 * builtins.c (expand_builtin_object_size): Add missing quoting to
32508 %D and like directives.
32509 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32510 (hsa_type_for_tree_type): Same.
32511 (verify_function_arguments): Same.
32512 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32513 * varasm.c (get_section): Same.
32516 2017-05-04 Martin Sebor <msebor@redhat.com>
32518 PR translation/80280
32519 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32521 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32523 * config/aarch64/aarch64.c (generic_addrcost_table):
32524 Change HI/TI mode setting.
32526 2017-05-04 Martin Jambor <mjambor@suse.cz>
32528 PR tree-optimization/80622
32529 * tree-sra.c (comes_initialized_p): New function.
32530 (build_accesses_from_assign): Only set write lazily when
32531 comes_initialized_p is false.
32532 (analyze_access_subtree): Use comes_initialized_p.
32533 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32534 instead of testing for PARM_DECL.
32536 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32538 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32539 constraint on operand 0 to allow more general addressing modes.
32540 Adjust output template.
32541 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32543 * config/aarch64/aarch64-protos.h
32544 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32545 * config/aarch64/constraints.md (Dp): New address constraint.
32546 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32549 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32551 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32552 update use of estimate_ipcp_clone_size_and_time.
32553 (estimate_local_effects): Update use of
32554 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32555 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32556 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32557 Return nonspecialized time.
32559 2017-05-04 Richard Biener <rguenther@suse.de>
32561 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32562 for the last VUSE which def dominates the PHI. Directly call
32564 (get_continuation_for_phi_1): Remove.
32566 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32568 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32569 to explain the use of truncating division. Cap the number of
32570 iterations to the maximum given by nb_iterations_upper_bound,
32573 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32575 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32576 * configure: Regenerate.
32577 * config.in: Regenerate.
32578 * config/i386/driver-mingw32.c: new file.
32579 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32580 * config.host: Link driver-mingw32.o on MinGW host.
32581 * doc/install.texi: Document new --enable-mingw-wildcard configure
32584 2017-05-04 Marek Polacek <polacek@redhat.com>
32586 PR tree-optimization/80612
32587 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32589 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32590 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32592 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32593 (movt splitter): Likewise.
32594 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32595 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32596 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32597 block for Thumb-1 with MOVT.
32598 (thumb2_legitimate_address_p): Move code block ...
32599 (can_avoid_literal_pool_for_label_p): ... into this new function.
32600 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32602 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32603 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32604 "M-profile targets with the MOVT instruction".
32606 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32608 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32609 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32610 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32612 2017-05-04 Martin Liska <mliska@suse.cz>
32614 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32615 variable cond_code.
32617 2017-05-04 Richard Biener <rguenther@suse.de>
32619 * tree.c (array_at_struct_end_p): Handle arrays at struct
32620 end with flexarrays more conservatively. Refactor and treat
32621 arrays of arrays or aggregates more strict. Fix
32622 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32623 * tree.c (array_at_struct_end_p): Adjust prototype.
32624 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32625 * gimple-fold.c (get_range_strlen): Likewise.
32626 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32628 2017-05-04 Richard Biener <rguenther@suse.de>
32630 PR tree-optimization/31130
32631 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32633 (supports_overflow_infinity): Likewise.
32634 (is_negative_overflow_infinity): Likewise.
32635 (is_positive_overflow_infinity): Likewise.
32636 (is_overflow_infinity): Likewise.
32637 (stmt_overflow_infinity): Likewise.
32638 (overflow_infinity_range_p): Likewise.
32639 (usable_range_p): Remove as always returning true.
32640 (make_overflow_infinity): Remove.
32641 (negative_overflow_infinity): Likewise.
32642 (positive_overflow_infinity): Likewise.
32643 (avoid_overflow_infinity): Likewise.
32644 (set_value_range): Adjust accordingly.
32645 (set_value_range_to_nonnegative): Likewise, remove now unused
32646 overflow_infinity arg.
32647 (vrp_operand_equal_p): Adjust.
32648 (update_value_range): Likewise.
32649 (range_int_cst_singleton_p): Likewise.
32650 (operand_less_p): Likewise.
32651 (compare_values_warnv): Likewise.
32652 (extract_range_for_var_from_comparison_expr): Likewise.
32653 (vrp_int_const_binop): Likewise.
32654 (zero_nonzero_bits_from_vr): Likewise.
32655 (extract_range_from_multiplicative_op_1): Likewise.
32656 (extract_range_from_binary_expr_1): Likewise.
32657 (extract_range_from_unary_expr): Likewise.
32658 (extract_range_from_comparison): Likewise.
32659 (extract_range_basic): Likewise.
32660 (adjust_range_with_scev): Likewise.
32661 (compare_ranges): Likewise.
32662 (compare_range_with_value): Likewise.
32663 (dump_value_range): Likewise.
32664 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32666 (simplify_cond_using_ranges): Adjust.
32668 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32670 * brig-builtins.def: Added a builtin for class_f64.
32671 * builtin-types.def: Added a builtin type needed by class_f64.
32673 2017-05-03 Jason Merrill <jason@redhat.com>
32675 * timevar.def: Add TV_CONSTEXPR.
32677 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32679 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32681 2017-05-03 Martin Jambor <mjambor@suse.cz>
32683 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32684 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32685 * ipa-cp.c (ipcp_propagate_stage): Do not call
32686 ipa_update_after_lto_read.
32687 * ipa-inline.c (ipa_inline): Likewise.
32689 2017-05-03 Martin Jambor <mjambor@suse.cz>
32691 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32692 tag. Added a default constructor and a destructor.
32693 (ipa_edge_args_sum_t): New class;
32694 (ipa_edge_args_sum): Declare.
32695 (ipa_edge_args_vector): Remove declaration.
32696 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32697 (ipa_free_edge_args_substructures): Remove declaration.
32698 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32699 (ipa_edge_args_info_available_for_edge_p): Likewise.
32700 * ipa-prop.c (ipa_edge_args_vector): Removed.
32701 (edge_removal_hook_holder): Likewise.
32702 (edge_duplication_hook_holder): Likewise.
32703 (ipa_edge_args_sum): New variable.
32704 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32705 ipa_edge_args_vector.
32706 (ipa_free_edge_args_substructures): Likewise.
32707 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32708 ipa_edge_args_vector.
32709 (ipa_edge_removal_hook): Turned into method
32710 ipa_edge_args_sum_t::remove.
32711 (ipa_edge_duplication_hook): Turned into method
32712 ipa_edge_args_sum_t::duplicate.
32713 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
32714 registering edge hooks.
32715 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
32716 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
32717 ipa_edge_args_sum instead of ipa_edge_args_vector.
32718 * ipa-profile.c (ipa_profile): Likewise.
32720 2017-05-03 Martin Jambor <mjambor@suse.cz>
32722 * symbol-summary.h (function_summary): New method exists.
32723 (function_summary::symtab_removal): Deallocate through release.
32724 (call_summary): New class.
32725 (gt_ggc_mx): New overload.
32726 (gt_pch_nx): Likewise.
32727 (gt_pch_nx): Likewise.
32729 2017-05-03 Jeff Law <law@redhat.com>
32731 PR tree-optimization/78496
32732 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
32733 from simplify_cond_using_ranges. Split off code to walk
32734 backwards through casts into ...
32735 (simplify_cond_using_ranges_2): New function.
32736 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
32737 (execute_vrp): After identifying jump threads, call
32738 simplify_cond_using_ranges_2.
32740 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32743 * ipa-inline.h (inline_summary): Add ctor.
32744 (create_ggc): Do not use ggc_cleared_alloc.
32746 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32747 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32749 * gcc.c (handle_braces): Support escaping in switch matching
32751 * doc/invoke.texi (Spec Files): Document it.
32752 Remove superfluous @code markup in items.
32754 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32756 * diagnostic-show-locus.c (struct column_range): New struct.
32757 (get_affected_columns): New function.
32758 (get_printed_columns): New function.
32759 (struct correction): New struct.
32760 (correction::ensure_capacity): New function.
32761 (correction::ensure_terminated): New function.
32762 (struct line_corrections): New struct.
32763 (line_corrections::~line_corrections): New dtor.
32764 (line_corrections::add_hint): New function.
32765 (layout::print_trailing_fixits): Reimplement in terms of the new
32767 (selftest::test_overlapped_fixit_printing): New function.
32768 (selftest::diagnostic_show_locus_c_tests): Call it.
32770 2017-05-03 Nathan Sidwell <nathan@acm.org>
32772 Canonicalize canonical type hashing
32773 * tree.h (type_hash_canon_hash): Declare.
32774 * tree.c (type_hash_list, attribute_hash_list): Move into
32775 type_hash_canon_hash.
32776 (build_type_attribute_qual_variant): Break out hash code calc into
32777 type_hash_canon_hash.
32778 (type_hash_canon_hash): New. Generic type hash computation.
32779 (build_range_type_1, build_array_type_1, build_function_type,
32780 build_method_type_directly, build_offset_type, build_complex_type,
32781 make_vector_type): Call it.
32783 2017-05-03 Richard Biener <rguenther@suse.de>
32785 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
32786 When all DRs have unknown misaligned do not always peel
32787 when there is a store but apply the same costing model as if
32788 there were only loads.
32790 2017-05-03 Richard Biener <rguenther@suse.de>
32793 PR tree-optimization/80492
32794 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32795 compare_base_decls returning dont-know properly.
32797 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32799 * config/arm/iterators.md (CCSI): New mode iterator.
32800 (arch): New mode attribute.
32801 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32802 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32803 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32804 code iterator for success result mode.
32805 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32806 the corresponding new insn generators.
32808 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32811 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32812 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32814 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32816 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32817 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32818 (DDR_A): Wrap DDR argument in brackets.
32819 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32820 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32821 (DDR_REVERSED_P): Likewise.
32823 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32825 PR tree-optimization/79472
32826 * tree-switch-conversion.c (struct switch_conv_info): Add
32827 contiguous_range and default_case_nonstandard fields.
32828 (collect_switch_conv_info): Compute contiguous_range and
32829 default_case_nonstandard fields, don't clear final_bb if
32830 contiguous_range and only the default case doesn't have the required
32832 (check_all_empty_except_final): Set default_case_nonstandard instead
32833 of failing if contiguous_range and the default case doesn't have empty
32835 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32836 and only the default case doesn't have the required constants. Skip
32838 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32839 if default_case_nonstandard.
32840 (build_constructors): Build constant 1 just once. Assert that default
32841 values aren't inserted in between cases if contiguous_range. Skip
32843 (build_arrays): Skip virtual phis.
32844 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32845 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32846 Handle virtual phis.
32847 (gen_inbound_check): Handle default_case_nonstandard case.
32848 (process_switch): Adjust check_final_bb caller. Call
32849 gather_default_values with the first non-default case instead of
32850 default case if default_case_nonstandard.
32852 2017-05-02 Nathan Sidwell <nathan@acm.org>
32854 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32855 check. Fix formatting.
32857 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32859 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32860 errors when comparing specialized and unspecialized times.
32862 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32864 * diagnostic-show-locus.c
32865 (layout::should_print_annotation_line_p): Make private.
32866 (layout::print_annotation_line): Make private.
32867 (layout::annotation_line_showed_range_p): Make private.
32868 (layout::show_ruler): Make private.
32869 (layout::print_source_line): Make private. Pass in line and
32870 line_width, rather than calling location_get_source_line. Drop
32872 (layout::print_leading_fixits): New method.
32873 (layout::print_any_fixits): Rename to...
32874 (layout::print_trailing_fixits): ...this, and make private.
32875 Don't print newline fixits.
32876 (diagnostic_show_locus): Move logic for printing one row into...
32877 (layout::print_line): ...this new function. Move the
32878 location_get_source_line call and error-handling from
32879 print_source_line to here. Call print_leading_fixits, and rename
32880 print_any_fixits to print_trailing_fixits.
32881 (selftest::test_fixit_insert_containing_newline): Update now that
32882 newlines are partially supported.
32883 (selftest::test_fixit_insert_containing_newline_2): New test.
32884 (selftest::test_fixit_replace_containing_newline): Update comments.
32885 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32886 * edit-context.c (class added_line): New class.
32887 (class edited_line): Describe newline handling in comment.
32888 (edited_line::actually_edited_p): New method.
32889 (edited_line::print_content): Delete redundant decl.
32890 (edited_line::m_predecessors): New field.
32891 (edited_file::print_content): Call edited_line::print_content.
32892 (edited_file::print_diff): Update to support newlines.
32893 (edited_file::print_diff_hunk): Likewise.
32894 (edited_file::print_run_of_changed_lines): New function.
32895 (edited_file::print_diff_line): Convert to...
32896 (print_diff_line): ...this.
32897 (edited_file::get_effective_line_count): New function.
32898 (edited_line::edited_line): Initialize new field m_predecessors.
32899 (edited_line::~edited_line): Clean up m_predecessors.
32900 (edited_line::apply_fixit): Handle newlines.
32901 (edited_line::get_effective_line_count): New function.
32902 (edited_line::print_content): New function.
32903 (edited_line::print_diff_lines): New function.
32904 (selftest::test_applying_fixits_insert_containing_newline): New
32906 (selftest::test_applying_fixits_replace_containing_newline): New
32908 (selftest::insert_line): New function.
32909 (selftest::test_applying_fixits_multiple_lines): Add example of
32911 (selftest::edit_context_c_tests): Call the new tests.
32913 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32915 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32916 parameter cand. Update dump information.
32917 (get_computation_cost): Update uses.
32919 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32921 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32922 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32923 (get_computation_at, rewrite_use_address): Update use of
32924 get_computation_aff.
32926 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32928 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32929 (get_computation): Delete.
32930 (get_computation_cost): Implement like get_computation_cost_at.
32931 Use get_computation_at.
32932 (get_computation_cost_at): Delete.
32933 (rewrite_use_nonlinear_expr): Use get_computation_at.
32934 (rewrite_use_compare, remove_unused_ivs): Ditto.
32936 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32938 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32940 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32942 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32943 (ivopts_global_cost_for_size): Rename parameter and update uses.
32944 (iv_ca_recount_cost): Update uses.
32945 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32946 candidates seperately in n_invs and n_cands.
32947 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32949 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32951 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32952 (find_inv_vars_cb): New.
32953 (find_depends): Renamed to ...
32954 (find_inv_vars): ... this.
32955 (add_candidate_1, force_var_cost): Call find_inv_vars.
32956 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32958 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32960 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32961 inv_vars. Add inv_exprs.
32962 (struct iv_cand): Rename depends_on to inv_vars.
32963 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32964 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32965 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32966 (dump_cand): Dump inv_vars.
32967 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32968 (record_invariant, find_depends, add_candidate_1): Ditto.
32969 (set_group_iv_cost, force_var_cost): Ditto.
32970 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32971 (get_computation_cost_at, get_computation_cost): Ditto.
32972 (determine_group_iv_cost_generic): Ditto.
32973 (determine_group_iv_cost_address): Ditto.
32974 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32975 (determine_group_iv_costs): Ditto.
32976 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32977 (iv_ca_set_remove_invariants): Renamed to ...
32978 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32979 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32980 (iv_ca_set_add_invariants): Renamed to ...
32981 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32982 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32983 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32984 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32985 (create_new_ivs): Remove useless dump.
32987 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32989 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32991 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32992 (iv_ca_set_no_cp, create_new_iv): Ditto.
32994 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32996 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32998 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33000 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33001 function tree_check2.
33003 2017-05-02 Martin Liska <mliska@suse.cz>
33005 * doc/gcov.texi: Add missing preposition.
33006 * gcov.c (function_info::function_info): Properly fill up
33007 all member variables.
33009 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33011 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33013 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33015 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33017 2017-05-02 Martin Liska <mliska@suse.cz>
33020 * lto-streamer-in.c (lto_read_tree_1): Remove
33021 LTO_STREAMER_DEBUG.
33022 * lto-streamer.c (struct tree_hash_entry): Likewise.
33023 (struct tree_entry_hasher): Likewise.
33024 (tree_entry_hasher::hash): Likewise.
33025 (tree_entry_hasher::equal): Likewise.
33026 (lto_streamer_init): Likewise.
33027 (lto_orig_address_map): Likewise.
33028 (lto_orig_address_get): Likewise.
33029 (lto_orig_address_remove): Likewise.
33030 * lto-streamer.h: Likewise.
33031 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33032 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33034 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33036 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33037 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33038 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33039 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33040 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33041 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33042 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33043 (mm_maskz_sub_ss): New intrinsics.
33044 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33045 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33046 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33047 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33048 (__builtin_ia32_subss_mask_round): New builtins.
33049 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33050 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33051 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33053 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33054 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33056 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33059 2017-05-02 Martin Jambor <mjambor@suse.cz>
33061 PR tree-optimization/78687
33062 * tree-sra.c (access): New field parent.
33063 (process_subtree_disqualification): New function.
33064 (disqualify_candidate): Call it.
33065 (build_accesses_from_assign): Reset write flag if creating an
33067 (build_access_subtree): Fill in parent field and also prpagate
33068 down grp_write flag.
33069 (create_artificial_child_access): New parameter set_grp_write, set
33070 grp_write to its value.
33071 (propagate_subaccesses_across_link): Also propagate grp_write flag
33073 (propagate_all_subaccesses): Push the closest parent back to work
33074 queue if add_access_to_work_queue returned true.
33076 2017-05-02 Richard Biener <rguenther@suse.de>
33078 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33079 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33080 -fstrict-overflow documentation.
33081 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33082 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33083 flag_strict_overflow.
33084 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33085 * lto-opts.c (lto_write_options): Do not stream it.
33086 * lto-wrapper.c (merge_and_complain): Do not handle it.
33087 * opts.c (default_options_table): Do not set -fstrict-overflow.
33088 (finish_options): Likewise do not clear it when sanitizing.
33089 * simplify-rtx.c (simplify_const_relational_operation): Do not
33090 test flag_strict_overflow.
33092 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33094 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33095 using enabled attribute.
33096 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33097 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33098 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33099 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33100 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33101 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33102 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33103 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33104 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33105 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33107 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33109 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33111 2017-05-02 Richard Biener <rguenther@suse.de>
33113 PR tree-optimization/80591
33115 2017-04-10 Richard Biener <rguenther@suse.de>
33117 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33120 2017-05-02 Richard Biener <rguenther@suse.de>
33122 PR tree-optimization/80549
33123 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33124 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33125 headers if they do not have a preheader.
33127 2017-05-02 Martin Liska <mliska@suse.cz>
33130 * common.opt: Fix typo.
33131 * doc/invoke.texi: Likewise.
33133 2017-05-01 Jan Beulich <jbeulich@suse.com>
33135 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33136 swapping, add (x,x,m,x,n) alternative.
33138 2017-05-01 Nathan Sidwell <nathan@acm.org>
33140 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33141 unnecessary unadjusted_alignment check.
33143 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33146 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33148 * gimplify.c (gimplify_cilk_detach): New function.
33149 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33150 * tree-core.h: Document EXPR_CILK_SPAWN.
33151 * tree.h (EXPR_CILK_SPAWN): Define.
33153 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33155 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33156 to use new fixit_hint representation, using the "replace" logic.
33157 (get_line_span_for_fixit_hint): Likewise.
33158 (layout::print_any_fixits): Likewise.
33159 (selftest::test_one_liner_many_fixits): Rename to...
33160 (selftest::test_one_liner_many_fixits_1): ...this, and update
33161 comment and expected output to reflect that the multiple fix-it
33162 hints are now consolidated into one insertion.
33163 (selftest::test_one_liner_many_fixits_2): New test.
33164 (selftest::test_diagnostic_show_locus_one_liner): Update for
33166 (selftest::test_fixit_consolidation): Update for fix-it API
33168 * diagnostic.c (print_parseable_fixits): Likewise.
33169 * edit-context.c (edited_line::m_line_events): Convert from
33170 auto_vec <line_event *> to auto_vec <line_event>.
33171 (class line_event): Convert from abstract base class to a concrete
33172 class, taking over the role of replace_event.
33173 (class insert_event): Delete.
33174 (class replace_event): Rename to class line_event. Convert to
33176 (edit_context::add_fixits): Reimplement.
33177 (edit_context::apply_insert): Delete.
33178 (edit_context::apply_replace): Rename to...
33179 (edit_context::apply_fixit): ...this. Convert to half-open range.
33180 (edited_file::apply_insert): Delete.
33181 (edited_file::apply_replace): Rename to...
33182 (edited_file::apply_fixit): ...this.
33183 (edited_line::~edited_line): Drop deletion of events.
33184 (edited_line::apply_insert): Delete.
33185 (edited_line::apply_replace): Rename to...
33186 (edited_line::apply_fixit): ...this. Convert to half-open range.
33187 Update for change to type of m_line_events.
33188 * edit-context.h (edit_context::apply_insert): Delete.
33189 (edit_context::apply_replace): Rename to...
33190 (edit_context::apply_fixit): ...this.
33192 2017-05-01 Martin Sebor <msebor@redhat.com>
33194 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33197 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33200 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33201 __get_cpuid_max returns 0.
33202 (__get_cpuid_count): Ditto.
33204 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33206 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33207 replacement expression is another instance of one of its arguments.
33209 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33212 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33213 check for stack push/pop autoinc.
33214 * config/i386/i386.c (ix86_agi_dependent): Return false
33215 if the only reason why modified_in_p returned true is that
33216 addr is SP based and set_insn is a push or pop.
33218 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33220 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33223 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33226 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33227 (account_size_time): Use two predicates - exec_pred and
33229 (evaluate_conditions_for_known_args): Compute both clause and
33231 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33232 (inline_summary_t::duplicate): Update.
33233 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33235 (compute_inline_parameters): Likewise.
33236 (estimate_edge_size_and_time): Update caluclation of time.
33237 (estimate_node_size_and_time): Compute both time and nonspecialized
33239 (estimate_ipcp_clone_size_and_time): Update.
33240 (inline_merge_summary): Update.
33241 (do_estimate_edge_time): Update.
33242 (do_estimate_edge_size): Update.
33243 (do_estimate_edge_hints): Update.
33244 (inline_read_section, inline_write_summary): Stream both new predicates.
33245 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33247 (compute_inlined_call_time): Cleanup.
33248 (big_speedup_p): Update.
33249 (edge_badness): Update.
33250 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33251 (size_time_entry): Replace predicate by exec_predicate and
33252 nonconst_predicate.
33253 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33254 (estimate_edge_time): Return also nonspec_time.
33255 (reset_edge_growth_cache): Update.
33257 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33259 PR rtl-optimization/80491
33260 * ifcvt.c (noce_process_if_block): When looking for x setter
33261 with missing else_bb, don't check only the insn right before
33262 cond_earliest, but look for the last insn that x is modified in
33263 within the same bb.
33265 PR rtl-optimization/80491
33266 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33268 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33270 PR tree-optimization/80487
33271 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33273 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33275 PR tree-optimization/79697
33276 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33277 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33278 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33280 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33281 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33283 2017-04-28 Martin Sebor <msebor@redhat.com>
33285 PR tree-optimization/80523
33286 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33287 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33289 (maybe_warn, format_directive, parse_directive): Use new functions.
33290 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33292 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33294 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33296 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33298 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33299 target_header_dir): Set correctly.
33300 * configure: Regenerated.
33301 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33302 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33303 instead of SYSTEM_HEADER_DIR.
33305 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33307 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33308 (estimate_local_effects): Likewise.
33309 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33310 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33311 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33312 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33313 do_estimate_edge_time, estimate_edge_time): Likewise.
33314 * ipa-inline-analysis.c (estimate_node_size_and_time,
33315 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33316 (estimate_time_after_inlining): Remove.
33318 2017-04-28 Martin Liska <mliska@suse.cz>
33320 * doc/gcov.texi: Enhance documentation of gcov.
33322 2017-04-28 Martin Liska <mliska@suse.cz>
33324 * doc/gcov.texi: Sort options in alphabetic order.
33325 * doc/gcov-dump.texi: Likewise.
33326 * doc/gcov-tool.texi: Likewise.
33327 * gcov.c (print_usage): Likewise.
33328 * gcov-dump.c (print_usage): Likewise.
33329 * gcov-tool.c (print_merge_usage_message): Likewise.
33330 (print_rewrite_usage_message): Likewise.
33331 (print_overlap_usage_message): Likewise.
33333 2017-04-28 Martin Liska <mliska@suse.cz>
33335 PR gcov-profile/53915
33336 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33338 2017-04-28 Martin Liska <mliska@suse.cz>
33340 PR gcov-profile/79891
33341 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33342 is marked by compiler as living on a line.
33343 (get_cycles_count): Remove usage of the union.
33344 (output_intermediate_file): Likewise.
33345 (find_source): Fix GNU coding style.
33346 (accumulate_line_counts): Remove old non-all block mode.
33347 (output_lines): Remove usage of the union.
33348 * profile.c (output_location): Include all BBs, even if
33349 belonging to a same line (and file) as a previous BB.
33351 2017-04-28 Martin Liska <mliska@suse.cz>
33353 * gcov.c (process_args): Handle new argument 'w'.
33354 (read_graph_file): Assign ID to BBs.
33355 (output_branch_count): Display BB # if verbose flag is set.
33356 (output_lines): Likewise for arcs.
33357 (print_usage): Add '--verbose' option help.
33358 * doc/gcov.texi: Document --verbose (-w) option.
33360 2017-04-28 Martin Liska <mliska@suse.cz>
33362 * gcov.c (struct block_location_info): New struct.
33363 (process_file): Fill up the new structure.
33364 (read_graph_file): Replace usage of encoding by the newly added
33366 (add_line_counts): Likewise.
33367 (accumulate_line_counts): Remove usage of the union.
33368 (function_info::function_info): New function.
33369 (function_info::~function_info): Likewise.
33370 (process_file): Call delete instead of release_function.
33371 (release_function): Release the function.
33372 (release_structures): Call delete instead of release_function.
33373 (solve_flow_graph): Replace usage of num_blocks.
33374 (find_exception_blocks): Likewise.
33375 (output_lines): Fix GNU coding style.
33377 2017-04-28 Martin Liska <mliska@suse.cz>
33380 * coverage.c (coverage_remove_note_file): New function.
33381 * coverage.h: Declare the function.
33382 * toplev.c (finalize): Clean if an error has been seen.
33384 2017-04-28 Martin Liska <mliska@suse.cz>
33386 PR gcov-profile/80031
33387 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33388 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33389 * gcov.c (read_graph_file): Read just number of blocks.
33390 * profile.c (branch_prob): Do not stream 0 flags per a basic
33393 2017-04-28 Martin Liska <mliska@suse.cz>
33395 * gcov-dump.c (tag_*): Add new argument to declarations.
33396 (dump_gcov_file): Likewise.
33397 (tag_blocks): Add and use new argument depth.
33398 (tag_arcs): Likewise.
33399 (tag_lines): Likewise.
33400 (tag_counters): Likewise.
33401 (tag_summary): Likewise.
33402 (dump_working_sets): Use depth to do a proper indentation.
33404 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33407 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33408 * symtab.c (symtab_node::debug_symtab): Move definition here.
33410 2017-04-28 Richard Biener <rguenther@suse.de>
33412 * lto-streamer.h (LTO_major_version): Bump to 7.
33414 2017-04-28 Richard Biener <rguenther@suse.de>
33416 * tree-vrp.c (assert_info): New struct.
33417 (add_assert_info): New helper.
33418 (register_edge_assert_for_2): Refactor to add asserts to a vector
33420 (register_edge_assert_for_1): Likewise.
33421 (register_edge_assert_for): Likewise.
33422 (finish_register_edge_assert_for): New helper actually registering
33423 asserts where live on edge.
33424 (find_conditional_asserts): Adjust.
33425 (find_switch_asserts): Likewise.
33426 (evrp_dom_walker::try_find_new_range): Generalize.
33427 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33429 2017-04-27 Marek Polacek <polacek@redhat.com>
33432 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33433 arg10 and arg11 to itype.
33435 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33437 * doc/extend.texi (Object Size Checking): Improve grammar.
33439 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33442 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33443 that the logic for permitting reciprocal estimates matches that
33446 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33449 * tree.c (type_cache_hasher::equal): Only compare
33450 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33451 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33452 non-aggregate element types.
33453 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33454 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33456 2017-04-27 Richard Biener <rguenther@suse.de>
33458 PR middle-end/80533
33459 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33460 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33461 keeping a reference to a trailing array.
33463 2017-04-27 Richard Biener <rguenther@suse.de>
33465 PR middle-end/80539
33466 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33467 being in loop-closed SSA form conservatively.
33468 (chrec_fold_multiply_poly_poly): Likewise.
33470 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33472 PR middle-end/79665
33473 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33474 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33476 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33479 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33480 (aarch64_function_arg_alignment): Return unsigned int again, but still
33481 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33482 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33483 Don't emit -Wpsabi note.
33484 (aarch64_function_arg_boundary): Likewise.
33485 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33488 2017-04-26 Nathan Sidwell <nathan@acm.org>
33490 * tree.h (crc32_unsigned_n): Declare.
33491 (crc32_unsigned, crc32_unsigned): Make inline.
33492 * tree.c (crc32_unsigned_bits): Replace with ...
33493 (crc32_unsigned_n): ... this.
33494 (crc32_unsigned, crc32_byte): Remove.
33495 (crc32_string): Remove unnecessary braces.
33497 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33499 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33500 * ipa-inline-analysis.c (MAX_TIME): Remove.
33501 (account_size_time): Use sreal for time.
33502 (dump_inline_summary): Update.
33503 (estimate_function_body_sizes): Update.
33504 (estimate_edge_size_and_time): Update.
33505 (estimate_calls_size_and_time): Update.
33506 (estimate_node_size_and_time): Update.
33507 (inline_merge_summary): Update.
33508 (inline_update_overall_summary): Update.
33509 (estimate_time_after_inlining): Update.
33510 (inline_read_section): Update.
33511 (inline_write_summary): Update.
33512 * ipa-inline.c (compute_uninlined_call_time): Update.
33513 (compute_inlined_call_time): Update.
33514 (recursive_inlining): Update.
33515 (inline_small_functions): Update.
33516 (dump_overall_stats): Update.
33517 * ipa-inline.h: Include sreal.h.
33518 (size_time_entry): Turn time to sreal.
33519 (inline_summary): Turn self_time nad time to sreal.
33521 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33523 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33525 (sreal::stream_out, sreal::stream_in): New.
33526 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33528 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33530 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33533 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33536 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33538 (dimode_scalar_chain::compute_convert_gain): Ditto.
33539 (dimode_scalar_chain::make_vector_copies): Ditto.
33540 (dimode_scalar_chain::convert_reg): Ditto.
33541 (dimode_scalar_chain::convert_insn): Ditto.
33542 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33543 (VI248_AVX512BW_1): New mode iterator.
33544 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33545 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33548 2017-04-25 Martin Sebor <msebor@redhat.com>
33550 PR tree-optimization/80497
33551 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33552 constants are representable in HOST_WIDE_INT.
33553 (parse_directive): Ditto.
33555 2017-04-25 Martin Sebor <msebor@redhat.com>
33558 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33559 (new_zero_array): Adjust signature.
33560 (dom_info::dom_init): Used unsigned rather that size_t.
33561 (dom_info::dom_info): Same.
33563 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33564 Jakub Jelinek <jakub@redhat.com>
33567 * config/arm/arm.c: Include gimple.h.
33568 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33569 returns negative, increment ncrn only if it returned positive.
33570 (arm_needs_doubleword_align): Return int instead of bool,
33571 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33572 members, but if there is any such non-FIELD_DECL
33573 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33574 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33575 returns negative, increment nregs only if it returned positive.
33576 (arm_setup_incoming_varargs): Likewise.
33577 (arm_function_arg_boundary): Emit -Wpsabi note if
33578 arm_needs_doubleword_align returns negative, return
33579 DOUBLEWORD_ALIGNMENT only if it returned positive.
33581 2017-04-25 Marek Polacek <polacek@redhat.com>
33584 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33585 first argument to type.
33587 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33590 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33591 type checks to test for compatibility instead of equality.
33593 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33594 Jakub Jelinek <jakub@redhat.com>
33597 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33599 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33600 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33601 the alignment computation, but return their maximum in warn_alignment.
33602 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33603 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33605 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33606 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33609 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33611 * config/arc/simdext.md (dmpyh): Fix typo.
33613 2017-04-25 Richard Biener <rguenther@suse.de>
33615 PR tree-optimization/80492
33616 * alias.c (compare_base_decls): Handle registers with asm
33617 specification conservatively.
33618 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33619 compare_base_decls returning dont-know properly.
33621 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33623 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33624 (legitimate_offset_address_p): New function.
33625 (arc_legitimate_address_p): Use above function.
33627 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33629 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33631 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33633 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33634 ACCH registers whenever they are available.
33636 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33638 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33639 double regs fix when not used.
33641 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33643 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33645 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33646 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33648 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33650 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33651 when emitting short ADD instructions.
33653 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33655 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33657 (cmpsi_cc_c_insn): Likewise.
33658 (cbranchsi4_scratch): Compute proper instruction length using
33659 compact_hreg_operand.
33660 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33662 2017-04-25 Richard Biener <rguenther@suse.de>
33664 PR middle-end/80509
33665 * passes.c (pass_manager::pass_manager): Initialize
33666 m_name_to_pass_map.
33668 2017-04-25 Richard Biener <rguenther@suse.de>
33670 PR tree-optimization/79201
33671 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33673 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33676 * config/s390/vector.md: Split MEM->GPR vector moves for
33677 non-s_operand addresses.
33679 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33682 * config/s390/predicates.md (reload_const_wide_int_operand): New
33684 * config/s390/s390.md ("movti"): Remove d/P alternative.
33685 ("movti_bigconst"): New pattern definition.
33687 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33690 * s390-protos.h (s390_expand_cs_hqi): Removed.
33691 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33692 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33693 modes as well as CCZ1mode and CCZmode.
33694 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33695 signature of s390_emit_compare_and_swap.
33696 (s390_expand_cs_hqi): Likewise, make static.
33697 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33698 compare-and-swap, in some cases.
33699 (s390_expand_cs): Wrapper function.
33700 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33702 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33703 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33704 patterns for small and large integers. Forbid symref memory operands.
33705 Move expander to s390.c. Require cc register.
33706 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33707 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33708 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33709 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33710 symref memory operands. Remove CC mode and call s390_match_ccmode
33712 ("atomic_exchange<mode>"): Allow and implement all integer modes.
33714 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33716 * config/s390/s390.md (define_peephole2): New peephole to help
33717 combining the load-and-test pattern with volatile memory.
33719 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33721 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
33722 with CCZmode for TARGET_Z196.
33724 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33726 PR rtl-optimization/80501
33727 * combine.c (make_compound_operation_int): Set subreg_code to SET
33728 even for AND with mask of the sign bit of mode.
33730 PR rtl-optimization/80500
33731 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
33732 sum's initial value.
33734 2017-04-25 Julian Brown <julian@codesourcery.com>
33735 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33737 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
33739 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33741 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33743 2017-04-25 Julian Brown <julian@codesourcery.com>
33744 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33746 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33747 (thunderx2t99_sha): New Reservation.
33749 2017-04-25 Julian Brown <julian@codesourcery.com>
33750 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33752 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33753 type for 1-element load.
33755 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33757 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33759 2017-04-24 Martin Jambor <mjambor@suse.cz>
33761 PR tree-optimization/80293
33762 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33763 char arrays not totally scalarizable if it is false.
33764 (analyze_all_variable_accesses): Pass correct value in the new
33765 parameter. Add a statistics counter.
33767 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
33769 PR middle-end/79931
33770 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
33772 2017-04-24 Richard Biener <rguenther@suse.de>
33774 PR tree-optimization/80494
33775 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33776 out for complex types.
33778 2017-04-24 Richard Biener <rguenther@suse.de>
33780 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
33781 * tree-ssa-sccvn.c (print_scc): Print SCC size.
33782 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
33783 (DFS): Adjust and never fail.
33784 (sccvn_dom_walker::fail): Remove.
33785 (sccvn_dom_walker::before_dom_children): Adjust.
33786 (run_scc_vn): Likewise and never fail.
33787 * tree-ssa-pre.c (pass_pre::execute): Adjust.
33788 (pass_fre::execute): Likewise.
33790 2017-04-24 Richard Biener <rguenther@suse.de>
33792 PR tree-optimization/79725
33793 * tree-ssa-sink.c (statement_sink_location): Return whether
33794 failure reason was zero uses. Move that check later.
33795 (sink_code_in_bb): Deal with zero uses by removing the stmt
33798 2017-04-24 Richard Biener <rguenther@suse.de>
33801 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33802 pointer-based references.
33804 2017-04-24 Richard Biener <rguenther@suse.de>
33807 * pass_manager.h (pass_manager::operator new): Remove.
33808 (pass_manager::operator delete): Likewise.
33809 * passes.c (pass_manager::operator new): Remove.
33810 (pass_manager::operator delete): Likewise.
33811 (pass_manager::pass_manager): Zero individual pass members.
33813 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33816 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33817 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33818 Check "XEXP (src, 1)" operand here.
33819 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33820 Check "XEXP (src, 1)" operand here.
33821 (dimode_scalar_chain::make_vector_copies): Detect count register
33822 of a shift instruction. Zero extend count register from QImode
33823 to DImode to satisfy vector shift pattern count operand predicate.
33824 Substitute vector shift count operand with a DImode copy.
33825 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33828 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33830 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33831 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33832 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33833 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33834 (UNSPEC_NOREX_MEM): Remove definition.
33836 2017-04-21 Richard Biener <rguenther@suse.de>
33838 PR tree-optimization/79547
33839 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33840 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33841 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33842 without any constraints.
33844 2017-04-21 Richard Biener <rguenther@suse.de>
33846 PR tree-optimization/78847
33847 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33849 2017-04-21 Richard Biener <rguenther@suse.de>
33851 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33852 (build_distinct_type_copy): Likewise.
33853 (build_variant_type_copy): Likewise.
33854 * tree.c (build_qualified_type): Pass down mem-stat info.
33855 (build_distinct_type_copy): Likewise.
33856 (build_variant_type_copy): Likewise.
33858 2017-04-21 Richard Biener <rguenther@suse.de>
33860 PR tree-optimization/80237
33861 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33863 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33864 for a simplified result.
33866 2016-04-21 Richard Biener <rguenther@suse.de>
33868 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33869 sth as strict as a simple_iv but a chrec without symbols and an
33870 operand defined in the loop we are peeling (and not some subloop).
33871 (propagate_constants_for_unrolling): Propagate all constants.
33873 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33876 * config/i386/i386.c (print_reg): Remove assert for disalowed
33877 regno values, call output_operand_lossage instead.
33879 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33882 * config/i386/constraints.md (Yc): New register constraint.
33883 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33884 Use Yc constraint for alternative 2 of operand 0. Remove
33885 preferred_for_speed attribute.
33887 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33889 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33890 lastprivate clauses in SIMT case.
33892 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33894 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33896 2017-04-20 Richard Biener <rguenther@suse.de>
33898 PR tree-optimization/57796
33899 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33900 as N scalar stores.
33901 (vect_model_load_cost): Cost gathers as N scalar loads.
33903 2017-04-20 Richard Biener <rguenther@suse.de>
33905 * ggc-page.c (ggc_allocated_p): Rename to ...
33906 (safe_lookup_page_table_entry): ... this and return the lookup
33908 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33910 2017-04-20 Richard Biener <rguenther@suse.de>
33912 PR tree-optimization/80453
33913 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33914 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33915 from the conditions.
33916 (vn_phi_eq): Pass them down.
33917 (vn_phi_lookup): Record them.
33918 (vn_phi_insert): Likewise.
33920 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33922 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33923 uninitialized variable warning to avoid buffer overrun.
33925 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33928 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33929 is suppressed for '{ 0 }' in C.
33931 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33933 * BASE-VER: Set to 8.0.0.
33935 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33937 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33938 priority .init_array and .fini_array section with SECTION_NOTYPE
33941 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33943 PR middle-end/80423
33944 * tree.h (build_array_type): Add typeless_storage default argument.
33945 * tree.c (type_cache_hasher::equal): Also compare
33946 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33947 (build_array_type): Add typeless_storage argument, set
33948 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33950 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33951 (build_array_type): Likewise. Add typeless_storage argument.
33953 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33954 Jakub Jelinek <jakub@redhat.com>
33956 PR tree-optimization/80426
33957 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33958 operation on symbolic operands, also compute the overflow for the
33959 invariant part when the operation degenerates into a negation.
33961 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33964 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33965 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33968 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33970 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33973 * config/avr/avr.c (tree.h): Include it.
33974 (cgraph.h): Include it.
33975 (avr_encode_section_info): Don't warn for uninitialized progmem
33976 variable if it's just an alias.
33978 2017-04-19 Richard Biener <rguenther@suse.de>
33981 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33982 when needed by AutoPGO.
33984 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33987 * doc/lto.texi: Remove an extra 'that'.
33989 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33991 PR rtl-optimization/80429
33992 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33993 are only used in debug insns.
33995 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33996 Vladimir Makarov <vmakarov@redhat.com>
33998 * config/sparc/predicates.md (input_operand): Add comment. Return
33999 true for any memory operand when LRA is in progress.
34000 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34002 2017-04-18 Jeff Law <law@redhat.com>
34005 * mips.md ({return,simple_return}_internal): Do not overwrite
34008 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34010 PR tree-optimization/80443
34011 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34012 instead of adding 1, subtract -1 and similarly instead of subtracting
34015 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34017 PR rtl-optimization/80357
34018 * haifa-sched.c (tmp_bitmap): New variable.
34019 (model_recompute): Handle duplicate use records.
34020 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34021 (free_global_sched_pressure_data): Free it.
34023 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34026 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34027 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34028 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34029 instead of SYSTEM_HEADER_DIR.
34031 2017-04-18 Jeff Law <law@redhat.com>
34033 PR middle-end/80422
34034 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34035 predecessors after walking up the insn chain.
34037 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34040 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34041 sizetype type into debug info.
34043 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34046 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34047 unneeded test for TARGET_UPPER_REGS_SF.
34048 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34050 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34053 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34054 instead of gsi_after_labels.
34056 2017-04-18 Jeff Law <law@redhat.com>
34058 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34062 2017-04-13 Jeff Law <law@redhat.com>
34063 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34064 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34066 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34069 * config/avr/avr.c (intl.h): Include it.
34070 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34072 2017-04-18 Martin Liska <mliska@suse.cz>
34074 PR gcov-profile/78783
34075 * gcov-tool.c (gcov_output_files): Validate that destination
34076 file is either removed by the tool or by a user.
34078 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34079 Guy Benyei <guybe@mellanox.com>
34081 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34082 block, and do not negate it, the stored id is already negative.
34084 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34086 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34088 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34091 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34092 masks of options that should be turned off if the VSX vector
34093 options are turned off.
34094 (OTHER_P8_VECTOR_MASKS): Likewise.
34095 (OTHER_VSX_VECTOR_MASKS): Likewise.
34096 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34097 rs6000_disable_incompatible_switches to validate no type switches
34099 (rs6000_incompatible_switch): New function to disallow turning on
34100 other vector options if -mno-vsx, -mno-power8-vector, or
34101 -mno-power9-vector are specified.
34103 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34105 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34107 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34109 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34110 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34111 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34112 (ARG_POINTER_CFA_OFFSET): Likewise.
34114 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34116 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34117 conditions to take advantage of various optimizations.
34119 2017-04-13 Jeff Law <law@redhat.com>
34121 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34122 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34123 (zero_extendsidi2_dext): Likewise.
34125 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34128 * fold-const.c (fold_ternary_loc): Revert
34129 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34132 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34134 PR rtl-optimization/80343
34135 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34136 new scratch pseudo.
34138 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34141 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34142 to ubsan_encode_value.
34144 2017-04-13 Jeff Law <law@redhat.com>
34146 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34147 appearing in DEBUG_INSNs.
34149 2017-04-13 Martin Liska <mliska@suse.cz>
34151 PR gcov-profile/80413
34152 * gcov-io.c (gcov_write_string): Copy to buffer just when
34153 allocated size is greater than zero.
34155 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34158 * dwarf2out.c (decls_for_scope): Ignore declarations of
34159 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34161 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34164 * ipa-visibility.c (non_local_p): Fix typos.
34165 (localize_node): When localizing symbol in same comdat group,
34166 dissolve the group only when we know external symbols are going
34168 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34170 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34172 PR tree-optimization/79390
34173 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34174 order does not result in usable sequence, retry with reversed operand
34180 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34181 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34182 op0 instead of fold_convert_loc (loc, type, arg0).
34184 2017-04-12 Jeff Law <law@redhat.com>
34186 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34187 has a delay slot in the generated code.
34189 * config/cris/cris.md (cris_preferred_reload_class): Return
34190 GENNONACR_REGS rather than GENERAL_REGS.
34192 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34195 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34196 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34197 signedness of the result type.
34199 2017-04-12 Richard Biener <rguenther@suse.de>
34200 Jeff Law <law@redhat.com>
34202 PR tree-optimization/80359
34203 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34204 trim stores to TARGET_MEM_REFs.
34206 2017-04-12 Richard Biener <rguenther@suse.de>
34208 PR tree-optimization/79390
34209 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34210 threading case even more.
34212 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34215 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34216 for quad_address_p for TImode, instead of just not indexed_address.
34218 2017-04-12 Richard Biener <rguenther@suse.de>
34219 Bernd Edlinger <bernd.edlinger@hotmail.de>
34221 PR middle-end/79671
34222 * alias.c (component_uses_parent_alias_set_from): Handle
34223 TYPE_TYPELESS_STORAGE.
34224 (get_alias_set): Likewise.
34225 * tree-core.h (tree_type_common): Add typeless_storage flag.
34226 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34227 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34228 for types containing members with TYPE_TYPELESS_STORAGE.
34229 (place_field): Likewise.
34230 (layout_type): Likewise for ARRAY_TYPE.
34231 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34232 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34233 TYPE_TYPELESS_STORAGE.
34234 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34236 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34239 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34240 first argument to type.
34242 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34246 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34247 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34248 (rs6000_expand_binop_builtin): Likewise.
34249 (rs6000_expand_ternop_builtin): Likewise; also add missing
34250 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34251 vshasigma built-ins.
34252 * doc/extend.texi: Document that vec_xxpermdi's third argument
34253 must be a constant.
34255 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34257 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34258 Use shift_const cost parameter when calculating gain of STV shifts.
34260 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34262 PR rtl-optimization/70478
34263 * lra-constraints.c (process_alt_operands): Check memory for
34264 disfavoring memory insn operand.
34266 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34268 PR middle-end/80100
34269 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34270 left shift in unsigned HOST_WIDE_INT type.
34272 PR rtl-optimization/80385
34273 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34274 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34277 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34278 if they have any depend clauses.
34280 2017-04-11 Martin Liska <mliska@suse.cz>
34283 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34284 * ipa-split.c (split_function): Create a local comdat symbol
34285 if caller is in a comdat group.
34287 2017-04-11 Martin Liska <mliska@suse.cz>
34290 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34293 2017-04-11 Martin Sebor <msebor@redhat.com>
34295 PR middle-end/80364
34296 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34297 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34299 (directive::set_width, directive::set_precision, format_character):
34301 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34304 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34307 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34308 conflict, set target->arch_name instead of target->cpu_name.
34310 2017-04-11 Richard Biener <rguenther@suse.de>
34312 PR tree-optimization/80374
34313 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34314 build_zero_cst, remove fold_convertible_p check again.
34316 2017-04-11 Martin Liska <mliska@suse.cz>
34319 * ubsan.c (instrument_object_size): Do not instrument register
34322 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34325 * config/i386/i386-builtin-types.def
34326 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34327 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34328 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34329 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34330 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34331 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34332 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34333 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34334 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34335 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34336 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34337 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34338 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34339 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34340 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34341 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34342 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34343 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34344 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34345 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34346 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34347 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34348 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34349 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34350 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34351 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34352 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34353 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34354 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34355 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34356 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34357 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34358 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34359 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34360 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34361 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34362 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34363 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34364 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34365 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34366 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34367 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34368 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34369 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34370 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34371 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34373 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34374 flag to second_arg_count, handle 4 argument function type _COUNT
34375 aliases, handle second_arg_count on second argument rather than last.
34377 2017-04-10 Jeff Law <law@redhat.com>
34379 PR tree-optimization/80374
34380 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34381 record anything if we can not convert integer_zero_node to the
34384 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34387 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34388 Enhance special handling given to the TARGET_P9_MINMAX option in
34389 relation to certain other options.
34391 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34393 PR tree-optimization/80153
34394 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34395 remove POINTER_PLUS_EXPR's base part directly, rather than through
34398 2017-04-10 Richard Biener <rguenther@suse.de>
34399 Bin Cheng <bin.cheng@arm.com>
34401 PR tree-optimization/80153
34402 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34403 the first element of pointer type aff_tree. Build result expr in
34405 (add_elt_to_tree): Convert to type unconditionally. Remove other
34406 fold_convert calls.
34407 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34408 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34410 2017-04-10 Richard Biener <rguenther@suse.de>
34412 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34415 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34417 PR rtl-optimization/70478
34418 * lra-constraints.c (curr_small_class_check): New.
34419 (update_and_check_small_class_inputs): New.
34420 (process_alt_operands): Update curr_small_class_check. Disfavor
34421 alternative insn memory operands. Check available regs for small
34424 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34427 * config/mips/mips.opt (-mvirt): Update description.
34428 * doc/invoke.texi (-mvirt): Likewise.
34430 2017-04-10 Richard Biener <rguenther@suse.de>
34432 PR middle-end/80362
34433 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34434 looking for NEGATE_EXPR in -A / -B to A / B folding.
34436 2017-04-10 Martin Liska <mliska@suse.cz>
34438 PR gcov-profile/80224
34439 * gcov.c (print_usage): Fix usage string.
34440 (get_gcov_intermediate_filename): Remove.
34441 (output_gcov_file): Use both for normal and intermediate format.
34442 (generate_results): Do not initialize special file for
34443 intermediate format.
34445 2017-04-10 Richard Biener <rguenther@suse.de>
34447 PR tree-optimization/80304
34448 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34451 2017-04-10 Nathan Sidwell <nathan@acm.org>
34454 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34455 (rs6000_init_builtins): Use it.
34457 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34459 * config/arm/arm.md (<mrc>): Add mode to SET source.
34460 (<mrrc>): Likewise.
34462 2017-04-10 Richard Biener <rguenther@suse.de>
34464 PR middle-end/80344
34465 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34467 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34470 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34471 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34472 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34473 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34474 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34475 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34476 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34477 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34478 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34479 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34480 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34481 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34482 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34483 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34484 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34485 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34486 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34487 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34488 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34489 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34490 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34491 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34492 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34494 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34496 PR rtl-optimization/70478
34497 * lra-constraints.c: Reverse the last patch.
34499 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34501 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34502 Add comment for WCHAR_T.
34504 2017-04-08 Martin Liska <mliska@suse.cz>
34507 2017-04-07 Martin Liska <mliska@suse.cz>
34510 * ipa-split.c (split_function): Add function part to a same comdat
34513 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34516 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34518 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34520 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34521 (vec_pairing): ...to this.
34522 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34523 (rs6000_sched_init): Adjust for name change.
34524 (struct rs6000_sched_context): Likewise.
34525 (rs6000_init_sched_context): Likewise.
34526 (rs6000_set_sched_context): Likewise.
34528 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34534 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34536 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34537 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34538 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34540 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34542 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34544 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34546 PR rtl-optimization/70703
34547 * ira-color.c (update_conflict_hard_regno_costs): Use
34548 int64_t instead of HOST_WIDE_INT.
34550 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34552 PR rtl-optimization/70478
34553 * lra-constraints.c (process_alt_operands): Disfavor alternative
34554 insn memory operands.
34556 2017-04-07 Jeff Law <law@redhat.com>
34558 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34559 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34561 2017-04-07 Martin Liska <mliska@suse.cz>
34564 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34565 Show error message instead of an ICE.
34567 2017-04-07 Martin Liska <mliska@suse.cz>
34570 * ipa-split.c (split_function): Add function part to a same comdat
34573 2017-04-07 Richard Biener <rguenther@suse.de>
34575 PR middle-end/80341
34576 * tree.c (get_unwidened): Also handle ! for_type case for
34578 * convert.c (do_narrow): Split out from ...
34579 (convert_to_integer_1): ... here. Do not pass final truncation
34580 type to get_unwidened for TRUNC_DIV_EXPR.
34582 2017-04-07 Richard Biener <rguenther@suse.de>
34584 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34586 (aff_combination_const): Adjust.
34587 (aff_combination_scale): Likewise.
34588 (aff_combination_add_elt): Likewise.
34589 (aff_combination_add_cst): Likewise.
34590 (aff_combination_convert): Likewise.
34591 (add_elt_to_tree): Likewise. Remove unused argument.
34592 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34594 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34596 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34598 * config/arm/arm.c (arm_default_short_enums): Use
34599 ARM_DEFAULT_SHORT_ENUMS.
34600 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34602 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34605 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34606 members with redundant out-of-class redeclaration.
34608 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34611 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34612 * config/i386/i386.md (*zero_extendsidi2):
34613 Add (?*x,*x) and (?*v,*v) alternatives.
34615 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34618 * config/i386/i386.c (ix86_expand_builtin)
34619 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34620 mode from insn data. Convert operands to insn operand mode.
34621 Copy operands that don't satisfy insn predicate to a register.
34623 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34625 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34628 2017-04-06 Richard Biener <rguenther@suse.de>
34630 PR tree-optimization/80334
34631 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34632 preserve alignment of accesses.
34634 2017-04-06 Richard Biener <rguenther@suse.de>
34636 PR tree-optimization/80262
34637 * tree-sra.c (build_ref_for_offset): Preserve address-space
34639 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34640 Drop useless address-space information on MEM_REF offsets.
34642 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34644 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34646 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34648 PR rtl-optimization/70703
34649 * ira-color.c (update_conflict_hard_regno_costs): Use
34650 HOST_WIDE_INT instead of long.
34652 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34655 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34656 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34658 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34659 for x86_64 target. Handle -m3dnowa option.
34661 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34663 PR rtl-optimization/70703
34664 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34665 (update_conflict_hard_regno_costs): Use long instead of unsigned
34666 arithmetic for cost calculation.
34668 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34669 Bernd Edlinger <bernd.edlinger@hotmail.de>
34672 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34675 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34678 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34679 ptr_mode with Pmode throughout.
34680 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34681 into probe_stack_range and use DImode.
34683 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34686 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34687 call_eh_return is true.
34689 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34691 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34692 Initialize last_match_fntype_index.
34694 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34697 * tree-nvr.c: Include internal-fn.h.
34698 (pass_return_slot::execute): Ignore internal calls without
34701 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34702 Richard Biener <rguenther@suse.de>
34705 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34706 captures used multiple times, except for the last use.
34707 * generic-match-head.c: Include gimplify.h.
34709 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34711 PR tree-optimization/79390
34712 * target.h (struct noce_if_info): Declare.
34713 * targhooks.h (default_noce_conversion_profitable_p): Declare.
34714 * target.def (noce_conversion_profitable_p): New target hook.
34715 * ifcvt.h (struct noce_if_info): New type, moved from ...
34716 * ifcvt.c (struct noce_if_info): ... here.
34717 (noce_conversion_profitable_p): Renamed to ...
34718 (default_noce_conversion_profitable_p): ... this. No longer
34720 (noce_try_store_flag_constants, noce_try_addcc,
34721 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
34722 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
34723 instead of noce_conversion_profitable_p.
34724 * config/i386/i386.c: Include ifcvt.h.
34725 (ix86_option_override_internal): Don't override
34726 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
34727 (ix86_noce_conversion_profitable_p): New function.
34728 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
34729 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
34730 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
34731 * doc/tm.texi: Regenerated.
34733 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34735 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
34738 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34741 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34742 instructions for small multiply cores.
34744 2017-04-04 Jeff Law <law@redhat.com>
34746 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34748 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34749 that are not set by the loop over the elements.
34751 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34754 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34755 int mode, convert_modes it to mode as unsigned, otherwise use
34756 lowpart_subreg to mode rather than SImode.
34757 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34758 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34759 Use DImode instead of SImode for the shift count operand.
34760 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34763 2017-04-04 Richard Biener <rguenther@suse.de>
34765 PR middle-end/80281
34766 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
34767 arithmetic done for the negate or the plus. Simplify.
34768 (A - (-B) -> A + B): Likewise.
34769 * fold-const.c (split_tree): Make sure to not negate pointers.
34771 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
34773 PR rtl-optimization/60818
34774 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34775 a compare of comparisons with the thing compared if this results
34776 in a different machine mode.
34778 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
34780 * alias.c (base_alias_check): Fix typo in comment.
34781 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
34782 * cgraphunit.c (symbol_table::compile): Likewise.
34783 * collect2.c (maybe_run_lto_and_relink): Likewise.
34784 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
34785 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
34786 * config/avr/avr.c (avr_map_op_t): Likewise.
34787 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
34788 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
34789 * config/epiphany/epiphany.md (movcc): Likewise.
34790 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
34791 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
34793 * config/mips/mips.c (mips_save_restore_reg): Likewise.
34794 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
34795 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
34796 * config/sh/sh.c (sh_rtx_costs): Likewise.
34797 * fold-const.c (fold_truth_andor): Likewise.
34798 * genautomata.c (collapse_flag): Likewise.
34799 * gengtype.h (struct type::u::s): Likewise.
34800 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34801 * input.c (FORMAT_AMOUNT): Likewise.
34802 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34803 (known_aggs_to_agg_replacement_list): Likewise.
34804 * ipa-inline-analysis.c: Likewise.
34805 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34806 * ipa-polymorphic-call.c
34807 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34808 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34809 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34811 * modulo-sched.c (apply_reg_moves): Likewise.
34812 * omp-expand.c (build_omp_regions_1): Likewise.
34813 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34814 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34815 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34816 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34817 * value-prof.c: Likewise.
34818 * var-tracking.c (val_reset): Likewise.
34820 2017-04-03 Richard Biener <rguenther@suse.de>
34822 PR tree-optimization/80275
34823 * fold-const.c (split_address_to_core_and_offset): Handle
34826 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34828 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34829 descriptors is at least equal to that of functions.
34831 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34833 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34835 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34838 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34839 (mov<IMOD4:mode>): New expander.
34840 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34842 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34844 PR rtl-optimization/79405
34845 * fwprop.c (propagations_left): New variable.
34846 (forward_propagate_into): Decrement it.
34847 (fwprop_init): Initialize it.
34848 (fw_prop): If the variable has reached zero, stop propagating.
34849 (fwprop_addr): Ditto.
34851 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34854 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34855 a FUNCTION_DECL, pass it as decl instead of origin to
34858 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34860 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34863 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34866 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34867 TARGET_VSX_SMALL_INTEGER.
34869 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34871 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34872 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34874 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34876 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34877 extraction from odd-numbered MSA register.
34879 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34881 PR middle-end/80173
34882 * expmed.c (store_bit_field_1): Don't attempt to create
34883 a word subreg out of hard registers wider than word if they
34884 have HARD_REGNO_NREGS of 1 for their mode.
34886 PR middle-end/80163
34887 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34888 conversions to integer types wider than word and pointer.
34891 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34892 (rtx_equal_for_cselib_p): Pass 0 to it.
34893 * cselib.c (cselib_hasher::equal): Likewise.
34894 (rtx_equal_for_cselib_1): Add depth argument. If depth
34895 is 128, don't look up VALUE locs and punt. Increment
34896 depth in recursive calls when walking VALUE locs.
34898 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34900 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34901 (make_gcov_file_name): Use the canonical path name for generating
34903 (read_line): Fix handling of files with ascii null bytes.
34905 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34907 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34908 to initialise a vector register instead
34909 of using a const_int.
34911 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34913 PR translation/80189
34914 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34915 diagnostic messages.
34917 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34920 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34921 (dfp_diex_<mode>): Update mode of operand 1.
34922 * doc/extend.texi (dxex, dxexq): Document change to return type.
34923 (diex, diexq): Document change to argument type.
34925 2017-03-30 Martin Jambor <mjambor@suse.cz>
34928 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34929 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34930 it reflects the signature changes performed at the callee side.
34931 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34932 to cgraph_build_function_type_skip_args.
34933 (build_function_decl_skip_args): Adjust call to the above function.
34935 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34938 * config/i386/sse.md
34939 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34940 register as dest whenever it is a MEM not rtx_equal_p to the
34941 corresponding dup operand, and when forcing into reg move the
34942 reg into the memory afterwards.
34943 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34944 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34945 for the force_reg mode.
34946 (avx512vl_vextractf128<mode>): Use register as dest either
34947 always when a MEM, or when it is a MEM not rtx_equal_p to the
34948 corresponding dup operand, or even not when it is a CONST_VECTOR
34949 depending on the mode and lo vs. hi.
34950 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34952 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34953 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34954 Likewise. Require that operands[2] is even.
34955 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34956 Remove extraneous parens. Require that operands[2] is a multiple
34958 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34959 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34960 disallow memory then.
34962 2017-03-30 Richard Biener <rguenther@suse.de>
34964 PR tree-optimization/77498
34965 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34966 to non-constants over backedges.
34968 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34970 PR rtl-optimization/80233
34971 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34972 as last_combined_insn. Do not test for BARRIER_P separately.
34974 2017-03-29 Andreas Schwab <schwab@suse.de>
34977 * calls.c (prepare_call_address): Convert funexp to Pmode before
34978 copying to temp reg.
34980 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34982 PR tree-optimization/80158
34983 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34984 Handle possible future case of more than one alternate
34986 (replace_rhs_if_not_dup): Likewise.
34987 (replace_one_candidate): Likewise.
34989 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34991 PR rtl-optimization/80193
34992 * ira.c (ira): Do not check allocation for LRA.
34994 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34996 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34997 (nvptx_output_simt_exit): Declare.
34998 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34999 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35000 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35001 (nvptx_declare_function_name): ...here. Emit declaration of local
35002 memory space buffer for omp_simt_enter insn.
35003 (nvptx_output_unisimt_switch): New.
35004 (nvptx_output_softstack_switch): New.
35005 (nvptx_output_simt_enter): New.
35006 (nvptx_output_simt_exit): New.
35007 * config/nvptx/nvptx.h (struct machine_function): New fields
35008 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35009 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35010 (UNSPECV_SIMT_EXIT): Ditto.
35011 (omp_simt_enter_insn): New insn.
35012 (omp_simt_enter): New expansion.
35013 (omp_simt_exit): New insn.
35014 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35016 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35017 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35018 (expand_GOMP_SIMT_EXIT): New.
35019 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35020 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35021 (GOMP_SIMT_EXIT): Ditto.
35022 * target-insns.def (omp_simt_enter): New insn.
35023 (omp_simt_exit): Ditto.
35024 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35026 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35027 (lower_rec_input_clauses): Likewise.
35028 (lower_lastprivate_clauses): Handle SIMT privatization.
35030 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35031 (ompdevlow_adjust_simt_enter): New.
35032 (find_simtpriv_var_op): New.
35033 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35034 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35036 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35037 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35038 (copy_decl_for_dup_finish): Ditto.
35040 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35042 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35045 * config/i386/i386.c (ix86_option_override_internal): Always
35046 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35048 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35050 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35052 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35054 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35055 mark new edge's irreducible flag accordign to it.
35056 (vect_do_peeling): Check loop preheader edge's irreducible flag
35057 and pass it to function slpeel_add_loop_guard.
35059 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35061 PR tree-optimization/80218
35062 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35063 Update block frequencies and counts.
35065 2017-03-28 Richard Biener <rguenther@suse.de>
35067 PR tree-optimization/78644
35068 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35069 of a simplification result we may not use it at all.
35071 2017-03-28 Richard Biener <rguenther@suse.de>
35074 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35075 without arguments, generate default definition of a SSA name.
35077 2017-03-28 Richard Biener <rguenther@suse.de>
35079 PR middle-end/80222
35080 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35081 TYPE_REF_CAN_ALIAS_ALL references.
35082 * fold-const.c (fold_indirect_ref_1): Likewise.
35084 2017-03-28 Martin Liska <mliska@suse.cz>
35087 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35088 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35090 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35091 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35093 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35094 (EXTRA_SPECS): Define.
35095 (SUBTARGET_EXTRA_SPECS): Likewise.
35096 (SUBTARGET_CPP_SPEC): Likewise.
35097 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35098 SUBTARGET_EXTRA_SPECS.
35099 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35101 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35103 * config/arc/simdext.md (vst64_insn): Update pattern.
35104 (vld32wh_insn): Likewise.
35105 (vld32wl_insn): Likewise.
35106 (vld64_insn): Likewise.
35107 (vld32_insn): Likewise.
35109 2017-03-28 Marek Polacek <polacek@redhat.com>
35112 * fold-const.c (fold_comparison): Use protected_set_expr_location
35113 instead of SET_EXPR_LOCATION.
35115 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35117 * tree.c (add_expr): Avoid name lookup warning.
35119 2017-03-27 Jeff Law <law@redhat.com>
35121 PR tree-optimization/80216
35122 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35123 function name. Limit recursion depth.
35124 (record_temporary_equivalences): Corresponding changes.
35126 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35128 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35131 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35134 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35136 * cfgcleanup.c (reg_note_cfa_p): New array.
35137 (insns_have_identical_cfa_notes): New function.
35138 (old_insns_match_p): Don't cross-jump in between /f
35139 and non-/f instructions. If both i1 and i2 are frame related,
35140 verify all CFA notes, their order and content.
35142 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35145 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35146 HImode and SImode with zero extend to DImode to one insn.
35147 (bswap<mode>2_extenddi): Likewise.
35148 (bswapsi2_extenddi): Likewise.
35149 (bswaphi2_extendsi): Likewise.
35150 (bswaphi2): Combine bswap HImode and SImode into one insn.
35151 Separate memory insns from swapping register.
35152 (bswapsi2): Likewise.
35153 (bswap<mode>2): Likewise.
35154 (bswaphi2_internal): Delete, no longer used.
35155 (bswapsi2_internal): Likewise.
35156 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35157 store, and gpr<-gpr swap insns.
35158 (bswap<mode>2_store): Likewise.
35159 (bswaphi2_reg): Register only splitter, combine with the splitter.
35160 (bswaphi2 splitter): Likewise.
35161 (bswapsi2_reg): Likewise.
35162 (bswapsi2 splitter): Likewise.
35163 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35164 the insns into load, store, and register/register insns.
35165 (bswapdi2_ldbrx): Likewise.
35166 (bswapdi2_load): Likewise.
35167 (bswapdi2_store): Likewise.
35168 (bswapdi2_reg): Likewise.
35170 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35172 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35173 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35175 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35178 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35180 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35181 special handling for target option conflicts between dform
35182 options (-mpower9-dform, -mpower9-dform-vector,
35183 -mpower9-dform-scalar) and -mno-direct-move.
35185 2017-03-27 Richard Biener <rguenther@suse.de>
35187 PR tree-optimization/80181
35188 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35190 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35192 * config/arc/predicates.md (move_double_src_operand): Replace the
35193 call to move_double_src_operand with a call to address_operand.
35195 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35197 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35198 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35199 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35201 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35203 * config/arc/predicates.md (long_immediate_loadstore_operand):
35204 Consider scaled addresses cases.
35206 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35208 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35209 restored when in interrupt.
35210 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35211 doesn't have delay slot.
35213 2017-03-27 Richard Biener <rguenther@suse.de>
35216 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35217 inlined thunk clones.
35219 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35222 * asan.c (instrument_derefs): Copy over last operand from
35223 original COMPONENT_REF to the new COMPONENT_REF with
35224 DECL_BIT_FIELD_REPRESENTATIVE.
35225 * ubsan.c (instrument_object_size): Likewise.
35227 2017-03-27 Richard Biener <rguenther@suse.de>
35229 PR tree-optimization/80170
35230 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35231 sure DR/SCEV didnt fold in constants we do not see when looking
35232 at the reference base alignment.
35234 2017-03-27 Richard Biener <rguenther@suse.de>
35236 PR middle-end/80171
35237 * gimple-fold.c (fold_ctor_reference): Properly guard against
35238 NULL return value from canonicalize_constructor_val.
35240 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35243 * config/i386/i386.c (ix86_expand_builtin)
35244 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35245 flags reg setting and flags reg using instructions.
35246 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35247 clobbering instructions to zero extend op2.
35249 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35251 * doc/install.texi (Configuration) <--with-aix-soname>:
35252 Update link to AIX ld.
35254 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35256 PR rtl-optimization/80160
35257 PR rtl-optimization/80159
35258 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35259 reg_alternate_class into account.
35261 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35264 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35265 to consider in curr_insn_transform.
35267 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35269 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35270 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35271 and emit_mode_inner.
35273 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35275 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35276 argument to the overloaded builtin variants. Use the new flag to
35277 deprecate certain builtin variants.
35278 * config/s390/s390-builtin-types.def: Add new builtin types.
35279 * config/s390/s390-builtins.h: Support new flags field for
35280 overloaded builtins.
35281 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35282 (s390_macro_to_expand): Enable vector float data type.
35283 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35284 builtins by incrementing the __VEC__ version number.
35285 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35287 (s390_resolve_overloaded_builtin): Emit error messages depending
35288 on the builtin flags.
35289 * config/s390/s390.c (s390_expand_builtin): Support additional
35290 flags argument. Change error message to match the messages
35291 emitted in s390-c.c.
35292 * config/s390/s390.md: New UNSPEC_* constants.
35293 (op_type): Add new instruction types.
35294 * config/s390/vecintrin.h: Add new builtins and test data class
35296 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35297 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35298 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35299 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35300 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35301 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35303 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35304 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35305 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35306 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35308 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35309 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35310 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35311 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35313 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35314 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35315 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35316 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35317 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35318 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35319 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35321 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35322 ("vec_scatter_element<V_HW_4:mode>_DI")
35323 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35324 ("vec_fpint<mode>", "vflls")
35325 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35326 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35327 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35328 ("*vec_cmphe<mode>_cc"): ... these.
35330 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35331 mode constant instead of magic value.
35333 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35335 * config/s390/s390.c (s390_expand_vec_compare): Support other
35336 vector floating point modes than just V2DF.
35337 (s390_expand_vcond): Likewise.
35338 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35339 (s390_cannot_change_mode_class): Prevent mode changes between TF
35340 and V1TF in vector registers.
35341 * config/s390/s390.md (DF, SF): New mode attributes.
35342 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35343 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35344 SFmode support for VRs.
35345 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35347 (VFT, VF_HW): New mode iterators.
35348 (vw, sdx): New mode attributes.
35349 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35350 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35351 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35352 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35353 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35354 also the new vector floating point modes. Renaming to ...
35356 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35357 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35358 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35359 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35360 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35361 ("vec_unordered<mode>"): ... these.
35363 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35364 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35365 ("*vec_extendv2df"): New insn definitions.
35367 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35369 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35370 ("mulditi3_2", "*muldi3_sign"): New patterns.
35371 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35372 rename the pattern definition.
35374 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35376 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35378 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35380 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35382 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35383 instruction if possible.
35384 * config/s390/vector.md (vec_halfnumelts): New mode
35386 ("*vec_vllezlf<mode>"): New pattern.
35388 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35390 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35391 ("popcountv4si2", "popcountv2di2"): Rename to ...
35392 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35393 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35395 ("popcount<mode>2_vxe"): New pattern.
35397 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35399 * common/config/s390/s390-common.c (processor_flags_table): Add
35401 * config.gcc: Add arch12.
35402 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35403 Default to arch12 for unknown CPU model numbers.
35404 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35405 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35406 PROCESSOR_max sanity check.
35407 * config/s390/s390-opts.h (enum processor_type): Add
35409 * config/s390/s390.c (processor_table): Add arch12.
35410 (s390_expand_builtin): Add check for B_VXE flag.
35411 (s390_issue_rate): Add PROCESSOR_ARCH12.
35412 (s390_get_sched_attrmask): Likewise.
35413 (s390_get_unit_mask): Likewise.
35414 (s390_sched_score): Enable z13 scheduling for arch12.
35415 (s390_sched_reorder): Likewise.
35416 (s390_sched_variable_issue): Likewise.
35417 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35419 (s390_tune_attr): Use z13 scheduling also for arch12.
35420 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35421 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35422 (TARGET_VXE_P): New macros.
35423 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35424 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35425 * config/s390/s390.opt: Add arch12 as processor_type.
35427 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35429 * config/s390/s390.md
35430 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35431 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35432 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35434 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35435 Rename expanders to ...
35437 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35438 ("fixuns_truncdddi2_emu"): ... these.
35440 ("fixuns_trunc<mode>si2_emu"): New expander.
35442 ("*fixuns_truncdfdi2_z13"): Rename to ...
35443 ("*fixuns_truncdfdi2_vx"): ... this.
35445 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35447 * config/s390/2964.md: Remove the single element vector compare
35448 instructions which are no longer used.
35449 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35451 (s390_canonicalize_comparison): Remove handling of DFmode
35453 (s390_expand_vec_compare_scalar): Remove function.
35454 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35455 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35457 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35459 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35461 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35463 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35464 will anyway by matched by mov<mode>_64dfp.
35466 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35468 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35469 vlef/vstef. Add missing operand to vleif.
35471 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35473 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35474 pair for all vector types with 64 bit elements.
35475 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35476 * config/s390/vector.md (V_HW_64): ... here.
35477 (V_128_NOSINGLE): New mode iterator.
35478 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35479 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35480 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35481 ("*vec_load_pairv2di"): Change to ...
35482 ("*vec_load_pair<mode>"): ... this one.
35484 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35486 * config/s390/constraints.md: Add comments.
35487 (jKK): Reject element sizes > 8 bytes.
35488 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35490 * config/s390/s390.md: Add the s_operand checks formerly in
35491 s390_split_ok_p to various splitters where they are still
35493 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35494 for 128 bit vectors. Plus two splitters.
35496 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35498 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35501 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35504 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35505 error if the boundary argument is not constant.
35507 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35509 PR rtl-optimization/80112
35510 * loop-doloop.c (doloop_condition_get): Don't check condition
35511 if cmp isn't SET with IF_THEN_ELSE src.
35513 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35515 PR tree-optimization/80158
35516 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35517 replacing a candidate statement, also replace it for the
35518 candidate's alternate interpretation.
35519 (replace_rhs_if_not_dup): Likewise.
35520 (replace_one_candidate): Likewise.
35522 2017-03-24 Richard Biener <rguenther@suse.de>
35524 PR tree-optimization/80167
35525 * graphite-isl-ast-to-gimple.c
35526 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35528 (translate_isl_ast_to_gimple::get_rename): Likewise.
35530 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35532 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35533 handling of certain combinations of target options, including the
35534 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35535 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35537 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35540 * config/arm/arm.md (*load_multiple): Add reload_completed to
35541 matching condition.
35543 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35544 Richard Biener <rguenth@suse.de>
35546 PR tree-optimization/79908
35547 PR tree-optimization/80136
35548 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35549 been cast away, gimplify_and_add suffices.
35551 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35553 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35555 2017-03-23 Richard Biener <rguenther@suse.de>
35557 PR tree-optimization/80032
35558 * gimplify.c (gimple_push_cleanup): Forced unconditional
35559 cleanups still have to go to the conditional_cleanups
35562 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35564 PR tree-optimization/80072
35565 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35567 (next_operand_entry_id): Change type to unsigned int.
35568 (sort_by_operand_rank): Make sure to return the right return value
35569 even if unsigned fields are bigger than INT_MAX.
35570 (struct oecount): Change cnt and id type to unsigned int.
35571 (oecount_hasher::equal): Formatting fix.
35572 (oecount_cmp): Make sure to return the right return value
35573 even if unsigned fields are bigger than INT_MAX.
35574 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35577 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35578 TREE_READONLY on result if writing it more than once.
35581 * doc/invoke.texi (-fsanitize=thread): Document that with
35582 -fnon-call-exceptions atomics are not able to throw
35586 * tsan.c: Include tree-eh.h.
35587 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35588 maybe_clean_or_replace_eh_stmt where needed.
35589 (instrument_memory_accesses): Add cfg_changed argument.
35590 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35591 if it returned true.
35592 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35594 PR rtl-optimization/63191
35595 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35596 wrapper function, moved the whole old content into ...
35597 (ix86_delegitimize_address_1): ... this. New inline function.
35598 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35599 true as last argument instead of ix86_delegitimize_address.
35601 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35603 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35604 cortexa57_branch_cost.
35606 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35608 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35610 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35613 * doc/md.texi (Constraints): Document wA constraint.
35614 * config/rs6000/constraints.md (wA): New.
35615 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35616 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35617 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35618 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35620 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35623 * gimplify.c (is_oacc_declared): New function.
35624 (oacc_default_clause): Use it to set default flags for acc declared
35625 variables inside parallel regions.
35626 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35627 declared variables.
35628 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35629 declare attribute to any decl as necessary.
35631 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35634 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35635 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35636 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35637 (arm_arch_lpae): This.
35638 * config/arm/arm.c (arm_arch7ve): Rename into ...
35639 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35640 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35643 2017-03-22 Martin Liska <mliska@suse.cz>
35646 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35647 error message instead of an ICE.
35649 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35651 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35653 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35655 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35657 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35660 2017-03-21 Martin Sebor <msebor@redhat.com>
35662 * doc/extend.texi: Use "cannot" instead of "can't."
35663 * doc/hostconfig.texi: Same.
35664 * doc/install.texi: Same.
35665 * doc/invoke.texi: Same.
35666 * doc/loop.texi: Same.
35667 * doc/md.texi: Same.
35668 * doc/objc.texi: Same.
35669 * doc/rtl.texi: Same.
35670 * doc/tm.texi: Same.
35671 * doc/tm.texi.in: Same.
35672 * doc/trouble.texi: Same.
35674 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35677 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35678 (collect_checksum_attributes): Set it.
35679 (die_checksum_ordered): Use it.
35681 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35683 PR tree-optimization/79908
35684 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35685 change: For a VA_ARG whose LHS has been cast away, use
35686 force_gimple_operand to construct the side effects.
35688 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35690 PR translation/80001
35691 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35692 more amenable to translation.
35693 (oacc_loop_auto_partitions): Likewise.
35695 2017-03-21 Marek Polacek <polacek@redhat.com>
35696 Martin Sebor <msebor@redhat.com>
35698 PR tree-optimization/80109
35699 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35700 on INTEGRAL_TYPE_P.
35702 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35703 Segher Boessenkool <segher@kernel.crashing.org>
35706 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35707 check reg_used_between_p between insn and one of succ or succ2
35708 depending on if succ is artificial insn not inserted into insn
35711 2017-03-21 Martin Liska <mliska@suse.cz>
35713 PR gcov-profile/80081
35714 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
35715 * doc/gcc.texi: Include gcov-dump stuff.
35716 * doc/gcov-dump.texi: New file.
35718 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
35720 PR rtl-optimization/79150
35721 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
35722 conditional jump, if the jump is the last insn of the loop.
35724 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35725 Richard Biener <rguenth@suse.de>
35727 PR tree-optimization/79908
35728 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35729 been cast away, use force_gimple_operand to construct the side
35732 2017-03-21 Martin Liska <mliska@suse.cz>
35734 PR libfortran/79956
35735 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
35738 2017-03-21 Brad Spengler <spender@grsecurity.net>
35741 * plugin.c (htab_hash_plugin): New function.
35742 (add_new_plugin): Use it and adjust.
35743 (parse_plugin_arg_opt): Adjust.
35744 (init_one_plugin): Likewise.
35746 2017-03-21 Richard Biener <rguenther@suse.de>
35748 PR tree-optimization/80032
35749 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35750 if set force the cleanup to happen unconditionally.
35751 (gimplify_target_expr): Push inserted clobbers with force_uncond
35752 to avoid them being removed by control-dependent DCE.
35754 2017-03-21 Richard Biener <rguenther@suse.de>
35756 PR tree-optimization/80122
35757 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35758 va_arg_pack_len when the inlined call stmt requires pack
35760 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35762 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35765 * tsan.c (instrument_builtin_call): If the memory model argument
35766 is not a constant, assume it is valid.
35769 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
35772 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35774 PR rtl-optimization/79910
35775 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35776 if its dest is used by an insn before I2 (other than the combined
35777 insns themselves, which are properly handled already).
35779 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35782 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35784 * combine.c (record_used_regs): New static function.
35785 (try_combine): Handle situations where there is an additional
35786 instruction between I2 and I3 which needs to have a LOG_LINK
35790 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35792 * combine.c (try_combine): Delete redundant i1 test. Call
35793 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35795 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35798 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35799 alternatives 13/14.
35801 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35803 PR tree-optimization/80054
35804 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35805 the optimization if a PHI or any of its arguments is not dominated
35806 by the candidate's basis. Use gphi* rather than gimple* as
35808 (replace_profitable_candidates): Clean up a gimple* variable that
35809 should be a gphi* variable.
35811 2017-03-20 Martin Sebor <msebor@redhat.com>
35814 * doc/extend.texi (attribute constructor): Document present limitation.
35816 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35819 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35820 __POWER9_VECTOR__ #ifdef control, change template definition to
35821 use Power9-specific built-in function.
35822 (vec_any_eq): Likewise.
35823 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35824 to control outcomes from this test.
35825 (vector_ae_<mode>p): For VEC_F modes, likewise.
35827 2017-03-20 Ian Lance Taylor <iant@google.com>
35829 * config/i386/i386.c (ix86_function_regparm): Save an extra
35830 register for -fsplit-stack with DECL_STATIC_CHAIN.
35832 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35835 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35836 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35838 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35840 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35842 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35845 2017-03-20 Marek Polacek <polacek@redhat.com>
35848 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35850 2017-03-20 Richard Biener <rguenther@suse.de>
35852 PR tree-optimization/80113
35853 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35854 allocate extra SSA name for PHI def.
35855 (add_close_phis_to_outer_loops): Likewise.
35856 (add_close_phis_to_merge_points): Likewise.
35857 (copy_loop_close_phi_args): Likewise.
35858 (copy_cond_phi_nodes): Likewise.
35860 2017-03-20 Martin Liska <mliska@suse.cz>
35862 PR middle-end/79753
35863 * tree-chkp.c (chkp_build_returned_bound): Do not build
35864 returned bounds for a LHS that's not a BOUNDED_P type.
35866 2017-03-20 Martin Liska <mliska@suse.cz>
35870 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35871 COMPLEX_CST and VECTOR_CST.
35873 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35876 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35877 target operand. A new splitter adds the clobber statement in case
35878 the target operand is dead anyway.
35880 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35882 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35883 to age-old versions of binutils and glibc.
35885 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35887 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35889 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35891 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35893 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35895 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35896 requirement for binutils 2.13.
35898 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35900 * combine.c (try_combine): Delete redundant i1 test. Call
35901 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35903 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35905 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35906 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35908 <riscv64-*-elf>: Re-arrange section
35909 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35910 <riscv32-*-linux>: Likewise.
35911 <riscv64-*-elf>: Likewise
35912 <riscv64-*-linux>: Likewise.
35914 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35917 * aarch64.opt(verbose-cost-dump): Fix typo.
35919 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35922 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35923 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35925 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35927 * reload.c (find_reloads): When reloading a nonoffsettable address,
35928 use RELOAD_OTHER for it and its address reloads.
35930 PR rtl-optimization/79910
35931 * combine.c (record_used_regs): New static function.
35932 (try_combine): Handle situations where there is an additional
35933 instruction between I2 and I3 which needs to have a LOG_LINK
35936 2017-03-17 Jeff Law <law@redhat.com>
35938 PR tree-optimization/71437
35939 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35940 conditional in the hash table first.
35941 (vrp_dom_walker::before_dom_children): Extract condition from
35942 ASSERT_EXPR. Record condition, its inverion and any implied
35943 conditions as well.
35945 2017-03-17 Marek Polacek <polacek@redhat.com>
35946 Markus Trippelsdorf <markus@trippelsdorf.de>
35948 PR tree-optimization/80079
35949 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35952 2017-03-17 Richard Biener <rguenther@suse.de>
35954 PR middle-end/80075
35955 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35956 Properly verify the LHS before the RHS possibly claims to be
35958 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35961 2017-03-17 Martin Jambor <mjambor@suse.cz>
35963 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35964 (List of -O2 options): Likewise.
35965 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35968 2017-03-17 Tom de Vries <tom@codesourcery.com>
35970 * gcov-dump.c (print_usage): Print bug_report_url.
35972 2017-03-17 Richard Biener <rguenther@suse.de>
35974 PR middle-end/80050
35975 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35976 (parser::peek): Likewise.
35978 2017-03-17 Richard Biener <rguenther@suse.de>
35980 PR tree-optimization/80048
35981 * sese.c (free_sese_info): Properly release rename_map and
35982 copied_bb_map elements.
35984 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35986 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35987 Add linked-list forward and backlinks. Insert on
35988 construction, remove on destruction.
35989 (class pass_store_merging): Add m_stores_head field.
35990 (pass_store_merging::terminate_and_process_all_chains):
35991 Iterate over m_stores_head list.
35992 (pass_store_merging::terminate_all_aliasing_chains):
35994 (pass_store_merging::execute): Check for debug stmts first.
35995 Push new chains onto the m_stores_head stack.
35997 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36000 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36001 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36002 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36004 2017-03-16 Jeff Law <law@redhat.com>
36006 PR tree-optimization/71437
36007 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36008 member function. Implementation moved into after_dom_children
36009 member function and into the threader's thread_outgoing_edges
36011 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36012 some code into new thread_outgoing_edges.
36013 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36014 definition. Simplify marker handling (do it here). Assume we always
36015 have the available expression and the const/copies tables.
36016 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36018 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36019 * tree-vrp.c (equiv_stack): No longer file scoped.
36020 (vrp_dom_walker): New class.
36021 (vrp_dom_walker::before_dom_children): New member function.
36022 (vrp_dom_walker::after_dom_children): Likewise.
36023 (identify_jump_threads): Setup domwalker. Use it rather than
36024 walking edges in a random order by hand. Simplify setup/finalization.
36025 (finalize_jump_threads): Remove.
36026 (vrp_finalize): Do not call identify_jump_threads here.
36027 (execute_vrp): Do it here instead and call thread_through_all_blocks
36030 PR tree-optimization/71437
36031 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36033 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36035 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36036 (dom_opt_dom_walker::thread_across_edge): Remove
36037 handle_dominating_asserts argument. All callers changed.
36038 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36039 changes. Remove calls to lhs_of_dominating_assert. Other
36040 uses of handle_dominating_asserts turn into unconditional code
36041 (simplify_control_stmt_condition_1): Likewise.
36042 (simplify_control_stmt_condition): Likewise.
36043 (thread_through_normal_block, thread_across_edge): Likewise.
36044 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36045 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36046 object if it is not an SSA_NAME.
36047 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36048 before calling into the VRP specific simplifiers.
36049 (identify_jump_threads): Remove handle_dominating_asserts
36052 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36055 * tree-diagnostic.c (default_tree_printer): No longer static.
36056 * tree-diagnostic.h (default_tree_printer): New prototype.
36058 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36060 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36061 Change ins into fmov.
36063 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36065 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36066 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36067 Use h_con constraint for operand 1.
36068 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36069 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36071 2017-03-15 Jeff Law <law@redhat.com>
36073 PR tree-optimization/71437
36074 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36075 (record_temporary_equivalences): Use it.
36077 PR tree-optimization/71437
36078 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36079 tree-ssa-scopedtables.
36080 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36081 (record_conditions, record_cond, vuse_eq): Likewise.
36082 (record_edge_info): Adjust to API tweak of record_conditions.
36083 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36084 (record_temporary_equivalences, optimize_stmt): Likewise.
36085 (eliminate_redundant_computations): Likewise.
36086 (record_equivalences_from_stmt): Likewise.
36087 * tree-ssa-scopedtables.c: Include options.h and params.h.
36088 (vuse_eq): New function, moved from tree-ssa-dom.c
36089 (build_and_record_new_cond): Likewise.
36090 (record_conditions): Likewise. Accept vector of conditions rather
36091 than edge_equivalence structure for first argument.
36092 for the first argument.
36093 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36094 from tree-ssa-dom.c.
36095 (avail_exprs_stack::record_cond): Likewise.
36096 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36097 from tree-ssa-dom.c.
36098 (avail_exprs_stack): Add new member functions lookup_avail_expr
36100 (record_conditions): Declare.
36102 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36105 * lra-constraints.c (process_alt_operands): Increase reject for
36106 reloading an input/output operand.
36108 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36111 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36112 insns to convert from signed/unsigned char/short to IEEE 128-bit
36114 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36116 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36119 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36120 subreg of inner mode for values already in registers.
36122 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36124 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36125 iteration reg is used after the loop.
36127 2017-03-14 Martin Sebor <msebor@redhat.com>
36129 PR tree-optimization/79800
36130 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36131 precision in negative-positive range.
36132 (format_floating): Call non-const overload with adjusted precision.
36134 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36137 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36140 2017-03-14 Martin Sebor <msebor@redhat.com>
36142 PR middle-end/80020
36143 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36144 * builtins.def (aligned_alloc): Use it.
36147 * Makefile.in (GTFILES): Add calls.c.
36148 * calls.c: Include "gt-calls.h".
36150 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36152 PR rtl-optimization/79728
36153 * regs.h (struct target_regs): New field
36154 x_contains_allocatable_regs_of_mode.
36155 (contains_allocatable_regs_of_mode): New macro.
36156 * reginfo.c (init_reg_sets_1): Initialize it, and change
36157 contains_reg_of_mode so it includes global regs as well.
36158 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36159 rather than contains_regs_of_mode.
36161 2017-03-14 Martin Liska <mliska@suse.cz>
36163 * doc/invoke.texi: Document options that can't be combined with
36164 -fcheck-pointer-bounds.
36166 2017-03-14 Martin Liska <mliska@suse.cz>
36168 PR middle-end/79831
36169 * doc/invoke.texi (-Wchkp): Document the option.
36171 2017-03-14 Martin Liska <mliska@suse.cz>
36173 * Makefile.in: Install gcov-dump.
36175 2017-03-14 Martin Liska <mliska@suse.cz>
36177 * multiple_target.c (expand_target_clones): Bail out for
36178 an invalid attribute.
36180 2017-03-14 Richard Biener <rguenther@suse.de>
36182 * alias.c (struct alias_set_entry): Pack properly.
36183 * cfgloop.h (struct loop): Likewise.
36184 * cse.c (struct set): Likewise.
36185 * ipa-utils.c (struct searchc_env): Likewise.
36186 * loop-invariant.c (struct invariant): Likewise.
36187 * lra-remat.c (struct cand): Likewise.
36188 * recog.c (struct change_t): Likewise.
36189 * rtl.h (struct address_info): Likewise.
36190 * symbol-summary.h (function_summary): Likewise.
36191 * tree-loop-distribution.c (struct partition): Likewise.
36192 * tree-object-size.c (struct object_size_info): Likewise.
36193 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36194 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36195 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36196 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36197 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36198 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36199 (struct _stmt_vec_info): Likewise.
36201 2017-03-14 Martin Liska <mliska@suse.cz>
36204 * multiple_target.c (create_dispatcher_calls): Check that
36205 a target can create a function dispatcher.
36207 2017-03-14 Martin Liska <mliska@suse.cz>
36210 * multiple_target.c (expand_target_clones): Drop local.local
36211 flag for default implementation.
36213 2017-03-14 Richard Biener <rguenther@suse.de>
36215 PR tree-optimization/80030
36216 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36218 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36220 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36221 gcc_fallthrough() instead of __attribute__((fallthrough));
36223 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36225 * doc/gcc.texi: Remove "up" link to (DIR).
36226 * doc/gccint.texi: Ditto.
36228 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36230 * doc/install.texi (Specific) <avr>: Remove reference to
36233 2017-03-13 Jeff Law <law@redhat.com>
36235 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36236 attribute rather than comments.
36238 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36239 match_scratch operand is highest.
36241 2017-03-13 Martin Liska <mliska@suse.cz>
36243 PR middle-end/78339
36244 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36245 is a CHKP clone, use original declaration.
36247 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36249 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36250 (arc_conditional_register_usage): Use a different allocation order
36251 when optimizing for size.
36252 * common/config/arc/arc-common.c (arc_option_optimization_table):
36253 Section anchors default on when optimizing for size.
36255 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36257 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36259 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36261 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36262 * config/arc/arc.md (cpu_facility): Add cd variant.
36263 (*movqi_insn): Add code density variant.
36264 (*movhi_insn): Likewise.
36265 (*movqi_insn): Likewise.
36266 (*addsi3_mixed): Likewise.
36267 (subsi3_insn): Likewise.
36269 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36271 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36273 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36275 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36276 expressions with MINUS and UNARY ops.
36278 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36281 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36283 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36284 between vec_select and vector argument.
36285 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36286 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36287 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36288 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36289 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36290 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36292 2017-03-13 Richard Biener <rguenther@suse.de>
36295 * params.def (vect-max-peeling-for-alignment): Fix typo.
36297 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36299 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36300 issue that only occurred with binutils below 2.18.
36302 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36304 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36305 refer to binutils 2.11/2.12 minimum.
36307 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36309 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36310 ftp.kernel.org and simplify binutils requirement.
36312 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36314 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36316 (Optimize Options): Ditto. Also remove redundancy.
36318 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36320 PR translation/79848
36321 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36323 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36324 to G_ to avoid double translation.
36326 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36328 PR translation/79923
36329 * auto-profile.c (get_combined_location): Convert leading
36330 character of diagnostics to lower case and remove trailing period.
36331 (read_profile): Likewise for various diagnostics.
36332 * config/arm/arm.c (arm_option_override): Remove trailing period
36333 from various diagnostics.
36334 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36335 (msp430_expand_delay_cycles): Likewise.
36337 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36340 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36341 full command-line argument, rather than just "str".
36342 (aarch64_validate_march): Likewise.
36343 (aarch64_validate_mtune): Likewise.
36345 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36347 PR rtl-optimization/78911
36348 * lra-assigns.c (must_not_spill_p): New function.
36349 (spill_for): Use it.
36351 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36353 PR tree-optimization/79981
36354 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36355 ATOMIC_COMPARE_EXCHANGE ifn result.
36356 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36357 IFN_ATOMIC_COMPARE_EXCHANGE.
36359 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36362 * opts.c (parse_sanitizer_options): Add missing question mark to
36363 "did you mean" message.
36365 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36367 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36369 (VMULEUH_UNS): Likewise.
36370 (VMULOUB_UNS): Likewise.
36371 (VMULOUH_UNS): Likewise.
36372 * config/rs6000/rs6000.c (builtin_function_type): Remove
36373 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36375 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36378 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36379 x with result of extra_parsing_for_operand_code_0.
36380 (function_reader::extra_parsing_for_operand_code_0): Convert
36381 return type from void to rtx, returning x. When reading
36382 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36383 larger size containing struct block_symbol.
36385 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36387 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36388 -mfloat128-hardware without -m64.
36390 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36393 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36394 entries to the case statement that marks unsigned arguments to
36395 overloaded functions.
36397 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36399 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36400 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36402 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36405 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36406 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36408 2017-03-10 Martin Liska <mliska@suse.cz>
36412 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36413 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36416 2017-03-10 Olivier Hainque <hainque@adacore.com>
36418 * tree-switch-conversion (array_value_type): Start by resetting
36419 candidate type to it's main variant.
36421 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36423 PR rtl-optimization/79909
36424 * combine.c (try_combine): Use simplify_replace_rtx on individual
36425 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36426 of the whole CALL_INSN_FUNCTION_USAGE.
36428 PR tree-optimization/79972
36429 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36430 get_range_info on SSA_NAMEs. Formatting fixes.
36432 2017-03-10 Richard Biener <rguenther@suse.de>
36433 Jakub Jelinek <jakub@redhat.com>
36435 PR tree-optimization/77975
36436 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36437 edge to be constant.
36438 (get_val_for): For constant x return it. Formatting fix.
36439 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36440 would use the same bases as the current one.
36442 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36444 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36445 instead of vec_select for V1TImode.
36446 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36448 (VSX_LE_128): Add V1TI to this mode iterator.
36449 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36450 (*vsx_le_perm_store_<mode>): Likewise.
36451 (pre-reload splitter for VSX stores): Likewise.
36452 (post-reload splitter for VSX stores): Likewise.
36453 (*vsx_xxpermdi2_le_<mode>): Likewise.
36454 (*vsx_lxvd2x2_le_<mode>): Likewise.
36455 (*vsx_stxvd2x2_le_<mode>): Likewise.
36457 2017-03-09 Michael Eager <eager@eagercon.com>
36459 Correct failures with --enable-checking=yes,rtl.
36461 * config/microblaze/microblaze.c (microblaze_expand_shift):
36462 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36463 test for const0_rtx.
36464 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36465 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36467 2017-03-09 Richard Biener <rguenther@suse.de>
36469 PR tree-optimization/79977
36470 * graphite-scop-detection.c (scop_detection::merge_sese):
36471 Handle the case of extra exits to blocks dominating the entry.
36473 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36475 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36478 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36480 PR rtl-optimization/79949
36481 * lra-constraints.c (process_alt_operands): Check memory when
36482 trying to predict a cycle. Print about the overall increase.
36484 2017-03-09 Richard Biener <rguenther@suse.de>
36486 PR middle-end/79971
36487 * gimple-expr.c (useless_type_conversion_p): Preserve
36488 TYPE_SATURATING for fixed-point types.
36490 2017-03-09 Richard Biener <rguenther@suse.de>
36493 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36494 alignment of BLKmode params.
36496 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36499 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36500 (VALL_NO_V2Q): Likewise.
36502 * config/aarch64/aarch64-simd.md
36503 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36505 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36506 VALL_NO_V2Q mode iterator.
36507 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36509 2017-03-09 Martin Liska <mliska@suse.cz>
36511 PR tree-optimization/79631
36512 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36513 tree_int_cst_sign_bit just for INTEGER constants.
36515 2017-03-09 Martin Liska <mliska@suse.cz>
36519 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36522 2017-03-09 Marek Polacek <polacek@redhat.com>
36525 * tree.c (inchash::add_expr): Handle TREE_VEC.
36527 2017-03-09 Martin Liska <mliska@suse.cz>
36530 (chkp_narrow_size_and_offset): New function.
36531 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36532 (void chkp_parse_bit_field_ref): New function.
36533 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36534 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36536 2017-03-09 Martin Liska <mliska@suse.cz>
36539 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36540 (chkp_find_bounds_1): Remove gcc_unreachable.
36542 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36545 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36546 BUILT_IN_SYNC*, determine the access type from the size suffix and
36547 always build a MEM_REF with that type. Handle forgotten
36548 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36551 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36552 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36553 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36554 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36555 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36556 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36557 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36558 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36559 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36560 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36561 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36562 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36563 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36564 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36565 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36566 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36567 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36568 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36569 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36570 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36571 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36572 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36573 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36574 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36575 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36576 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36577 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36578 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36579 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36580 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36581 definitions outside of __OPTIMIZE__ guarded section.
36584 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36585 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36586 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36587 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36590 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36592 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36593 ("vfenez<mode>"): Add missing constraints.
36595 2017-03-08 Martin Sebor <msebor@redhat.com>
36598 * config/nds32/nds32.c (nds32_option_override):
36599 Fix misspelled diagnostic.
36601 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36604 * gimplify.c (gimplify_omp_for): Replace index var in outer
36605 taskloop statement with an artificial variable and add
36606 OMP_CLAUSE_PRIVATE clause for it.
36608 2017-03-08 Richard Biener <rguenther@suse.de>
36610 PR tree-optimization/79955
36611 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36612 for accesses that are completely outside of the variable.
36614 2017-03-08 Andrew Haley <aph@redhat.com>
36616 PR tree-optimization/79943
36617 * tree-ssa-loop-split.c (compute_new_first_bound): When
36618 calculating the new upper bound, (END-BEG) should be added, not
36621 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36623 * config/avr/avr.md (setmemhi): Make sure match_dup
36624 operand number comes before match_scratch.
36626 2017-03-08 Richard Biener <rguenther@suse.de>
36628 PR tree-optimization/79920
36629 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36630 with ncopies == 1 to ...
36631 (vect_transform_slp_perm_load): ... here. Properly compute
36632 all element loads by iterating VF times over the group. Do
36633 not handle ncopies (computed in a broken way) in
36634 vect_create_mask_and_perm.
36636 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36639 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36640 is a uniform vector, use uniform_vector_p return value instead of
36641 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36643 2017-03-07 Marek Polacek <polacek@redhat.com>
36645 PR middle-end/79809
36646 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36647 (alloca_call_type): Likewise.
36649 2017-03-07 Martin Liska <mliska@suse.cz>
36651 * gcov.c (process_args): Put comment to correct location.
36653 2017-03-07 Martin Liska <mliska@suse.cz>
36655 PR middle-end/68270
36656 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36657 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36658 (chkp_narrow_bounds_for_field): Likewise.
36659 (chkp_parse_array_and_component_ref): Pass one more argument to
36662 2017-03-07 Richard Biener <rguenther@suse.de>
36664 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36667 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36669 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36670 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36672 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36675 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36676 to end of description.
36677 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36679 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36681 PR rtl-optimization/79901
36682 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36684 (*avx512f_<code><mode>3<mask_name>): ... this.
36685 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36686 iterator instead of VI8_AVX2_AVX512BW.
36688 PR rtl-optimization/79901
36689 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36690 min/max expander, expand it using expand_vec_cond_expr.
36693 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36696 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36699 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36700 to void * for PCH reasons.
36701 * dwarf2out.c (output_loc_operands, output_die): Cast
36702 v.val_vec.array to unsigned char *.
36704 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36707 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36710 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36712 PR rtl-optimization/79571
36713 * lra-constraints.c (process_alt_operands): Calculate static
36714 reject and subtract it from overall when only addresses will be
36717 2017-03-06 Julia Koval <julia.koval@intel.com>
36720 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
36721 incoming stack boundary to 128 for 64-bit targets.
36723 2017-03-06 Richard Biener <rguenther@suse.de>
36725 PR tree-optimization/79894
36726 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
36727 to NULL after folding it.
36729 2017-03-06 Richard Biener <rguenther@suse.de>
36731 PR tree-optimization/79824
36732 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
36733 check disabling peeling for gaps.
36735 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
36737 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
36738 attributes): Document gettimeofday.
36740 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36742 * config/s390/s390.c (s390_option_override_internal): Set
36743 PARAM_MIN_VECT_LOOP_BOUND
36745 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36747 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36748 * config/s390/s390.md: Likewise.
36750 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36753 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36754 (<avx2_avx512>_perm<mode>): Rename to ...
36755 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36757 (<avx512>_perm<mode>_mask): Rename to ...
36758 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36760 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36761 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36762 instead of VI8F_256_512.
36763 (avx512f_perm<mode>): New define_expand.
36764 (avx512f_perm<mode>_mask): Likewise.
36765 (avx512f_perm<mode>_1<mask_name>): New define_insn.
36766 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
36768 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36770 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
36771 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
36773 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36775 2017-03-06 Martin Liska <mliska@suse.cz>
36778 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
36779 when having a SSA NAME w/o VAR_DECL assigned to it.
36781 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36783 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
36784 msa_dpsub_<su>_d): Fix MODE for vec_select.
36786 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36788 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
36790 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
36792 2017-03-06 Richard Biener <rguenther@suse.de>
36794 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
36795 * plugin.c (register_plugin_info): Likewise.
36796 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36798 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36800 * config/i386/sse.md (sse_storehps, sse_storelps,
36801 avx_<castmode><avxsizesuffix>_<castmode>,
36802 avx512f_<castmode><avxsizesuffix>_<castmode>,
36803 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36804 in condition that at least one operand is not a MEM.
36806 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36808 PR middle-end/79805
36809 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36810 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36812 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36813 gimple_call_nothrow_p flag based on whether original builtin can throw.
36814 If it can, emit following stmts on the fallthrough edge.
36815 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36816 don't create new bb if inserting just debug stmts on the edge, try to
36817 insert them on the fallthru bb or just reset debug stmts.
36819 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36822 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36823 restore recog_data (including the operand rtxes inside it) around
36824 the call to get_insn_template.
36826 2017-03-03 Martin Sebor <msebor@redhat.com>
36828 PR tree-optimization/79699
36829 * context.c (context::~context): Free MPFR caches to avoid
36830 a memory leak on program exit.
36832 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36834 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36835 Use wide_int::ulow () instead of .elt (0).
36837 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36839 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36840 (*pushxf): Limit oF constraint to 32bit targets and add oC
36841 constraint for 64bit targets.
36842 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36843 (*pushdf): Change rmF constraint to rmC.
36845 2017-03-03 Martin Liska <mliska@suse.cz>
36847 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36848 Remove unused variable.
36850 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36853 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36854 is a memory operand, increase num_memory.
36855 (ix86_expand_args_builtin): Likewise.
36857 2017-03-03 Jan Hubicka <jh@suse.cz>
36860 * ipa-devirt.c (maybe_record_node): Properly handle
36861 __cxa_pure_virtual visibility.
36863 2017-03-03 Martin Liska <mliska@suse.cz>
36865 PR tree-optimization/79803
36866 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36868 (pass_loop_prefetch::execute): Disabled optimization if an
36869 assumption about L1 cache size is not met.
36871 2017-03-03 Martin Liska <mliska@suse.cz>
36873 PR rtl-optimization/79574
36874 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36875 (hash_scan_set): Likewise.
36876 (dump_hash_table): Likewise.
36877 (hoist_code): Likewise.
36879 2017-03-03 Richard Biener <rguenther@suse.de>
36881 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36882 in place of uhigh ().
36883 (fixed_convert_from_real): Likewise.
36885 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36888 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36890 2017-03-03 Richard Biener <rguenther@suse.de>
36892 PR middle-end/79818
36893 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36894 TYPE_OVERFLOW_UNDEFINED check.
36896 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36898 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36900 (vector_ae_<mode>_p): Likewise.
36901 (vector_nez_<mode>_p): Likewise.
36902 (vector_ne_v2di_p): Likewise.
36903 (vector_ae_v2di_p): Likewise.
36904 (vector_ne_<mode>_p): Likewise.
36905 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36907 (vsx_tsqrt<mode>2_fe): Likewise.
36909 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36912 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36914 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36916 PR rtl-optimization/79780
36917 * cprop.c (one_cprop_pass): When second and further conditional trap
36918 in a single basic block is turned into an unconditional trap, turn it
36919 into a deleted note to avoid RTL verification failures.
36921 2017-03-02 Richard Biener <rguenther@suse.de>
36923 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36925 2017-03-02 Richard Biener <rguenther@suse.de>
36927 PR tree-optimization/79345
36929 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36930 param and abort the walk, returning -1 if it is hit.
36931 (walk_aliased_vdefs): Take a limit param and pass it on.
36932 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36933 defaulting to 0 and return a signed int.
36934 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36935 (check_defs): New helper.
36936 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36937 about uninitialized memory.
36938 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36939 bogus uninitialized warning.
36940 (fixed_convert_from_real): Likewise.
36942 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36944 PR tree-optimization/66768
36945 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36946 iv_use if base object can't be determined.
36948 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36950 PR tree-optimization/79345
36951 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36952 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36953 (get_pattern_stats): Initialize it.
36954 * genemit.c (gen_expand): Verify match_scratch numbers come after
36955 match_operand/match_dup numbers.
36956 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36957 match_scratch numbers.
36958 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36960 * config/s390/s390.md (trunctdsd2): Likewise.
36962 2017-03-02 Richard Biener <rguenther@suse.de>
36964 * wide-int.h (wide_int_storage::operator=): Implement in terms
36967 2017-03-02 Richard Biener <rguenther@suse.de>
36969 PR tree-optimization/79777
36970 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36971 the to insert expression to sth existing.
36973 2017-03-01 Martin Sebor <msebor@redhat.com>
36975 PR middle-end/79692
36976 * gimple-ssa-sprintf.c
36977 (directive::known_width_and_precision): New function.
36978 (format_integer): Use it.
36979 (get_mpfr_format_length): Consider the full range of precision
36980 when computing %g output with the # flag. Set the likely byte
36981 count to 3 rather than 1 when precision is indeterminate.
36982 (format_floating): Correct the lower bound of precision.
36984 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36986 * doc/invoke.texi: Document default code model for 64-bit Linux.
36988 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36991 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36992 udiv rather than div since input pattern is unsigned.
36994 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36996 * config/i386/i386.c (print_reg): Warn for values of
36997 unsupported size in integer register.
36999 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37002 * config/rs6000/predicates.md (current_file_function_operand): Do
37003 not allow self calls to be local if the function is replaceable.
37005 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37008 * config/rs6000/altivec.h (vec_ctz and others): Change the
37009 preprocessor macro that controls conditional compilation from
37010 _ARCH_PWR9 to __POWER9_VECTOR__.
37011 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37012 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37013 control (instead of _ARCH_PWR9 control) so that template
37014 definition uses power9-specific function.
37015 (vec_any_eq): Likewise.
37016 (vec_all_ne): Change macro definition to use a power9-specific
37017 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37018 _ARCH_PWR9 control).
37019 (vec_any_eq) Likewise.
37020 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37021 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37022 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37023 support for xvcmpnedp instruction.
37024 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37025 macro expansion so that Power9 implementation of vec_all_ne does
37026 not use the AltiVec predicate framework.
37027 (VCMPNEH_P): Likewise.
37028 (VCMPNEW_P): Likewise.
37029 (VCMPNED_P): Likewise.
37030 (VCMPNEFP_P): Likewise.
37031 (VCMPNEDP_P): Likewise.
37032 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37033 implementation of vec_any_eq to not use AltiVec predicate
37035 (VCMPAEH_P): Likewise.
37036 (VCMPAEW_P): Likewise.
37037 (VCMPAED_P): Likewise.
37038 (VCMPAEFP_P): Likewise.
37039 (VCMPAEDP_P): Likewise.
37040 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37041 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37042 not use the AltiVec predicate framework.
37043 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37044 of vec_any_eq to not use AltiVec predicate framework.
37045 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37046 support for predefined __POWER9_VECTOR__ macro to indicate that
37047 Power9 instruction selection is enabled.
37048 (altivec_overloaded_builtins): Remove extraneous
37049 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37050 function argument types RS6000_BTI_bool_V16QI and
37051 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37052 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37053 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37054 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37055 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37056 Power9 for implementations of vec_cmpne. Change the signature for
37057 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37058 (representing vec_all_ne) to remove the previously described first
37059 argument of type RS6000_BTI_INTSI, as this was an artifact of
37060 reliance on the AltiVec predicate framework, which is no longer
37061 used in the implementation of these functions. Add
37062 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37063 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37064 since, unlike the AltiVec predicate framework implementation, we
37065 do not share function descriptors between vec_alle and vec_anyeq.
37066 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37067 set of modes that receive special treatment even when
37068 TARGET_P9_VECTOR is true. The special treatment emits code that
37069 does not depend on Power9 instructions.
37070 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37071 define_expand to not rely on AltiVec predicate framework.
37072 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37074 (vector_ne_v2di_p): Change this define_expand to not rely on
37075 AltiVec predicate framework.
37076 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37078 (vector_ne_<mode>_p): Change this define_expand to not rely on
37079 AltiVec predicate framework.
37080 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37082 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37083 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37084 define_insn pattern.
37085 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37086 define_insn pattern because the xvcmpne<VSs>. instruction is not
37088 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37089 instruction is not supported.
37091 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37093 * config/nvptx/nvptx.c: Include intl.h.
37095 2017-03-01 Martin Jambor <mjambor@suse.cz>
37098 * ipa-prop.h (ipa_bits): Removed field known.
37099 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37100 to pointers. Adjusted their comments to warn about their sharing.
37101 (ipcp_transformation_summary): Change bits to a vector of pointers.
37102 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37103 (ipa_get_ipa_bits_for_value): Declare.
37104 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37105 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37106 (ipa_bits_hash_table): Likewise.
37107 (ipa_vr_ggc_hash_traits): Likewise.
37108 (ipa_vr_hash_table): Likewise.
37109 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37110 being pointers and vr_known being removed.
37111 (ipa_set_jf_unknown): Likewise.
37112 (ipa_get_ipa_bits_for_value): New function.
37113 (ipa_set_jfunc_bits): Likewise.
37114 (ipa_get_value_range): New overloaded functions.
37115 (ipa_set_jfunc_vr): Likewise.
37116 (ipa_compute_jump_functions_for_edge): Use the above functions to
37117 construct bits and vr parts of jump functions.
37118 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37119 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37121 (ipcp_grow_transformations_if_necessary): Also allocate
37122 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37124 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37125 them. Fix too long lines.
37126 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37127 vr_known being removed.
37128 (ipa_read_jump_function): Use new setter functions to construct bits
37129 and vr parts of jump functions or set them to NULL.
37130 (write_ipcp_transformation_info): Adjust for bits being pointers.
37131 (read_ipcp_transformation_info): Likewise.
37132 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37134 Include gt-ipa-prop.h.
37135 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37137 (ipcp_store_bits_results): Likewise.
37138 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37139 Do not write to existing jump functions but use a temporary instead.
37141 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37144 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37145 attempt to use its first operand as BIT_FIELD_REF base.
37147 2017-03-01 Richard Biener <rguenther@suse.de>
37149 PR middle-end/79721
37150 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37151 interpolating formula in wrapping arithmetic.
37152 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37154 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37156 PR tree-optimization/79734
37157 * tree-vect-generic.c (expand_vector_condition): Optimize
37158 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37159 Handle VEC_COND_EXPR where comparison has different inner width from
37160 type's inner width.
37162 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37164 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37165 markup, and similar issues. Remove @opindex entries for things
37166 that aren't options. Add missing -mmpy-option entries.
37168 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37170 PR tree-optimization/79737
37171 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37172 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37173 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37174 instead of byte_size. Formatting fix.
37175 (shift_bytes_in_array_right): Formatting fix.
37177 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37180 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37181 condition on optimize for the leaf function test.
37183 2017-02-28 Martin Liska <mliska@suse.cz>
37186 * read-rtl-function.c (function_reader::handle_unknown_directive):
37187 Bail out when one uses -flto.
37189 2017-02-28 Martin Liska <mliska@suse.cz>
37191 * common.opt: Replace space with tabular for options of <number>
37193 * config/i386/i386.opt: Show <number> value for
37194 -mlarge-data-threshold.
37195 * opts.c (print_filtered_help): Do not display number in hexadecimal
37198 2017-02-28 Martin Liska <mliska@suse.cz>
37200 * common.opt: Fix --help=option -Q for options which are of
37203 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37205 * config/i386/i386.c (print_reg): Error out for values
37206 of 8-bit size in invalid integer register.
37208 2017-02-28 Martin Sebor <msebor@redhat.com>
37210 PR tree-optimization/79691
37211 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37213 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37216 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37217 gcc_unreachable with output_operand_lossage.
37219 2017-02-28 Richard Biener <rguenther@suse.de>
37221 PR tree-optimization/79740
37222 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37224 (visit_nary_op): Insert the nary into the hashtable if we
37225 pattern-matched sth.
37226 * tree-ssa-pre.c (eliminate_insert): Robustify.
37228 2017-02-28 Richard Biener <rguenther@suse.de>
37230 PR middle-end/79731
37231 * fold-const.c (decode_field_reference): Reject out-of-bound
37234 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37236 * config/i386/i386.c: Include intl.h.
37237 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37238 instead of just cond ? "..." : "...".
37239 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37240 * coverage.c (read_counts_file): Likewise.
37241 * omp-offload.c: Include intl.h.
37242 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37243 of just cond ? "..." : "...".
37244 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37245 of just cond ? "..." : "...".
37247 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37250 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37252 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37253 'tune for' CPU name.
37254 * config/arm/arm-cpu-data.h: Regenerated.
37256 2017-02-28 Richard Biener <rguenther@suse.de>
37258 PR tree-optimization/79732
37259 * tree-inline.c (expand_call_inline): Do not shadow var.
37261 2017-02-28 Richard Biener <rguenther@suse.de>
37263 PR tree-optimization/79723
37264 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37265 address-space properly.
37267 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37269 * doc/optinfo.texi (Optimization groups): Fix option used for
37271 * doc/invoke.texi (-fopt-info): Document "omp".
37272 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37273 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37274 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37275 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37276 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37278 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37280 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37281 OPTGROUP_OMP with "omp".
37283 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37286 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37287 for arithmetic shift of unsigned V2DI.
37289 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37291 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37292 arc/linux.h headers.
37293 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37294 (LINK_SPEC): Likewise.
37295 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37296 (EXTRA_SPECS): Likewise.
37297 (STARTFILE_SPEC): Likewise.
37298 (ENDFILE_SPEC): Likewise.
37299 (LIB_SPEC): Likewise.
37300 (TARGET_SDATA_DEFAULT): Likewise.
37301 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37302 (MULTILIB_DEFAULTS): Likewise.
37303 (DWARF2_UNWIND_INFO): Likewise.
37304 * config/arc/big.h: New file.
37305 * config/arc/elf.h: Likewise.
37306 * config/arc/linux.h: Likewise.
37307 * config/arc/t-uClibc: Remove.
37309 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37311 PR tree-optimization/77536
37312 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37313 (tree_transform_and_unroll_loop): Use above function to compute the
37314 estimated niter of unrolled loop and use it when scaling profile.
37315 Also use count info rather than frequency if it's non-zero.
37316 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37317 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37318 (vect_transform_loop): Call above function.
37320 2017-02-27 Richard Biener <rguenther@suse.de>
37322 PR tree-optimization/45397
37323 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37324 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37325 (visit_nary_op): Add pattern matching for CSEing sign-changed
37326 or truncated operations with wider ones.
37328 2017-02-27 Richard Biener <rguenther@suse.de>
37330 PR tree-optimization/79690
37331 * tree-vect-stmts.c (vectorizable_store): Use vector type
37332 built from the DR with address-space.
37334 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37336 * doc/invoke.texi (Optimize Options): Refine the description
37337 of asan-use-after-return.
37339 2017-02-25 Alan Modra <amodra@gmail.com>
37341 PR rtl-optimization/79584
37342 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37343 base, not ad->base_term, the reg within base. Remove assertion
37344 that ad->base == ad->base_term. Replace gen_int_mode using
37345 bogus mode with const0_rtx.
37347 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37349 PR middle-end/79396
37350 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37351 FMA_EXPR like tcc_binary or tcc_unary.
37353 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37356 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37358 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37359 (output_loc_operands): Handle DW_OP_call_ref and
37360 DW_OP_GNU_variable_value.
37361 (struct variable_value_struct): New type.
37362 (struct variable_value_hasher): Likewise.
37363 (variable_value_hash): New variable.
37364 (string_types): Remove.
37365 (copy_loc_descr): New function.
37366 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37367 (prepend_loc_descr_to_each): New function.
37368 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37369 instead of add_loc_descr_to_each if the first argument is single
37370 location list and the second has multiple.
37371 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37372 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37373 when looking for variable value which doesn't have other location info.
37374 (loc_list_from_tree): Formatting fix.
37375 (gen_array_type_die): Simplify DW_AT_string_length handling.
37376 (adjust_string_types): Remove.
37377 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37378 string_types. Call resolve_variable_values.
37379 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37380 (resolve_addr_in_expr): Likewise. Add A argument.
37381 (copy_deref_exprloc): Remove deref argument. Adjust for the
37382 original expression being DW_OP_GNU_variable_value with optionally
37383 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37384 optionally after it.
37385 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37386 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37387 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37388 (variable_value_hasher::hash, variable_value_hasher::equal): New
37390 (resolve_variable_value_in_expr, resolve_variable_value,
37391 resolve_variable_values, note_variable_value_in_expr,
37392 note_variable_value): New functions.
37393 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37396 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37399 * opts.h (handle_generated_option): Add GENERATED_P argument.
37400 * opts-common.c (handle_option): Adjust function comment.
37401 (handle_generated_option): Add GENERATED_P argument, pass it to
37403 (control_warning_option): Pass false to handle_generated_option
37405 * opts.c (maybe_default_option): Pass true to handle_generated_option
37407 * optc-gen.awk: Likewise.
37409 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37411 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37412 a REG, look at the REG it is a SUBREG of.
37413 (splitter for cmpeqsi_t): Ditto.
37415 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37417 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37418 the special USEs with the pattern of the insn, not the insn itself.
37420 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37423 * doc/invoke.texi: Document -mload-store-pairs.
37425 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37426 Sandra Loosemore <sandra@codesourcery.com>
37428 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37429 argument isn't a CONST_INT.
37430 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37431 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37432 (nios2_validate_compare): Bypass alternate compare logic if *op2
37433 is not a CONST_INT.
37434 (ldstwm_operation_p): Return false if first_base is not a REG or
37435 if first_offset is not a CONST_INT.
37437 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37439 * config/cris/cris.md: Use correct operand in a define_peephole2.
37441 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37443 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37445 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37447 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37448 this_insn if it is an INSN or JUMP_INSN.
37449 (force_offsettable): Look at base, not at addr.
37450 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37451 on things that aren't necessarily CONST_INTs.
37453 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37455 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37456 -mfpmath=sse is the default also for x86-32 targets with SSE2
37457 instruction set when @option{-ffast-math} is enabled
37459 2017-02-24 Jeff Law <law@redhat.com>
37461 PR rtl-optimizatoin/79286
37462 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37465 2017-02-24 Richard Biener <rguenther@suse.de>
37467 PR tree-optimization/79389
37468 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37471 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37473 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37474 function comment to reflect reality.
37475 (loop_exits_before_overflow): Fix typo in function description.
37477 2017-02-24 Richard Biener <rguenther@suse.de>
37479 PR tree-optimization/79389
37480 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37481 properly that a threading opportunity exists. Detect conditional
37482 copy/constant propagation opportunities.
37484 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37486 * config/visium/visium.md (type): Add trap.
37487 (b): New mode attribute.
37488 (*btst): Rename into...
37489 (*btst<mode>): ...this and adjust.
37490 (*cbranchsi4_btst_insn): Rename into...
37491 (*cbranch<mode>4_btst_insn): ...this and adjust.
37492 (trap): New define_insn.
37494 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37496 PR tree-optimization/79389
37497 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37498 (noce_reversed_cond_code): New function.
37499 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37500 reversed_comparison_code. Formatting fix.
37501 (noce_try_store_flag): Test rev_cond != NULL in addition to
37502 reversed_comparison_code.
37503 (noce_try_store_flag_constants): Likewise.
37504 (noce_try_store_flag_mask): Likewise.
37505 (noce_try_addcc): Use rev_cond if non-NULL instead of
37506 reversed_comparison_code.
37507 (noce_try_cmove_arith): Likewise. Formatting fixes.
37508 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37509 (noce_find_if_block): Initialize rev_cond.
37510 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37511 instead of false as last argument never attempt to reverse it
37514 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37516 PR tree-optimization/79663
37517 * tree-predcom.c (combine_chains): Process refs in reverse order
37518 only for ZERO length chains, and add explaining comment.
37520 2017-02-23 Jeff Law <law@redhat.com>
37522 PR tree-optimization/79578
37523 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37524 in call to operand_equal_p.
37526 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37529 * config/i386/cpuid.h: Fix another undefined behavior.
37531 2017-02-23 Richard Biener <rguenther@suse.de>
37533 PR tree-optimization/79683
37534 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37535 vector types for data-refs.
37537 2017-02-23 Martin Liska <mliska@suse.cz>
37539 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37541 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37543 PR middle-end/79665
37544 * internal-fn.c (get_range_pos_neg): Moved to ...
37545 * tree.c (get_range_pos_neg): ... here. No longer static.
37546 * tree.h (get_range_pos_neg): New prototype.
37547 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37548 are known to be in between 0 and signed maximum inclusive, try to
37549 expand both unsigned and signed divmod and use the cheaper one from
37552 2017-02-22 Jeff Law <law@redhat.com>
37554 PR tree-optimization/79578
37555 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37556 to compare base operands.
37558 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37561 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37562 gpc_reg_operand instead of fpr_reg_operand.
37564 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37566 * config/mips/mips.c (mips_return_in_memory): Force FP
37567 vector types to be returned in memory for o32 ABI.
37569 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37571 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37572 instead of DW_TAG_member for static data member declarations and don't
37573 set no_linkage_name for static inline data members.
37574 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37577 2017-02-22 Martin Liska <mliska@suse.cz>
37579 * doc/invoke.texi: Replace inequality signs with square brackets
37582 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37585 * lra-constraints.c (simplify_operand_subreg): Handle
37586 WORD_REGISTER_OPERATIONS targets.
37588 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37591 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37592 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37593 elimination by swapping fld*.
37595 2017-02-22 Richard Biener <rguenther@suse.de>
37597 PR tree-optimization/79673
37598 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37599 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37600 irrelevant address-space qualifiers and avoiding a
37601 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37603 2017-02-22 Richard Biener <rguenther@suse.de>
37605 PR tree-optimization/79666
37606 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37607 to not symbolically negate if that may introduce undefined
37610 2017-02-22 Martin Liska <mliska@suse.cz>
37613 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37614 * data-streamer-out.c (streamer_write_gcov_count_stream):
37616 * value-prof.c (stream_out_histogram_value): Make assert more
37617 precise based on type of counter.
37619 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37622 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37623 Use nonimmediate_operand instead of memory_operand for operand 1.
37624 (float-extend standard_x87sse_constant_load splitter): Ditto.
37626 2017-02-21 Jeff Law <law@redhat.com>
37628 PR tree-optimization/79621
37629 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37630 blocks with edges to themselves.
37632 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37635 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37636 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37637 Use gimple_call_builtin_p.
37640 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37641 on temporarily removed DEBUG_INSNs.
37643 PR tree-optimization/79649
37644 * tree-loop-distribution.c (classify_partition): Give up on
37645 non-generic address space loads/stores.
37647 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37649 * doc/loop.texi (Loop manipulation): Remove nonexistent
37650 tree_ssa_loop_version from the documentation.
37651 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37653 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37656 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37657 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37658 * config/rs6000/rs6000.c: Include except.h.
37659 (rs6000_expand_split_stack_prologue): Call
37660 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37662 2017-02-21 Martin Jambor <mjambor@suse.cz>
37665 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37666 have been analyzed.
37668 2017-02-21 Martin Jambor <mjambor@suse.cz>
37670 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37671 for backward compatibility only.
37672 * doc/invoke.texi (Option Summary): Remove all references to
37673 -fipa-cp-alignment.
37675 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37679 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37681 * lra-constraints.c (curr_insn_transform): Handle
37682 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37684 2017-02-21 Martin Liska <mliska@suse.cz>
37686 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37688 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37690 PR translation/79638
37691 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37693 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37696 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37697 (arm_function_ok_for_sibcall): Return false for an indirect call by
37698 descriptor if all the argument registers are used.
37699 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37700 alignment of the function.
37702 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37704 PR tree-optimization/61441
37705 * simplify-rtx.c (simplify_const_unary_operation): For
37706 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37707 the sNaN unmodified.
37709 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37711 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37712 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
37713 instead of SYSTEM_HEADER_DIR.
37715 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
37716 Martin Liška <mliska@suse.cz>
37718 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
37719 Fix typos and grammar, use active voice, and clarify.
37721 2017-02-20 Marek Polacek <polacek@redhat.com>
37723 PR middle-end/79537
37724 * gimplify.c (gimplify_expr): Handle unused *&&L;.
37727 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
37729 2017-02-20 Jakub Jelinek <jakub@redhat.com>
37732 * config/i386/i386.c (ix86_expand_builtin): Handle
37733 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
37734 ix86_builtins_isa[fcode].isa as a requirement of those
37735 flags and any other flag in the bitmask.
37736 (ix86_init_mmx_sse_builtins): Use 0 instead of
37737 ~OPTION_MASK_ISA_64BIT as mask.
37738 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
37739 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
37740 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37741 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37743 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37746 * lra-constraints.c (split_reg): Check requested split mode
37747 is supported by the register.
37749 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37751 * lra-constraints.c (simplify_operand_subreg): Remove early
37754 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37757 * lra-constraints.c (curr_insn_transform): Tighten condition
37758 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37760 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37763 * lra-constraints.c (curr_insn_transform): Handle
37764 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37766 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
37769 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
37771 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
37773 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37776 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
37779 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37781 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
37784 2017-02-18 Jakub Jelinek <jakub@redhat.com>
37786 * final.c (last_columnnum, override_columnnum): New variables.
37787 (final_start_function): Set last_columnnum, pass it to begin_prologue
37788 hook and pass 0 to dwarf2out_begin_prologue.
37789 (final_scan_insn): Update override_columnnum. Pass last_columnnum
37790 to source_line debug hook.
37791 (notice_source_line): Compute last_columnnum and for debug_column_info
37792 return true on column changes.
37793 * debug.h (struct gcc_debug_hooks): Add column argument to
37794 source_line and begin_prologue hooks.
37795 (debug_nothing_int_charstar_int_bool): Remove prototype.
37796 (debug_nothing_int_int_charstar,
37797 debug_nothing_int_int_charstar_int_bool): New prototypes.
37798 (dwarf2out_begin_prologue): Add column argument.
37799 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37800 begin_prologue hooks.
37801 (debug_nothing_int_charstar_int_bool): Remove.
37802 (debug_nothing_int_int_charstar,
37803 debug_nothing_int_int_charstar_int_bool): New functions.
37804 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37805 through to dwarf2out_source_line.
37806 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37807 (dwarf2out_source_line): Add column argument, emit it if requested.
37808 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37810 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37811 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37812 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37813 through to dwarf2out_begin_prologue.
37814 (vmsdbgout_source_line): Add column argument, pass it through to
37815 dwarf2out_source_line.
37816 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37817 dbxout_source_line caller.
37818 (dbxout_source_line): Add column argument.
37820 * common.opt (gno-column-info, gcolumn-info): New options.
37821 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37822 (check_die): Also test for multiple DW_AT_decl_column attributes.
37823 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37824 DW_AT_decl_column if requested.
37825 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37827 (gen_variable_die): Likewise.
37828 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37829 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37832 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37833 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37834 (ix86_handle_option): Handle OPT_m3dnowa.
37835 * doc/invoke.texi (-m3dnowa): Document.
37836 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37837 -m3dnowa instead of -m3dnow -march=athlon.
37840 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37841 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37843 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37846 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37847 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37848 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37849 generator for vsx_xxpermdi_<mode>_be.
37850 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37851 force big-endian semantics.
37852 (vsx_xxpermdi_<mode>_be): New define_expand with same
37853 implementation as previous version of vsx_xxpermdi_<mode>.
37855 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37857 PR tree-optimization/79327
37858 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37859 variable, its initialization and use.
37861 2017-02-17 Julia Koval <julia.koval@intel.com>
37863 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37864 (OPTION_MASK_ISA_PKU_UNSET): New.
37865 (ix86_handle_option): Handle -mrdpid.
37866 * config/i386/cpuid.h (bit_RDPID): New.
37867 * config/i386/driver-i386.c (host_detect_local_cpu):
37868 Detect RDPID feature.
37869 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37870 * config/i386/i386-c.c (ix86_target_macros_internal):
37872 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37873 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37874 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37875 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37876 * config/i386/i386.md (define_insn "rdpid"): New.
37877 * config/i386/i386.opt Add -mrdpid.
37878 * config/i386/immintrin.h (_rdpid_u32): New.
37880 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37882 PR rtl-optimization/79541
37883 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37884 instead of transforming it into USE.
37886 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37888 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37889 If HONOR_SNANS (SFmode) force the input to a register.
37890 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37891 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37892 an frsp or similar insn.
37894 2017-02-17 Martin Liska <mliska@suse.cz>
37896 PR rtl-optimization/79577
37897 * params.def (selsched-max-sched-times): Increase minimum to 1.
37899 2017-02-17 Martin Liska <mliska@suse.cz>
37901 PR rtl-optimization/79574
37902 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37904 2017-02-17 Martin Liska <mliska@suse.cz>
37906 PR tree-optimization/79529
37907 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37908 ssa_defined_default_def_p to handle cases which are implicitly
37910 * tree-ssa.c (ssa_defined_default_def_p): New function.
37911 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37912 which are implicitly defined.
37913 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37915 2017-02-17 Richard Biener <rguenther@suse.de>
37917 PR middle-end/79576
37918 * params.def (max-ssa-name-query-depth): Limit to 10.
37920 2017-02-17 Richard Biener <rguenther@suse.de>
37922 PR tree-optimization/79552
37923 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37926 2017-02-17 Richard Biener <rguenther@suse.de>
37929 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37931 2017-02-17 Marek Polacek <polacek@redhat.com>
37933 PR middle-end/79536
37934 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37935 (fold_negate_expr): New wrapper.
37937 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37939 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37940 Correct terminology and de-emphasize pre-standard behavior.
37942 2017-02-16 Alan Modra <amodra@gmail.com>
37944 PR rtl-optimization/79286
37945 * ira.c (def_dominates_uses): New function.
37946 (update_equiv_regs): Don't create an equivalence for insns that
37947 may trap where the register def does not dominate the use.
37949 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37951 PR rtl-optimization/78127
37952 * lra.c (lra): Call lra_eliminate before finish the loop after
37955 2017-02-16 Richard Biener <rguenther@suse.de>
37957 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37959 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37960 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37961 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37962 (isl_val_int_from_wi): New function.
37963 (extract_affine_gmp): Rename to ...
37964 (extract_affine_wi): ... this, take a widest_int.
37965 (extract_affine_int): Just wrap extract_affine_wi.
37966 (add_param_constraints): Use isl_val_int_from_wi.
37967 (add_loop_constraints): Likewise, and extract_affine_wi.
37969 2017-02-15 Jeff Law <law@redhat.com>
37971 PR middle-end/79521
37972 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37973 ira_init_register_move_cost_if_necessary.
37975 2017-02-15 Martin Sebor <msebor@redhat.com>
37977 PR middle-end/32003
37978 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37979 removed in a prior commit.
37981 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37983 PR tree-optimization/79347
37984 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37985 counters during peeling.
37987 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37989 * Makefile.in (site.exp): Remove "set ISLVER".
37991 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37994 * real.c (real_from_integer): Call real_convert even for decimal.
37996 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37999 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38001 2017-02-14 Andrew Pinski <apinski@cavium.com>
38003 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38004 cores and change the partno/implementer to be correct.
38005 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38006 the 'B" as the implementer.
38007 * config/aarch64/aarch64-tune.md: Regenerate.
38009 2017-02-14 Carl Love <cel@us.ibm.com>
38011 * config/rs6000/rs6000.c: Add case statement entry to make the
38012 xvcvuxdsp built-in argument unsigned.
38013 * config/rs6000/vsx.md: Fix the source and return operand types so they
38014 match the instruction definitions from the ISA document. Fix typo
38015 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38018 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38021 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38022 member early_clobber_alts.
38023 * lra-lives.c (reg_early_clobber_p): New.
38024 (process_bb_lives): Use it.
38025 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38026 (debug_operand_data): Initialize early_clobber_alts.
38027 (setup_operand_alternative): Set up early_clobber_alts.
38028 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38029 alternatives to new_insn_reg.
38030 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38032 (lra_update_insn_regno_info): Pass the new arg.
38034 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38036 PR middle-end/79505
38037 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38038 (new_oacc_loop_raw): Don't clear already cleared fields.
38041 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38042 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38043 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38045 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38048 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38050 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38053 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38054 the extra instruction to the right place to store 128-bit constant
38057 2017-02-14 Martin Sebor <msebor@redhat.com>
38059 PR middle-end/79448
38060 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38061 warning for strings of unknown length.
38063 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38065 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38067 2017-02-14 Jeff Law <law@redhat.com>
38070 * ira-costs.c (scan_one_insn): Initialize register move costs
38071 for pseudos seen in USE/CLOBBER insns.
38073 PR tree-optimization/79095
38074 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38075 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38076 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38077 if the operands are known to be not equal, then the resulting range
38079 (intersect_ranges): If the new range is ~[0,0] and the old range is
38080 wide, then prefer ~[0,0].
38081 * tree-vrp.c (overflow_comparison_p_1): New function.
38082 (overflow_comparison_p): New function.
38083 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38084 if NAME is used in an overflow test.
38085 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38086 overflow check that can be expressed as an equality test, then adjust
38087 ops to be that equality test.
38089 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38091 * config/s390/s390-builtin-types.def: Remove flags argument.
38092 * config/s390/s390.c (s390_init_builtins): Likewise.
38094 2017-02-14 Martin Liska <mliska@suse.cz>
38096 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38097 vector. Fix trailing white spaces.
38099 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38101 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38104 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38106 PR rtl-optimization/68664
38107 * config/arm/arm.c (arm_sched_can_speculate_insn):
38108 New function. Declare prototype.
38109 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38111 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38113 PR rtl-optimization/68664
38114 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38116 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38118 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38120 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38121 max skip bytes for function, loop and jump.
38123 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38125 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38126 ABS_EXPR for gimple dump.
38128 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38131 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38133 PR tree-optimization/79408
38134 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38135 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38136 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38137 also if rhs1 is INTEGER_CST.
38139 2017-02-14 Richard Biener <rguenther@suse.de>
38141 PR middle-end/79432
38142 * tree-into-ssa.c (insert_phi_nodes): When the function can
38143 have abnormal edges rewrite SSA names with broken use-def
38144 dominance out of SSA and register them for PHI insertion.
38146 2017-02-13 Martin Sebor <msebor@redhat.com>
38148 PR middle-end/79496
38149 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38150 clearing info.nowrite flag when snprintf size argument is a range.
38152 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38154 * cprop.c (cprop_jump): Add missing space in string literal.
38155 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38156 (get_constraint_for_component_ref): Likewise.
38157 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38158 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38159 * lra-constraints.c (process_alt_operands): Likewise.
38160 * ipa-inline.c (inline_small_functions): Likewise.
38161 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38162 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38163 * trans-mem.c (diagnose_tm_1_op): Likewise.
38164 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38165 (grid_parallel_clauses_gridifiable): Likewise.
38167 * config/nvptx/mkoffload.c (process): Add space in between
38170 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38171 "MOD4_SSE_REGS" and "ALL_REGS".
38173 * spellcheck.c (test_data): Add , in between "foo" and "food".
38175 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38178 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38179 boundary crossing check and subsequent code generation agree.
38181 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38183 * config/aarch64/aarch64.c (has_memory_op): Delete.
38184 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38187 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38189 PR rtl-optimization/79388
38190 PR rtl-optimization/79450
38191 * combine.c (distribute_notes): When removing TEM_INSN for which
38192 corresponding dest has last value recorded, invalidate that last
38195 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38197 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38198 of explicit '@'. Add missing assembly comment marker on branch costs
38201 2017-02-13 Nathan Sidwell <nathan@acm.org>
38203 * gengtype-lex.l (<in_struct>): Add '/'.
38205 2017-02-13 Martin Liska <mliska@suse.cz>
38208 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38210 2017-02-13 Richard Biener <rguenther@suse.de>
38212 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38214 * configure: Re-generate.
38215 * config.in: Likewise.
38216 * graphite-dependences.c: Simplify as if
38217 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38218 * graphite-isl-ast-to-gimple.c: Likewise.
38219 * graphite-optimize-isl.c: Likewise.
38220 * graphite-poly.c: Likewise.
38221 * graphite-sese-to-poly.c: Likewise.
38222 * graphite.h: Likewise.
38223 * toplev.c: Include isl/version.h and use isl_version () for
38224 printing the ISL version.
38225 * doc/install.texi: Update ISL requirement.
38227 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38229 * doc/standards.texi (Standards): Update reference to
38232 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38234 * doc/extend.texi (Named Address Spaces): sourceware.org now
38236 * doc/install.texi (Binaries): Ditto.
38239 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38241 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38242 standard terminology "stringize"/"stringizing" throughout.
38243 * doc/cppinternals.texi: Likewise.
38245 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38247 * doc/extend.texi: Fix some spelling mistakes and typos.
38248 * doc/invoke.texi: Likewise.
38250 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38253 * params.def (inline-min-speedup) Change from 10 to 8.
38255 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38257 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38260 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38263 * ipa-inline-analysis.c (get_minimal_bb): New function.
38264 (record_modified): Use it.
38265 (remap_edge_change_prob): Handle also ancestor functions.
38267 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38269 * doc/contrib.texi (Contributors): Remove broken link into
38270 the Mauve CVS repository.
38272 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38274 PR middle-end/79454
38275 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38276 result computation whenever lhs doesn't have vector mode, not
38277 just when it has BLKmode.
38279 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38281 * doc/makefile.texi (profiledbootstrap): Refer to the
38282 installation instructions only in textual form.
38284 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38287 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38289 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38291 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38292 (Specific): Update mingw-w64 reference.
38294 (Specific): Remove broken link to Renesas RX processor.
38296 2017-02-10 Richard Biener <rguenther@suse.de>
38298 * toplev.c (process_options): Do not mention obsolete graphite
38299 options when printing sorry message about missing graphite support.
38300 Mention -floop-nest-optimize.
38302 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38304 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38305 (vtst_p16): Likewise.
38306 (vtstq_p8): Likewise.
38307 (vtstq_p16): Likewise.
38309 (vtstq_p64): Likewise.
38310 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38311 (vset_lane_p64): New.
38312 (vsetq_lane_p64): New.
38314 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38316 PR tree-optimization/79411
38317 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38318 stmt operands are SSA_NAMEs used in abnormal phis.
38319 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38322 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38325 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38328 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38330 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38332 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38334 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38338 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38339 not arbitrary TREE_CONSTANT.
38342 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38343 "omp declare target link" attribute unless is_global_var.
38344 * omp-offload.c (find_link_var_op): Likewise.
38346 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38347 Chung-Lin Tang <cltang@codesourcery.com>
38349 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38351 (gimplify_adjust_omp_clauses): Don't delete TILE.
38352 (gimplify_omp_for): Deal with TILE.
38353 * internal-fn.c (expand_GOACC_TILE): New function.
38354 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38356 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38357 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38359 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38360 avoid DIV for outermost collapse var.
38361 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38362 Remove out of date comments, fix whitespace.
38363 * omp-general.c (omp_extract_for_data): Deal with tiling.
38364 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38365 adjust OLF_DIM_BASE value.
38366 (struct omp_for_data): Add tiling field.
38367 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38368 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38369 for auto loops. Remove default auto determining, moved to
38370 oacc_loop_fixed_partitions.
38371 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38372 stmts, add e_mask field.
38373 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38374 (oacc_thread_numbers): Use oacc_dim_call.
38375 (oacc_xform_tile): New.
38376 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38377 (finish_oacc_loop): Adjust for ifns vector.
38378 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38379 add case for GOACC_TILE fns.
38380 (oacc_loop_xform_loop): Delete.
38381 (oacc_loop_process): Iterate over call list directly, and add
38382 handling for GOACC_TILE fns.
38383 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38385 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38386 vector partitioning to outer loops. Assign 2 partitions to loops
38387 when available. Add TILE handling.
38388 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38389 (execite_oacc_device_lower): Process GOACC_TILE fns,
38390 ignore unknown specs.
38391 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38392 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38393 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38395 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38397 * configure.ac (ACX_BUGURL): Update.
38398 * configure: Regenerate.
38400 2017-02-09 Richard Biener <rguenther@suse.de>
38402 PR tree-optimization/69823
38403 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38404 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38406 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38408 * config/arc/arc-c.def: Add __NPS400__ definition.
38409 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38410 (TARGET_NPS400): Define.
38412 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38414 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38416 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38417 pointer, arch_info.
38418 (arc_cpu_types): Fill the arch_info field with a pointer into the
38419 arc_arch_types table.
38420 (arc_selected_cpu): Declare.
38421 * config/arc/arc.c (arc_selected_cpu): Make global.
38422 (arc_selected_arch): Delete.
38423 (arc_base_cpu): Delete.
38424 (arc_override_options): Remove references to deleted variables,
38425 update access to arch information.
38426 (ARC_OPT): Update access to arch information.
38427 (ARC_OPTX): Likewise.
38428 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38429 (TARGET_ARC600): Update access to arch information.
38430 (TARGET_ARC601): Likewise.
38431 (TARGET_ARC700): Likewise.
38432 (TARGET_EM): Likewise.
38433 (TARGET_HS): Likewise.
38434 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38437 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38440 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38441 condition/operands for integer GE/LE/GEU/LEU operations.
38443 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38445 PR translation/79397
38446 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38449 2017-02-08 Martin Jambor <mjambor@suse.cz>
38452 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38453 whether allocation happened.
38454 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38455 nothing was allocated.
38457 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38459 PR tree-optimization/79408
38460 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38461 constant, but SSA_NAME with a known integer range, use the minimum
38462 of that range instead of op1 to determine if modulo can be replaced
38463 with its first operand.
38465 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38467 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38469 2017-02-08 Richard Biener <rguenther@suse.de>
38471 PR tree-optimization/71824
38472 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38473 Check all loops contained in the merged region.
38475 2017-02-07 Andrew Pinski <apinski@cavium.com>
38477 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38479 2017-02-07 Andrew Pinski <apinski@cavium.com>
38481 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38482 (thunderxt88): Likewise.
38483 (thunderxt81): Disable LSE and change v8.1 to v8.
38484 (thunderxt83): Likewise.
38486 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38487 Richard Biener <rguenther@suse.de>
38489 PR middle-end/79399
38490 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38491 type from int to size_t.
38492 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38494 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38496 PR rtl-optimization/79386
38497 * cprop.c (bypass_conditional_jumps): Initialize
38498 bypass_last_basic_block already before splitting bbs after
38499 unconditional traps...
38500 (bypass_conditional_jumps): ... rather than here.
38503 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38504 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38505 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38506 fix -masm=intel patterns.
38508 2017-02-07 Richard Biener <rguenther@suse.de>
38510 PR tree-optimization/79256
38511 PR middle-end/79278
38512 * builtins.c (get_object_alignment_2): Use min_align_of_type
38513 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38514 and ADJUST_FIELD_ALIGN.
38516 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38518 * doc/tm.texi: Regenerate.
38519 * stor-layout.c (layout_decl): Adjust.
38520 (update_alignment_for_field): Likewise.
38521 (place_field): Likewise.
38522 (min_align_of_type): Likewise.
38523 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38524 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38525 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38526 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38527 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38528 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38529 * config/i386/i386.c (x86_field_alignment): Likewise.
38530 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38531 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38532 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38533 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38534 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38535 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38539 2017-01-30 Richard Biener <rguenther@suse.de>
38541 PR tree-optimization/79256
38542 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38543 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38546 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38548 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38549 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38550 builtins to SImode and emit a zero-extend, if necessary.
38552 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38554 * docs/invoke.texi (RISC-V Options): Alphabetize.
38556 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38558 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38561 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38563 * config/riscv/riscv.c: New file.
38564 * common/config/riscv/riscv-common.c: Likewise.
38565 * config.gcc: Likewise.
38566 * config/riscv/constraints.md: Likewise.
38567 * config/riscv/elf.h: Likewise.
38568 * config/riscv/generic.md: Likewise.
38569 * config/riscv/linux.h: Likewise.
38570 * config/riscv/multilib-generator: Likewise.
38571 * config/riscv/peephole.md: Likewise.
38572 * config/riscv/pic.md: Likewise.
38573 * config/riscv/predicates.md: Likewise.
38574 * config/riscv/riscv-builtins.c: Likewise.
38575 * config/riscv/riscv-c.c: Likewise.
38576 * config/riscv/riscv-ftypes.def: Likewise.
38577 * config/riscv/riscv-modes.def: Likewise.
38578 * config/riscv/riscv-opts.h: Likewise.
38579 * config/riscv/riscv-protos.h: Likewise.
38580 * config/riscv/riscv.h: Likewise.
38581 * config/riscv/riscv.md: Likewise.
38582 * config/riscv/riscv.opt: Likewise.
38583 * config/riscv/sync.md: Likewise.
38584 * config/riscv/t-elf-multilib: Likewise.
38585 * config/riscv/t-linux: Likewise.
38586 * config/riscv/t-linux-multilib: Likewise.
38587 * config/riscv/t-riscv: Likewise.
38588 * configure.ac: Likewise.
38589 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38590 Waterman as RISC-V maintainers.
38591 * doc/install.texi: Add RISC-V entries.
38592 * doc/invoke.texi: Add RISC-V options section.
38593 * doc/md.texi: Add RISC-V constraints section.
38594 * configure: Regenerated.
38596 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38599 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38600 false values to be constant vectors with all 0 or all 1 bits set.
38601 (vcondu<mode><mode>): Likewise.
38602 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38604 (fpmask_comparison_operator): Update comment.
38605 (vecint_comparison_operator): New predicate.
38606 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38607 vector conditionals when the true and false values are constant
38608 vectors with all 0 bits or all 1 bits set.
38610 2017-02-06 Martin Sebor <msebor@redhat.com>
38612 PR tree-optimization/79376
38613 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38615 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38617 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38618 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38619 to simplify split condition.
38621 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38623 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38624 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38627 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38629 PR rtl-optimization/68664
38630 * target.def (can_speculate_insn): New hook.
38631 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38632 * doc/tm.texi: Regenerate.
38633 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38634 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38635 (rs6000_sched_can_speculate_insn): New function.
38637 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38639 PR tree-optimization/79284
38640 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38641 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38642 vectorizable_mask_load_store, vectorizable_operation,
38643 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38644 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38645 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38646 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38647 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38648 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38649 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38650 is_gimple_assign (stmt). Replace another such test with
38651 is_gimple_assign (stmt).
38653 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38656 * config/avr/avr.c (rtl-iter.h): Include it.
38657 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38658 (avr_legitimate_combined_insn): ...and implementation.
38660 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38662 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38663 * config/s390/s390.c (s390_const_operand_ok)
38664 (s390_canonicalize_comparison, s390_extract_part)
38665 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38666 (s390_contiguous_bitmask_p, s390_rtx_costs)
38667 (legitimize_pic_address): Likewise.
38668 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38669 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38670 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38671 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38672 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38674 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38676 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38677 REGNO($0) == REGNO($1).
38679 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38681 * config/s390/linux.h(SIZE_TYPE): Add comment.
38683 2017-02-06 Julian Brown <julian@codesourcery.com>
38684 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38685 Virendra Pathak <virendra.pathak@broadcom.com>
38687 * config/aarch64/aarch64-cores.def: Change the scheduler
38689 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38690 * config/aarch64/thunderx2t99.md: New file.
38692 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38694 * doc/standards.texi (Go Language): Update link to language
38697 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38699 * tree-eh.c (lower_resx): Sanitize profile.
38700 (cleanup_empty_eh_move_lp): Likewise.
38702 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38705 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38707 * cfgloopmanip.h (loop_version): Update prototype.
38708 * modulo-sched.c (sms_schedule): Update call of loop_version.
38709 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38710 * tree-parloops.c (gen_parallel_loop): Likewise.
38711 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38712 * tree-ssa-loop-split.c (split_loop): Likewise.
38713 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
38714 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
38716 2017-02-05 Martin Liska <mliska@suse.cz>
38719 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
38721 (print_operand_address): Initialize a struct to zero.
38723 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38725 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
38726 garbage collector only in textual form.
38728 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38730 * doc/extend.texi (x86 specific memory model extensions for
38731 transactional memory): Simplify a phrase.
38733 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
38736 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
38737 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
38738 (atomic_storedi_1): Likewise.
38740 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38742 PR tree-optimization/79338
38743 * tree-parloops.c (gather_scalar_reductions): Don't call
38744 vect_analyze_loop_form for loop->inner before destroying loop's
38747 2017-02-03 Martin Sebor <msebor@redhat.com>
38749 PR tree-optimization/79327
38750 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38751 when precision has resulted in leading zeros.
38752 (format_integer): Adjust the likely counter to assume an unknown
38753 argument that may be zero is non-zero.
38755 2017-02-03 Jason Merrill <jason@redhat.com>
38758 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38759 avoid copying non-taken branch.
38761 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38763 PR tree-optimization/79340
38764 * tree-vect-loop.c (vectorizable_reduction): Release
38765 vec_defs elements after safe_splicing them into other vectors.
38768 PR tree-optimization/79327
38769 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
38770 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
38772 (format_integer): Use wide_int_to_tree instead of build_int_cst
38773 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38774 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38775 of shortest and longest sequence.
38777 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
38779 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
38780 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
38782 2017-02-03 Walter Lee <walt@tilera.com>
38785 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
38786 after initial stackframe link reg save.
38787 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
38789 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38792 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
38793 wu for stxssp alternative.
38795 2017-02-03 Martin Sebor <msebor@redhat.com>
38797 PR tree-optimization/79352
38798 * gimple-fold.c (get_range_strlen): Add argument.
38799 (get_range_strlen): Change return type to bool.
38800 (get_maxval_strlen): Pass in a dummy argument.
38801 * gimple-fold.h (get_range_strlen): Change return type to bool.
38802 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38803 * tree.h (array_at_struct_end_p): Add argument.
38804 * tree.c (array_at_struct_end_p): Handle it.
38806 2017-02-03 Martin Liska <mliska@suse.cz>
38809 * multiple_target.c (create_dispatcher_calls): Redirect edge
38810 from a caller of a dispatcher.
38811 (expand_target_clones): Make the clones local.
38812 (ipa_target_clone): Do both target clones and resolvers.
38813 (ipa_dispatcher_calls): Remove the pass.
38814 (pass_dispatcher_calls::gate): Likewise.
38815 (make_pass_dispatcher_calls): Likewise.
38816 * passes.def (pass_target_clone): Put as very first IPA early
38819 2017-02-03 Martin Liska <mliska@suse.cz>
38821 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38822 in case of a function with ifunc attribute.
38824 2017-02-03 Martin Liska <mliska@suse.cz>
38826 * cgraph.c (cgraph_node::dump): Dump function version info.
38827 * symtab.c (symtab_node::dump_base): Add missing new line.
38829 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38831 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38832 (ifcombine_ifandif): Use it.
38834 2017-02-03 Martin Liska <mliska@suse.cz>
38836 * doc/invoke.texi: Document default value for
38837 use-after-scope-direct-emission-threshold.
38839 2017-02-03 Martin Liska <mliska@suse.cz>
38841 PR tree-optimization/79339
38842 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38843 (format_floating): Likewise.
38845 2017-02-03 Martin Liska <mliska@suse.cz>
38848 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38850 (ipa_node_params_t::remove): Likewise.
38851 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38852 initialization from removed ipa_node_params_t::insert.
38853 (ipa_node_params::~ipa_node_params): Move from removed
38854 ipa_node_params_t::release.
38855 * symbol-summary.h (symbol_summary::m_released): New member.
38856 Do not release a summary twice. Do not allow to call finalizer
38857 for types of a summary that live in GGC memory.
38859 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38861 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38864 2017-02-02 Martin Sebor <msebor@redhat.com>
38866 PR middle-end/79275
38867 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38868 (format_string): Tighten up the range of output for non-constant
38869 strings and correct the expected range for wide non-constant strings.
38871 2017-02-02 Martin Sebor <msebor@redhat.com>
38873 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38875 PR middle-end/32003
38876 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38878 (-fdump-tree-@var): Add to index and document how to come up
38879 with pass-specific option and dump file names.
38880 (-fdump-passes): Clarify where to look for output.
38882 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38884 PR middle-end/77445
38885 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38886 statistics of the analyzed path; allow threading for speed when
38887 any of BBs along the path are optimized for speed.
38889 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38891 PR middle-end/78468
38892 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38893 settings of the virtual registers.
38896 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38898 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38899 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38902 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38904 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38905 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38907 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38909 * config/s390/s390.md: Add missing comments with the expanded
38911 * config/s390/vector.md: Likewise.
38912 * config/s390/vx-builtins.md: Likewise.
38914 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38917 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38918 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38919 conditions on a single line.
38921 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38923 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38924 __S390_VX__ to __VX__.
38926 2017-02-01 Andrew Pinski <apinski@cavium.com>
38928 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38929 stmt_info to record_stmt_cost.
38930 (vect_get_known_peeling_cost): Pass stmt_info if known to
38932 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38933 cpu_vector_cost field into
38934 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38935 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38936 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38937 splitting of scalar_stmt_cost and vec_stmt_cost.
38938 (thunderx_vector_cost): Likewise.
38939 (cortexa57_vector_cost): LIkewise.
38940 (exynosm1_vector_cost): Likewise.
38941 (xgene1_vector_cost): Likewise.
38942 (thunderx2t99_vector_cost): Improve after the splitting of the two
38944 (aarch64_builtin_vectorization_cost): Update for the splitting of
38945 scalar_stmt_cost and vec_stmt_cost.
38947 2017-02-01 Torvald Riegel <triegel@redhat.com>
38948 Richard Henderson <rth@redhat.com>
38950 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38951 conditional on existance of a fast atomic load.
38952 * optabs-query.c (can_atomic_load_p): New function.
38953 * optabs-query.h (can_atomic_load_p): Declare it.
38954 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38955 no fast atomic load is available for the particular size of access.
38956 (expand_atomic_compare_and_swap): Likewise.
38957 (expand_atomic_load): Likewise.
38958 (expand_atomic_store): Likewise.
38959 (expand_atomic_fetch_op): Likewise.
38960 * testsuite/lib/target-supports.exp
38961 (check_effective_target_sync_int_128): Remove x86 because it provides
38962 no fast atomic load.
38963 (check_effective_target_sync_int_128_runtime): Likewise.
38965 2017-02-01 Richard Biener <rguenther@suse.de>
38967 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38968 (graphite_transform_loops): Provide opt-info for optimized nests.
38969 * tree-parloop.c (parallelize_loops): Provide opt-info for
38970 parallelized loops.
38972 2017-02-01 Richard Biener <rguenther@suse.de>
38974 PR middle-end/79315
38975 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38976 was not set before.
38978 2017-02-01 Richard Biener <rguenther@suse.de>
38980 PR tree-optimization/71824
38981 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38982 Verify the loops are valid in the merged SESE region.
38983 (scop_detection::can_represent_loop_1): Check analyzing the
38984 evolution of the number of iterations in the region succeeds.
38986 2017-01-31 Ian Lance Taylor <iant@golang.org>
38988 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38989 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38991 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38993 PR preprocessor/79210
38994 * input.c (get_substring_ranges_for_loc): Replace line_width
38995 assertion with error-handling.
38997 2017-01-31 Richard Biener <rguenther@suse.de>
38999 PR tree-optimization/77318
39000 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39001 (create_pw_aff_from_tree): Take loop parameter.
39002 (add_condition_to_pbb): Pass loop of the condition to
39003 create_pw_aff_from_tree.
39005 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39007 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39008 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39010 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39014 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39016 (convert_int_to_float128): Likewise.
39017 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39018 (convert_int_to_float128): Likewise.
39019 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39020 (UNSPEC_IEEE128_CONVERT): Likewise.
39021 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39022 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39023 Use local variables for IBM extended format.
39024 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39025 (fix_trunc<mode>si2_fprs): Likewise.
39026 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39027 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39028 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39029 to know that we can now have integers of all sizes in vector
39031 (fix<uns>_<mode>di2_hw): Likewise.
39032 (float<uns>_<mode>si2_hw): Likewise.
39033 (fix_<mode>si2_hw): Likewise.
39034 (fixuns_<mode>si2_hw): Likewise.
39035 (float<uns>_<mode>di2_hw): Likewise.
39036 (float_<mode>di2_hw): Likewise.
39037 (float_<mode>si2_hw): Likewise.
39038 (floatuns_<mode>di2_hw): Likewise.
39039 (floatuns_<mode>si2_hw): Likewise.
39040 (xscvqp<su>wz_<mode>): Delete, no longer used.
39041 (xscvqp<su>dz_<mode>): Likewise.
39042 (xscv<su>dqp_<mode>): Likewise.
39043 (ieee128_mfvsrd_64bit): Likewise.
39044 (ieee128_mfvsrd_32bit): Likewise.
39045 (ieee128_mfvsrwz): Likewise.
39046 (ieee128_mtvsrw): Likewise.
39047 (ieee128_mtvsrd_64bit): Likewise.
39048 (ieee128_mtvsrd_32bit): Likewise.
39050 2017-01-31 Martin Liska <mliska@suse.cz>
39053 * ipa-prop.c (ipa_free_all_node_params): Call release method
39054 instead of ~sumbol_summary to not to trigger double times
39057 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39059 PR tree-optimization/71691
39060 * bitmap.h (class auto_bitmap): New.
39061 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39062 is_maybe_undefined instead of ssa_undefined_value_p.
39064 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39066 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39067 __S390_ARCH_LEVEL__ to __ARCH__.
39069 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39071 PR tree-optimization/79267
39072 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39073 if should_remove_lhs_p is true.
39075 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39078 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39079 (add_alignment_attribute): New.
39080 (base_type_die): Add alignment attribute.
39081 (subrange_type_die): Likewise.
39082 (modified_type_die): Likewise.
39083 (gen_array_type_die): Likewise.
39084 (gen_descr_array_type_die: Likewise.
39085 (gen_enumeration_type_die): Likewise.
39086 (gen_subprogram_die): Likewise.
39087 (gen_variable_die): Likewise.
39088 (gen_field_die): Likewise.
39089 (gen_ptr_to_mbr_type_die): Likewise.
39090 (gen_struct_or_union_type_die): Likewise.
39091 (gen_subroutine_type_die): Likewise.
39092 (gen_typedef_die): Likewise.
39093 (base_type_cmp): Compare alignment attribute.
39095 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39098 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39099 (setb_unsigned) New pattern for setb with CCUNS.
39100 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39101 subfc./subfe sequence to avoid overflow problems. Generate a
39102 shorter sequence with cmpld/setb for power9.
39103 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39104 for generating subfc. instruction.
39105 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39106 now uses this instruction.
39108 2017-01-30 Ian Lance Taylor <iant@google.com>
39111 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39112 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39114 2017-01-30 Martin Sebor <msebor@redhat.com>
39116 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39117 Move constant to the right of a relational operator.
39118 (get_mpfr_format_length, format_character, format_string): Ditto.
39119 (should_warn_p, maybe_warn): Same.
39121 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39123 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39126 * asan.c (get_translation_unit_decl): Remove function.
39127 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39129 2017-01-30 Martin Liska <mliska@suse.cz>
39131 PR gcov-profile/79259
39132 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39133 -fprofile-generate.
39135 2017-01-30 Martin Liska <mliska@suse.cz>
39138 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39139 Initialize variables with NULL value.
39141 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39144 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39146 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39148 2017-01-30 Richard Biener <rguenther@suse.de>
39150 PR tree-optimization/79276
39151 * tree-vrp.c (process_assert_insertions): Properly adjust common
39152 when removing a duplicate.
39154 2017-01-30 Richard Biener <rguenther@suse.de>
39156 PR tree-optimization/79256
39157 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39158 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39160 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39162 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39165 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39166 ("*r<noxa>sbg_<mode>_sll_bitmask")
39167 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39168 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39169 Use contiguous_bitmask_nowrap_operand.
39171 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39174 * config/rs6000/altivec.h (vec_xl): Revise #define.
39175 (vec_xst): Likewise.
39177 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39179 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39181 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39183 PR rtl-optimization/79194
39184 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39185 traps before call to bypass_conditional_jumps.
39187 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39189 PR tree-optimization/71374
39190 * lra-constraints.c (check_conflict_input_operands): New.
39191 (match_reload): Use it.
39193 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39196 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39197 account to calculate conflict_set.
39199 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39201 PR rtl-optimization/78559
39202 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39203 other_insn in combine.
39205 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39207 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39208 uint16_type_node for BT_UINT16.
39210 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39212 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39213 "RTL Tests" to menu.
39214 (GIMPLE Tests): New node.
39215 (RTL Tests): New node.
39217 2017-01-27 Richard Biener <rguenther@suse.de>
39219 PR tree-optimization/79245
39220 * tree-loop-distribution.c (distribute_loop): Apply cost
39221 modeling also to detected patterns.
39223 2017-01-27 Richard Biener <rguenther@suse.de>
39225 PR tree-optimization/71433
39226 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39227 (compare_assert_loc): New function.
39228 (process_assert_insertions): Sort and optimize assert locations
39229 to remove duplicates and push down identical assertions on
39230 edges to their destination block.
39232 2017-01-27 Richard Biener <rguenther@suse.de>
39234 PR tree-optimization/79244
39235 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39236 out SSA names even if abnormal.
39238 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39240 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39241 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39242 instead of MPFR_RNDN.
39244 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39247 * arm.c (arm_option_override): Don't call build_target_option_node
39248 until after doing all option overrides.
39249 (arm_valid_target_attribute_tree): Likewise.
39251 2017-01-27 Martin Liska <mliska@suse.cz>
39253 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39254 for {cd}tors and C++ {cd}tors.
39256 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39258 * config/s390/s390.md ("*setmem_long_and")
39259 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39261 2017-01-26 Martin Sebor <msebor@redhat.com>
39263 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39266 2017-01-26 Martin Sebor <msebor@redhat.com>
39268 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39269 HAVE_DFmode before using XFmode or DFmode.
39270 (parse_directive): Avoid using the z length modifier to avoid
39271 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39273 PR middle-end/78703
39274 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39275 to accept adjustment as an array.
39276 (get_int_range): New function.
39277 (struct directive): Make width and prec arrays.
39278 (directive::set_width, directive::set_precision): Call get_int_range.
39279 (format_integer, format_floating): Handle width and precision ranges.
39280 (format_string, parse_directive): Same.
39282 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39285 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39286 comdat_type_p set, just clone them, but keep the children in the
39290 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39291 which have direct callers with -fvar-tracking-assignments enabled
39293 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39294 inside of type units.
39296 2017-01-26 Martin Sebor <msebor@redhat.com>
39298 PR middle-end/78703
39299 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39301 (struct format_result): Replace number_chars, number_chars_min,
39302 and number_chars_max with a single member of struct result_range.
39304 (format_result::operator+=): Adjust.
39305 (struct fmtresult): Remove bounded. Handle likely and unlikely
39307 (fmtresult::adjust_for_width_or_precision): New function.
39308 (fmtresult:type_max_digits): New function.
39309 (bytes_remaining): Handle likely and unlikely counters.
39310 (min_bytes_remaining): Remove.
39311 (format_percent): Simplify.
39312 (format_integer, format_floating): Set likely and unlikely counters.
39313 (get_string_length, format_character, format_string): Same.
39314 (format_plain, should_warn_p): New function.
39315 (maybe_warn): Call should_warn_p. Update diagnostic messages
39316 and handle those for all directives, including plain strings.
39317 (format_directive): Handle likely and unlikely counters.
39318 Remove unnecessary quoting from diagnostics. Add an informational
39320 (add_bytes): Remove.
39321 (pass_sprintf_length::compute_format_length): Simplify.
39322 (try_substitute_return_value): Handle likely and unlikely counters.
39324 2017-01-26 Carl Love <cel@us.ibm.com>
39326 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39327 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39329 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39332 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39333 endianess for subregs into account.
39334 * lra-constraints.c (lra_constraints): Do risky transformations
39335 always on the first iteration.
39336 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39337 last_call_used_reg_set.
39338 (process_bb_lives): Define and use last_call_used_reg_set.
39339 * lra.c (lra): Always continue after lra_constraints on the first
39342 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39344 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39346 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39348 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39350 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39351 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39352 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39353 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39354 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39355 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39356 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39357 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39358 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39360 2017-01-26 Marek Polacek <polacek@redhat.com>
39363 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39364 for the third operand.
39366 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39368 PR middle-end/79236
39369 * omp-low.c (struct omp_context): Add simt_stmt field.
39370 (scan_omp_for): Return omp_context *.
39371 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39372 context to the _simt_ SIMD stmt.
39373 (lower_omp_for): For combined SIMD with sibling _simt_
39374 SIMD, make sure to use the same decls in _looptemp_
39375 clauses as in the sibling.
39377 2017-01-26 David Sherwood <david.sherwood@arm.com>
39379 PR middle-end/79212
39380 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39383 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39386 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39387 emit fld b; fld a; if possible.
39389 * brig-builtins.def: Update copyright years.
39390 * config/arm/arm_acle_builtins.def: Update copyright years.
39392 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39395 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39396 constraint instead of o for the stxsd instruction.
39398 2017-01-25 Carl Love <cel@us.ibm.com>
39400 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39401 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39403 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39405 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39407 2017-01-25 Richard Biener <rguenther@suse.de>
39409 PR tree-optimization/69264
39410 * target.def (vector_alignment_reachable): Improve documentation.
39411 * doc/tm.texi: Regenerate.
39412 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39414 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39415 earlier changes with respect to TYPE_USER_ALIGN.
39416 (vector_alignment_reachable_p): Likewise. Improve dumping.
39418 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39421 * config/arm/arm.md (xordi3): Force constant operand into a register
39424 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39426 * doc/invoke.texi (-fstore-merging): Correct default optimization
39427 levels at which it is enabled.
39428 (-O): Move -fstore-merging from list to...
39431 2017-01-25 Richard Biener <rguenther@suse.de>
39434 * omp-expand.c: Include debug.h.
39435 (expand_omp_taskreg): Make sure to generate early debug before
39436 outlining anything from a function.
39437 (expand_omp_target): Likewise.
39438 (grid_expand_target_grid_body): Likewise.
39440 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39443 * asan.c (get_translation_unit_decl): New function.
39444 (asan_add_global): Extract modules file name from globals
39445 TRANSLATION_UNIT_DECL name.
39447 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39450 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39451 for long calls with APCS frame and VFP.
39453 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39455 * cfg.c (original_copy_tables_initialized_p): New function.
39456 * cfg.h (original_copy_tables_initialized_p): New decl.
39457 * cfgrtl.c (relink_block_chain): Guard the call to
39458 free_original_copy_tables with a call to
39459 original_copy_tables_initialized_p.
39460 * cgraph.h (symtab_node::native_rtl_p): New decl.
39461 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39462 (symtab_node::needed_p): Don't assert for early assembly output
39463 for __RTL functions.
39464 (cgraph_node::finalize_function): Set "force_output" for __RTL
39466 (cgraph_node::analyze): Bail out early for __RTL functions.
39467 (analyze_functions): Update assertion to support __RTL functions.
39468 (cgraph_node::expand): Bail out early for __RTL functions.
39469 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39471 * function.h (struct function): Update comment for field
39473 * gimple-expr.c: Include "tree-pass.h".
39474 (gimple_has_body_p): Return false for __RTL functions.
39475 * Makefile.in (OBJS): Add run-rtl-passes.o.
39476 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39478 (gcc::pass_manager::get_clean_slate): New accessor.
39479 * passes.c: Include "insn-addr.h".
39480 (should_skip_pass_p): Add logging. Update logic for running
39481 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39482 property-provider override so it is only done for gimple passes.
39484 (skip_pass): New function.
39485 (execute_one_pass): Call skip_pass when skipping passes.
39486 * read-md.c (md_reader::read_char): Support filtering
39487 the input to a subset of line numbers.
39488 (md_reader::md_reader): Initialize fields
39489 m_first_line and m_last_line.
39490 (md_reader::read_file_fragment): New function.
39491 * read-md.h (md_reader::read_file_fragment): New decl.
39492 (md_reader::m_first_line): New field.
39493 (md_reader::m_last_line): New field.
39494 * read-rtl-function.c (function_reader::create_function): Only
39495 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39496 curr_properties. Set DECL_INITIAL to a dummy block.
39497 (read_rtl_function_body_from_file_range): New function.
39498 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39500 * run-rtl-passes.c: New file.
39501 * run-rtl-passes.h: New file.
39503 2017-01-24 Jeff Law <law@redhat.com>
39505 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39508 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39510 PR tree-optimization/79159
39511 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39512 (record_nonwrapping_iv): Improve boundary using above function if no
39513 value range information.
39515 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39516 Martin Jambor <mjambor@suse.cz>
39518 * brig-builtins.def: New file.
39519 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39520 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39521 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39522 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39523 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39524 * builtin-types.def (BT_INT8): New.
39525 (BT_INT16): Likewise.
39526 (BT_UINT8): Likewise.
39527 (BT_UINT16): Likewise.
39528 (BT_FN_ULONG): Likewise.
39529 (BT_FN_UINT_INT): Likewise.
39530 (BT_FN_UINT_ULONG): Likewise.
39531 (BT_FN_UINT_LONG): Likewise.
39532 (BT_FN_UINT_PTR): Likewise.
39533 (BT_FN_ULONG_PTR): Likewise.
39534 (BT_FN_INT8_FLOAT): Likewise.
39535 (BT_FN_INT16_FLOAT): Likewise.
39536 (BT_FN_UINT32_FLOAT): Likewise.
39537 (BT_FN_UINT16_FLOAT): Likewise.
39538 (BT_FN_UINT8_FLOAT): Likewise.
39539 (BT_FN_UINT64_FLOAT): Likewise.
39540 (BT_FN_UINT16_UINT32): Likewise.
39541 (BT_FN_UINT32_UINT16): Likewise.
39542 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39543 (BT_FN_INT_PTR_INT): Likewise.
39544 (BT_FN_UINT_PTR_UINT): Likewise.
39545 (BT_FN_LONG_PTR_LONG): Likewise.
39546 (BT_FN_ULONG_PTR_ULONG): Likewise.
39547 (BT_FN_VOID_UINT64_UINT64): Likewise.
39548 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39549 (BT_FN_INT8_INT8_INT8): Likewise.
39550 (BT_FN_INT16_INT16_INT16): Likewise.
39551 (BT_FN_INT_INT_INT): Likewise.
39552 (BT_FN_UINT_FLOAT_UINT): Likewise.
39553 (BT_FN_FLOAT_UINT_UINT): Likewise.
39554 (BT_FN_ULONG_UINT_UINT): Likewise.
39555 (BT_FN_ULONG_UINT_PTR): Likewise.
39556 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39557 (BT_FN_UINT_UINT_UINT): Likewise.
39558 (BT_FN_VOID_UINT_PTR): Likewise.
39559 (BT_FN_UINT_UINT_PTR: Likewise.
39560 (BT_FN_UINT32_UINT64_PTR): Likewise.
39561 (BT_FN_INT_INT_UINT_UINT): Likewise.
39562 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39563 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39564 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39565 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39566 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39567 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39568 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39569 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39570 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39571 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39572 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39573 * doc/frontends.texi: List BRIG FE.
39574 * doc/install.texi (Testing): Add BRIG tesring requirements.
39575 * doc/invoke.texi (Overall Options): Mention BRIG.
39576 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39578 2017-01-24 Richard Biener <rguenther@suse.de>
39580 PR translation/79208
39581 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39583 2017-01-24 Martin Jambor <mjambor@suse.cz>
39586 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39587 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39588 and known_contexts.
39590 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39592 PR middle-end/79123
39593 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39594 casts from signed to unsigned really don't have a range.
39596 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39598 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39599 GMP_RNDx for compatiblity.
39601 2017-01-24 Martin Liska <mliska@suse.cz>
39604 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39605 that would prevent us to call alloca with -1 as argument.
39607 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39609 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39610 Avoid -Wformat-security warning.
39612 2017-01-23 Andrew Pinski <apinski@cavium.com>
39614 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39617 2017-01-23 Martin Sebor <msebor@redhat.com>
39619 PR middle-end/78703
39620 * gimple-ssa-sprintf.c (warn_level): New global.
39621 (format_integer): Use it here and throughout the rest of the file.
39622 Use the same switch to compute sign as base.
39623 (maybe_warn): New function.
39624 (format_directive): Factor out warnings into maybe_warn.
39625 Add debugging output. Use warn_level.
39626 (add_bytes): Use warn_level.
39627 (pass_sprintf_length::compute_format_length): Add debugging output.
39628 (try_substitute_return_value): Same.
39629 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39631 PR middle-end/78703
39632 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39633 (struct fmtresult, format_integer, format_floating): Adjust.
39634 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39635 (get_string_length, format_string,format_directive): Same.
39636 (pass_sprintf_length::compute_format_length): Same.
39637 (try_substitute_return_value): Simplify slightly.
39639 PR middle-end/78703
39640 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39641 (fmtresult::operator+=): Outlined.
39642 (struct fmtresult): Add ctors.
39643 (struct conversion_spec): Rename...
39644 (struct directive): ...to this. Add and remove data members.
39645 (directive::set_width, directive::set_precision): New functions.
39646 (format_percent): Use fmtresult ctor.
39647 (get_width_and_precision): Remove.
39648 (format_integer): Make naming changes. Avoid computing width and
39650 (format_floating): Same. Adjust indentation.
39651 (format_character, format_none): New functions.
39652 (format_string): Moved character handling to format_character.
39653 (format_directive): Remove arguments, change return type.
39654 (parse_directive): New function.
39655 (pass_sprintf_length::compute_format_length): Move directive
39656 parsing to parse_directive.
39658 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39660 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39661 (assign_assembler_name_if_needed): ... this.
39662 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39663 (assign_assembler_name_if_needed): ... this.
39664 (free_lang_data_in_cgraph): Adjust callers.
39665 * cgraphunit.c (cgraph_node::analyze): Likewise.
39666 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39669 2017-01-23 Richard Biener <rguenther@suse.de>
39671 PR tree-optimization/79088
39672 PR tree-optimization/79188
39673 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39674 resetting loop bounds after last path deletion. Reset loop
39675 bounds of the target loop, make code match the comments.
39676 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39677 Make sure loops need no fixups.
39679 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39681 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39682 exponent support with double type for first argument.
39683 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39684 type returned by __builtin_vec_extract_sig,
39685 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39686 functions from "vector int" to "vector unsigned int" or from
39687 "vector long long int" to "vector unsigned long long int".
39688 Changed type returned by __builtin_vec_extract_exp,
39689 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39690 functions from "vector int" to "vector unsigned int" or from
39691 "vector long long int" to "vector unsigned long long int".
39692 Changed return type of __builtin_vec_test_data_class,
39693 __builtin_vec_test_data_class_sp, and
39694 __builtin_vec_test_data_class_dp from "vector int" to
39695 "vector bool int" or from "vector long long int" to "vector bool
39696 long long int" and changed second argument type from "unsigned
39697 int" to "int". Added new overloaded function forms "vector float
39698 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39699 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39700 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39701 double, vector unsigned long long int)" and "vector double
39702 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39703 long int)". Changed return type of
39704 __builtin_scalar_test_data_class and
39705 __builtin_scalar_test_data_class_sp and
39706 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39707 int" and changed second argument from "unsigned int" to "int".
39708 Changed type returned by __builtin_scalar_test_neg,
39709 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39710 from "int" to "bool int". Added new overloaded function form
39711 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39712 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
39713 exponent double-precision with floating point first argument.
39714 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
39715 documentation of scalar_test_data_class, scalar_test_neg,
39716 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
39717 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
39718 vec_test_data_class built-in functions to reflect refinements in
39719 their type signatures.
39721 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
39723 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
39725 (aarch64_elf_asm_destructor): Likewise.
39727 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
39729 PR rtl-optimization/78634
39730 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
39731 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
39732 * ifcvt.c (noce_try_cmove): Add missing cost check.
39734 PR rtl-optimization/71724
39735 * combine.c (if_then_else_cond): Look for situations where it is
39736 beneficial to undo the work of one of the recursive calls.
39738 2017-01-23 Bin Cheng <bin.cheng@arm.com>
39740 PR tree-optimization/70754
39741 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39742 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39743 combined stmt before it if not NULL.
39744 (combine_chains): Process refs reversely and compute dominance point
39747 2017-01-23 Martin Liska <mliska@suse.cz>
39749 PR tree-optimization/79196
39750 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39751 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39753 (strlen_optimize_stmt): Call the renamed function.
39755 2017-01-23 Michael Matz <matz@suse.de>
39757 PR tree-optimization/78384
39758 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39760 2017-01-23 Richard Biener <rguenther@suse.de>
39762 PR tree-optimization/79186
39763 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39764 both incoming edges before moving an assert.
39766 2017-01-23 Martin Jambor <mjambor@suse.cz>
39768 * ipa-prop.c (load_from_param_1): Removed.
39769 (load_from_unmodified_param): Bits from load_from_param_1 put back
39771 (load_from_param): Removed.
39772 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
39773 with stmt. Reverted back to use of load_from_unmodified_param.
39775 2017-01-23 Martin Jambor <mjambor@suse.cz>
39778 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
39779 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
39780 field a pointer to garbage collected vector, mark lattices and
39781 ipcp_orig_node with GTY((skip)).
39782 (ipa_get_param_count): Adjust to descriptors being a pointer.
39783 (ipa_get_param): Likewise.
39784 (ipa_get_type): Likewise.
39785 (ipa_get_param_move_cost): Likewise.
39786 (ipa_set_param_used): Likewise.
39787 (ipa_get_controlled_uses): Likewise.
39788 (ipa_set_controlled_uses): Likewise.
39789 (ipa_is_param_used): Likewise.
39790 (ipa_node_params_t): Move into garbage collector. New methods insert
39792 (ipa_node_params_sum): Annotate wth GTY(()).
39793 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
39795 (ipa_load_from_parm_agg): Adjust declaration.
39796 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
39797 * ipa-profile.c (ipa_profile): Likewise.
39798 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39799 (ipa_populate_param_decls): Make descriptors parameter garbage
39801 (ipa_dump_param): Adjust to descriptors being a pointer.
39802 (ipa_alloc_node_params): Likewise.
39803 (ipa_initialize_node_params): Likewise.
39804 (load_from_param_1): Make descriptors parameter garbage collected.
39805 (load_from_unmodified_param): Likewise.
39806 (load_from_param): Likewise.
39807 (ipa_load_from_parm_agg): Likewise.
39808 (ipa_node_params::~ipa_node_params): Removed.
39809 (ipa_free_all_node_params): Remove call to delete operator.
39810 (ipa_node_params_t::insert): New.
39811 (ipa_node_params_t::remove): Likewise.
39812 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39813 copy known_csts and known_contexts vectors.
39814 (ipa_read_node_info): Adjust to descriptors being a pointer.
39815 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39817 (ipcp_transform_function): Make descriptors variable garbage
39820 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39822 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39823 * config/i386/avx512dqintrin.h: Ditto.
39824 * config/i386/avx512fintrin.h: Ditto.
39825 * config/i386/i386.c: Handle new builtins.
39826 * config/i386/i386-builtin.def: Add new builtins.
39827 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39828 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39830 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39831 Martin Liska <mliska@suse.cz>
39833 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39834 * asan.c (asan_expand_poison_ifn): Support stores and use
39835 appropriate ASAN report function.
39836 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39837 * internal-fn.def (ASAN_POISON_USE): Declare.
39838 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39839 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39840 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39841 ASAN_POISON calls w/o LHS.
39842 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39843 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39844 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39845 * gimplify.c (asan_poison_variables): Add attribute
39846 use_after_scope_memory to variables that really needs to live
39848 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39849 having the attribute.
39851 2017-01-23 Martin Liska <mliska@suse.cz>
39853 * asan.c (create_asan_shadow_var): New function.
39854 (asan_expand_poison_ifn): Likewise.
39855 * asan.h (asan_expand_poison_ifn): New declaration.
39856 * internal-fn.c (expand_ASAN_POISON): Likewise.
39857 * internal-fn.def (ASAN_POISON): New builtin.
39858 * sanopt.c (pass_sanopt::execute): Expand
39859 asan_expand_poison_ifn.
39860 * tree-inline.c (copy_decl_for_dup_finish): Make function
39862 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39863 * tree-ssa.c (is_asan_mark_p): New function.
39864 (execute_update_addresses_taken): Rewrite local variables
39865 (identified just by use-after-scope as addressable) into SSA.
39867 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39869 * doc/install.texi (Specific): opensource.apple.com uses https
39870 now. Remove trailing slash.
39872 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39874 * README.Portability: Remove note on an Irix compatibility issue.
39876 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39878 * gcov.c (INCLUDE_ALGORITHM): Define.
39879 (INCLUDE_VECTOR): Define.
39880 No longer include <vector> and <algorithm> directly.
39882 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39884 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39886 * doc/invoke.texi (Code Gen Options): Ditto.
39888 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39891 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39893 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39895 rtl-optimization/79125
39896 * cprop.c (local_cprop_pass): Handle cases where we make an
39897 unconditional trap.
39899 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39903 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39904 read from, for big endian.
39906 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39908 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39909 register pauth builtins for LP64 only.
39911 2017-01-20 Marek Polacek <polacek@redhat.com>
39914 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39917 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39919 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39921 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39922 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39923 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39925 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39928 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39929 in big-endian mode when they are not a single duplicated value.
39931 2017-01-20 Richard Biener <rguenther@suse.de>
39933 * BASE-VER: Bump to 7.0.1.
39935 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39937 * omp-low.c (omplow_simd_context): New struct. Use it...
39938 (lower_rec_simd_input_clauses): ...here and...
39939 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39940 references to idx, lane, max_vf, is_simt.
39942 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39944 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39947 2017-01-20 Martin Jambor <mjambor@suse.cz>
39949 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39950 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39952 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39953 (GTFILES): Rename hsa.c to hsa-common.c.
39954 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39955 * hsa-dump.c: Likewise.
39956 * hsa-gen.c: Likewise.
39957 * hsa-regalloc.c: Likewise.
39958 * ipa-hsa.c: Likewise.
39959 * omp-expand.c: Likewise.
39960 * omp-low.c: Likewise.
39961 * toplev.c: Likewise.
39963 2017-01-20 Marek Polacek <polacek@redhat.com>
39966 * doc/invoke.texi: Document -Wduplicated-branches.
39967 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39968 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39969 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39970 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39971 return 0 only when not OEP_LEXICOGRAPHIC.
39972 (fold_build_cleanup_point_expr): Use the expression
39973 location when building CLEANUP_POINT_EXPR.
39974 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39975 * tree.c (add_expr): Handle error_mark_node.
39977 2017-01-20 Martin Liska <mliska@suse.cz>
39980 * tree-profile.c (init_ic_make_global_vars): Do not call
39982 (gimple_init_gcov_profiler): Likewise.
39984 2017-01-20 Martin Liska <mliska@suse.cz>
39987 * cgraph.h (maybe_create_reference): Remove argument and
39989 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39991 * ipa-cp.c (create_specialized_node): Likewise.
39992 * symtab.c (symtab_node::maybe_create_reference): Handle
39993 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39995 2017-01-20 Martin Liska <mliska@suse.cz>
39997 * read-rtl-function.c (function_reader::create_function): Use
39998 build_decl instread of build_decl_stat.
40000 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40002 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40003 * config/i386/avx512dqintrin.h: Ditto.
40004 * config/i386/avx512fintrin.h: Ditto.
40005 * config/i386/i386-builtin-types.def: Add new types.
40006 * config/i386/i386.c: Handle new types.
40007 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40008 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40009 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40010 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40011 (__builtin_ia32_kshiftridi): New.
40012 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40014 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40018 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40019 define to rs6000_init_stack_protect_guard.
40020 (rs6000_init_stack_protect_guard): New function.
40022 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40023 Yunqiang Su <yunqiang.su@imgtec.com>
40025 * config.gcc (supported_defaults): Add madd4.
40026 (with_madd4): Add validation.
40027 (all_defaults): Add madd4.
40028 * config/mips/mips.opt (mmadd4): New option.
40029 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40031 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40033 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40034 (ISA_HAS_FUSED_MADD4): Likewise.
40035 * doc/invoke.texi (-mmadd4): Document the new option.
40036 * doc/install.texi (--with-madd4): Document the new option.
40038 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40040 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40041 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40042 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40043 (aarch64_init_pauth_hint_builtins): New.
40044 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40045 (aarch64_expand_builtin): Expand new builtins.
40047 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40049 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40050 * combine-stack-adj.c (no_unhandled_cfa): Handle
40051 REG_CFA_TOGGLE_RA_MANGLE.
40052 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40053 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40054 info for return address signing.
40055 (aarch64_expand_epilogue): Likewise.
40057 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40059 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40060 * config/aarch64/aarch64-protos.h
40061 (aarch64_return_address_signing_enabled): New declaration.
40062 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40064 (aarch64_expand_prologue): Sign return address before it's pushed onto
40066 (aarch64_expand_epilogue): Authenticate return address fetched from
40068 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40069 (aarch64_attributes): New function attributes for "sign-return-address".
40070 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40071 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40072 ("*do_return"): Generate combined instructions according to key index.
40073 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40074 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40076 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40077 * config/aarch64/aarch64.opt (msign-return-address=): New.
40078 * doc/extend.texi (AArch64 Function Attributes): Documents
40079 "sign-return-address=".
40080 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40082 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40084 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40085 overall option summary.
40087 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40089 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40090 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40091 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40092 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40094 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40096 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40097 -mpower9-minmax by default for -mcpu=power9.
40098 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40099 128-bit floating point.
40101 2017-01-20 Alan Modra <amodra@gmail.com>
40103 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40104 optimizing for size.
40106 2017-01-20 Alan Modra <amodra@gmail.com>
40109 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40110 for strcmp and strncmp from corresponding builtin decl.
40112 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40114 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40115 instead of i386/rtems-64.h.
40116 * config/i386/rtems-64.h: Remove.
40118 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40122 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40124 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40126 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40128 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40129 Change int to HOST_WIDE_INT.
40130 * config/aarch64/aarch64-protos.h
40131 (aarch64_simd_gen_const_vector_dup): Likewise.
40132 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40134 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40136 * langhooks-def.h (lhd_type_for_size): New decl.
40137 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40138 * langhooks.c (lhd_type_for_size): New function, taken from
40141 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40143 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40144 define_bypass for CR latency.
40145 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40146 (power9-alu2): Add define_bypass for CR latency.
40148 (power9-mul): Update insn latency.
40149 (power9-mul-compare): Update insn latency, bypass latency and remove
40152 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40154 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40156 * config/aarch64/aarch64.md
40157 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40158 aarch64_nopcrelative_literal_loads.
40159 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40161 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40163 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40164 TARGET_LOONGSON_3A.
40165 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40167 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40170 * config.gcc (supported_defaults): Add lxc1-sxc1.
40171 (with_lxc1_sxc1): Add validation.
40172 (all_defaults): Add lxc1-sxc1.
40173 * config/mips/mips.opt (mlxc1-sxc1): New option.
40174 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40176 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40177 __mips_no_lxc1_sxc1.
40178 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40179 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40180 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40182 2017-01-19 Richard Biener <rguenther@suse.de>
40184 PR tree-optimization/72488
40185 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40186 sure to restore SSA info.
40187 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40189 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40191 PR rtl-optimization/79121
40192 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40193 of the inner type when shifting an extended value.
40195 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40198 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40199 interposable aliases.
40201 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40204 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40205 Use the evmergelohi instruction.
40206 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40207 (mov_sitf_e500_subreg8_2_be): Likewise.
40208 (mov_sitf_e500_subreg12_2_le): Likewise.
40209 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40210 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40211 (mov_sitf_e500_subreg8_2_le): Likewise.
40212 (mov_sitf_e500_subreg12_2_be): Likewise.
40214 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40216 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40217 attribute from vecsimple to vecperm.
40218 (altivec_vbpermq2): Likewise.
40220 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40223 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40225 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40226 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40227 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40228 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40229 case where N arg is SIZE_MAX.
40230 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40231 (cmpstrsi): Add pattern.
40233 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40235 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40236 __builtin_vec_revb builtins.
40237 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40238 built-in functions to support generation of the ISA 3.0 XXBR<x>
40239 vector byte reverse instructions.
40240 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40241 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40242 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40243 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40244 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40245 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40246 (P9V_BUILTIN_VEC_REVB): Likewise.
40247 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40248 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40249 (p9_xxbrq_v16qi): Likewise.
40250 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40251 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40252 (p9_xxbrh_v8hi): Likewise.
40253 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40254 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40255 vec_revb built-in functions.
40257 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40259 PR rtl-optimization/78952
40260 * config/i386/i386.md (any_extract): New code iterator.
40261 (*insvqi_2): Use any_extract for source operand.
40262 (*insvqi_3): Use any_shiftrt for source operand.
40264 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40266 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40268 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40270 2017-01-18 Matthias Klose <doko@ubuntu.com>
40272 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40274 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40276 * config/rs6000/altivec.h (vec_bperm): Change #define.
40277 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40278 (altivec_vbpermq2): New define_insn.
40279 (altivec_vbpermd): Likewise.
40280 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40281 function interface.
40282 (VBPERMD): Likewise.
40283 (VBPERM): New polymorphic function interface.
40284 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40285 Add entries for P9V_BUILTIN_VEC_VBPERM.
40286 * doc/extend.texi: Add interfaces for vec_bperm.
40288 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40290 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40291 first letter of error messages.
40292 (s390_resolve_overloaded_builtin): Likewise.
40293 * config/s390/s390.c (s390_expand_builtin): Likewise.
40294 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40295 (s390_valid_target_attribute_inner_p): Likewise.
40296 * config/s390/s390.md ("tabort"): Likewise.
40298 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40300 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40301 (ISA_AVOID_DIV_HILO): New macro.
40302 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40303 (ISA_HAS_DDIV): Likewise.
40305 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40307 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40309 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40311 * doc/invoke.texi (fabi-version): Spelling fix.
40313 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40316 * doc/invoke.texi (fabi-version): Mention mangling fix for
40319 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40322 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40324 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40327 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40328 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40330 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40332 (stack_protect_test): Ditto.
40333 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40334 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40336 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40337 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40338 -mstack-protector-guard-offset=.
40339 (RS/6000 and PowerPC Options): Ditto.
40341 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40343 * config/i386/i386.h (MASK_CLASS_P): New define.
40344 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40345 there are no registers from different register sets also when
40346 mask registers are used. Update function comment.
40347 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40348 to (*k/*r) and (*k/*km) alternatives.
40350 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40352 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40353 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40354 (EH_RETURN_HANDLER_RTX): New define.
40355 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40356 Force frame pointer in EH return functions.
40357 (aarch64_expand_epilogue): Add barrier for eh_return.
40358 (aarch64_final_eh_return_addr): Remove.
40359 (aarch64_eh_return_handler_rtx): New function.
40360 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40362 (aarch64_eh_return_handler_rtx): New prototype.
40364 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40366 * config/rs6000/altivec.h (vec_rlmi): New #define.
40367 (vec_vrlnm): Likewise.
40368 (vec_rlnm): Likewise.
40369 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40370 (UNSPEC_VRLNM): Likewise.
40371 (VIlong): New mode iterator.
40372 (altivec_vrl<VI_char>mi): New define_insn.
40373 (altivec_vrl<VI_char>nm): Likewise.
40374 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40376 (VRLDNM): Likewise.
40377 (RLNM): New polymorphic function entry.
40378 (VRLWMI): New monomorphic function entry.
40379 (VRLDMI): Likewise.
40380 (RLMI): New polymorphic function entry.
40381 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40382 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40383 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40386 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40389 * dwarf2out.c (field_byte_offset): Restore the
40390 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40391 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40392 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40395 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40398 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40400 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40403 * dwarf2out.c (add_data_member_location_attribute): For constant
40404 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40405 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40406 DW_AT_byte_size attributes.
40408 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40410 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40411 after forcing to constant memory when the code model is medium.
40413 2017-01-17 Julia Koval <julia.koval@intel.com>
40416 * config/i386/avx512fintrin.h
40417 (_mm512_i32gather_ps): Change __addr type to void const*.
40418 (_mm512_mask_i32gather_ps): Ditto.
40419 (_mm512_i32gather_pd): Ditto.
40420 (_mm512_mask_i32gather_pd): Ditto.
40421 (_mm512_i64gather_ps): Ditto.
40422 (_mm512_mask_i64gather_ps): Ditto.
40423 (_mm512_i64gather_pd): Ditto.
40424 (_mm512_mask_i64gather_pd): Ditto.
40425 (_mm512_i32gather_epi32): Ditto.
40426 (_mm512_mask_i32gather_epi32): Ditto.
40427 (_mm512_i32gather_epi64): Ditto.
40428 (_mm512_mask_i32gather_epi64): Ditto.
40429 (_mm512_i64gather_epi32): Ditto.
40430 (_mm512_mask_i64gather_epi32): Ditto.
40431 (_mm512_i64gather_epi64): Ditto.
40432 (_mm512_mask_i64gather_epi64): Ditto.
40433 (_mm512_i32scatter_ps): Change __addr type to void*.
40434 (_mm512_mask_i32scatter_ps): Ditto.
40435 (_mm512_i32scatter_pd): Ditto.
40436 (_mm512_mask_i32scatter_pd): Ditto.
40437 (_mm512_i64scatter_ps): Ditto.
40438 (_mm512_mask_i64scatter_ps): Ditto.
40439 (_mm512_i64scatter_pd): Ditto.
40440 (_mm512_mask_i64scatter_pd): Ditto.
40441 (_mm512_i32scatter_epi32): Ditto.
40442 (_mm512_mask_i32scatter_epi32): Ditto.
40443 (_mm512_i32scatter_epi64): Ditto.
40444 (_mm512_mask_i32scatter_epi64): Ditto.
40445 (_mm512_i64scatter_epi32): Ditto.
40446 (_mm512_mask_i64scatter_epi32): Ditto.
40447 (_mm512_i64scatter_epi64): Ditto.
40448 (_mm512_mask_i64scatter_epi64): Ditto.
40449 * config/i386/avx512pfintrin.h
40450 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40451 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40452 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40453 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40454 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40455 (_mm512_prefetch_i32scatter_ps): Ditto.
40456 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40457 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40458 (_mm512_prefetch_i64scatter_pd): Ditto.
40459 (_mm512_prefetch_i64scatter_ps): Ditto.
40460 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40461 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40462 * config/i386/avx512vlintrin.h
40463 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40464 (_mm_mmask_i32gather_ps): Ditto.
40465 (_mm256_mmask_i32gather_pd): Ditto.
40466 (_mm_mmask_i32gather_pd): Ditto.
40467 (_mm256_mmask_i64gather_ps): Ditto.
40468 (_mm_mmask_i64gather_ps): Ditto.
40469 (_mm256_mmask_i64gather_pd): Ditto.
40470 (_mm_mmask_i64gather_pd): Ditto.
40471 (_mm256_mmask_i32gather_epi32): Ditto.
40472 (_mm_mmask_i32gather_epi32): Ditto.
40473 (_mm256_mmask_i32gather_epi64): Ditto.
40474 (_mm_mmask_i32gather_epi64): Ditto.
40475 (_mm256_mmask_i64gather_epi32): Ditto.
40476 (_mm_mmask_i64gather_epi32): Ditto.
40477 (_mm256_mmask_i64gather_epi64): Ditto.
40478 (_mm_mmask_i64gather_epi64): Ditto.
40479 (_mm256_i32scatter_ps): Change __addr type to void*.
40480 (_mm256_mask_i32scatter_ps): Ditto.
40481 (_mm_i32scatter_ps): Ditto.
40482 (_mm_mask_i32scatter_ps): Ditto.
40483 (_mm256_i32scatter_pd): Ditto.
40484 (_mm256_mask_i32scatter_pd): Ditto.
40485 (_mm_i32scatter_pd): Ditto.
40486 (_mm_mask_i32scatter_pd): Ditto.
40487 (_mm256_i64scatter_ps): Ditto.
40488 (_mm256_mask_i64scatter_ps): Ditto.
40489 (_mm_i64scatter_ps): Ditto.
40490 (_mm_mask_i64scatter_ps): Ditto.
40491 (_mm256_i64scatter_pd): Ditto.
40492 (_mm256_mask_i64scatter_pd): Ditto.
40493 (_mm_i64scatter_pd): Ditto.
40494 (_mm_mask_i64scatter_pd): Ditto.
40495 (_mm256_i32scatter_epi32): Ditto.
40496 (_mm256_mask_i32scatter_epi32): Ditto.
40497 (_mm_i32scatter_epi32): Ditto.
40498 (_mm_mask_i32scatter_epi32): Ditto.
40499 (_mm256_i32scatter_epi64): Ditto.
40500 (_mm256_mask_i32scatter_epi64): Ditto.
40501 (_mm_i32scatter_epi64): Ditto.
40502 (_mm_mask_i32scatter_epi64): Ditto.
40503 (_mm256_i64scatter_epi32): Ditto.
40504 (_mm256_mask_i64scatter_epi32): Ditto.
40505 (_mm_i64scatter_epi32): Ditto.
40506 (_mm_mask_i64scatter_epi32): Ditto.
40507 (_mm256_i64scatter_epi64): Ditto.
40508 (_mm256_mask_i64scatter_epi64): Ditto.
40509 (_mm_i64scatter_epi64): Ditto.
40510 (_mm_mask_i64scatter_epi64): Ditto.
40511 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40512 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40513 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40514 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40515 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40516 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40517 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40518 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40519 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40520 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40521 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40522 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40523 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40524 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40525 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40526 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40527 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40528 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40529 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40530 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40531 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40532 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40533 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40534 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40535 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40536 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40537 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40538 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40539 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40540 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40541 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40542 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40543 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40544 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40545 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40546 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40547 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40548 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40549 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40550 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40551 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40552 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40553 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40554 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40555 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40556 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40557 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40558 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40559 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40560 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40561 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40562 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40563 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40564 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40565 definitions accordingly.
40567 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40568 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40571 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40574 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40577 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40578 hard regs for inner regno.
40580 2017-01-17 Martin Liska <mliska@suse.cz>
40583 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40584 assumption and add comment.
40586 2017-01-17 Nathan Sidwell <nathan@acm.org>
40588 * ipa-visibility.c (localize_node): New function, broken out of ...
40589 (function_and_variable_visibility): ... here. Call it.
40591 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40593 PR middle-end/77445
40594 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40595 correctly set frequency of oudgoing edge.
40596 (duplicate_thread_path): Fix profile updating.
40598 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40601 * configure.ac: Add GCC_BASE_VER.
40602 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40603 version from BASE-VER file.
40604 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40605 (gcc.o): Depend on $(BASEVER).
40606 * common.opt (dumpfullversion): New option.
40607 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40608 * doc/invoke.texi: Document -dumpfullversion.
40609 * doc/install.texi: Document --with-gcc-major-version-only.
40610 * configure: Regenerated.
40612 2017-01-17 Richard Biener <rguenther@suse.de>
40614 PR tree-optimization/71433
40615 * tree-vrp.c (register_new_assert_for): Merge same asserts
40616 on all incoming edges.
40617 (process_assert_insertions_for): Handle insertions at the
40620 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40622 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40623 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40625 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40628 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40631 2017-01-17 Alan Modra <amodra@gmail.com>
40634 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40635 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40636 symbolic stack limit when pic.
40638 2017-01-16 Martin Sebor <msebor@redhat.com>
40640 PR tree-optimization/78608
40641 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40643 2017-01-16 Jeff Law <law@redhat.com>
40646 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40647 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40648 for several include directories that may be relative to sysroot.
40649 * config/i386/x-mingw32 (gplus_includedir): Define.
40650 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40651 (native_system_includedir): Likewise.
40652 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40653 override if TARGET_SYSTEM_ROOT is defined.
40654 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40656 PR tree-optimization/79090
40657 PR tree-optimization/33562
40658 PR tree-optimization/61912
40659 PR tree-optimization/77485
40660 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40661 and computed trims into the dump file.
40663 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40665 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40667 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40670 * gimplify.c (gimplify_init_constructor): If want_value and
40671 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40675 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40676 sequence. Formatting fixes.
40677 (doloop_optimize): Formatting fixes.
40680 * gcc.c (debug_level_greater_than_spec_func): New function.
40681 (static_spec_functions): Add debug-level-gt spec function.
40682 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40684 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40685 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40686 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40687 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40688 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40689 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40691 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40693 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40694 QImode fixups to general and mask registers only.
40696 2017-01-16 Carl Love <cel@us.ibm.com>
40698 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40699 for built-in functions
40700 vector signed char vec_nabs (vector signed char)
40701 vector signed short vec_nabs (vector signed short)
40702 vector signed int vec_nabs (vector signed int)
40703 vector signed long long vec_nabs (vector signed long long)
40704 vector float vec_nabs (vector float)
40705 vector double vec_nabs (vector double)
40706 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40708 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40709 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40710 * doc/extend.texi: Update the documentation file for the new built-in
40713 2017-01-16 Martin Sebor <msebor@redhat.com>
40715 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
40718 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40720 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
40721 UNSPEC_VSX__XXSPLTD to require special splat handling.
40723 2017-01-16 David Malcolm <dmalcolm@redhat.com>
40726 * system.h: Poison strndup.
40728 2017-01-16 Alan Modra <amodra@gmail.com>
40731 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
40734 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
40736 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
40738 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
40740 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40741 call recog here. Assert that INSN_CODE (insn) is non-negative.
40743 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40746 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40748 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40749 in the currently scheduled RTL fragment.
40751 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40753 PR rtl-optimization/78751
40754 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40757 2017-01-14 Jeff Law <law@redhat.com>
40759 PR tree-optimization/79090
40760 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40761 variable length stores.
40762 (compute_trims): Delete dead assignment to *trim_tail.
40763 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
40766 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
40768 PR rtl-optimization/78626
40769 PR rtl-optimization/78727
40770 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
40771 of a block, and split such blocks after everything else is finished.
40773 2017-01-14 Alan Modra <amodra@gmail.com>
40776 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40777 target legitimate_combined_insn.
40778 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
40779 (rs6000_legitimate_combined_insn): New function.
40780 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
40782 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
40783 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
40784 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
40786 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
40788 * doc/frontends.texi (G++ and GCC): Remove references to Java.
40790 2017-01-13 Jeff Law <law@redhat.com>
40792 PR tree-optimization/33562
40793 PR tree-optimization/61912
40794 PR tree-optimization/77485
40795 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
40797 (delete_dead_assignment): Likewise.
40798 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40799 statement to delete_dead_call and delete_dead_assignment.
40801 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40804 * substring-locations.c (format_warning_va): Strengthen case 1 so
40805 that both endpoints of the substring must be within the format
40806 range for just the substring to be printed.
40808 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40810 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40811 * config/i386/i386.c (ix86_target_string): Add missing options
40812 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40813 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40814 flags_other and ix86_target_other to flags2_other. Display unknown
40816 (ix86_valid_target_attribute_inner_p): Add missing options and
40817 reorder options by implied ISAs, as in ix86_target_string.
40819 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40821 * hash-table.h (hash_table::too_empty_p): New function.
40822 (hash_table::expand): Use it.
40823 (hash_table::traverse): Likewise.
40824 (hash_table::empty_slot): Use sizeof (value_type) instead of
40825 sizeof (PTR) to convert bytes to elements. Shrink the table
40826 if the current size is excessive for the current number of
40829 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40831 * ira-costs.c (record_reg_classes): Break from the inner loop
40832 early once alt_fail is known to be true. Update outer loop
40833 handling accordingly.
40835 2017-01-13 Jeff Law <law@redhat.com>
40837 * tree-ssa-dse.c (decrement_count): New function.
40838 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40839 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40840 when we know the partially dead statement is a mem* function.
40842 PR tree-optimization/61912
40843 PR tree-optimization/77485
40844 * tree-ssa-dse.c: Include expr.h.
40845 (maybe_trim_constructor_store): New function.
40846 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40848 PR tree-optimization/33562
40849 PR tree-optimization/61912
40850 PR tree-optimization/77485
40851 * doc/invoke.texi: Document new dse-max-object-size param.
40852 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40853 * tree-ssa-dse.c: Include params.h.
40854 (dse_store_status): New enum.
40855 (initialize_ao_ref_for_dse): New, partially extracted from
40857 (valid_ao_ref_for_dse, normalize_ref): New.
40858 (setup_live_bytes_from_ref, compute_trims): Likewise.
40859 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40860 (maybe_trim_partially_dead_store): Likewise.
40861 (maybe_trim_complex_store): Likewise.
40862 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40863 Track what bytes live from the original store. Return tri-state
40864 for dead, partially dead or live.
40865 (dse_dom_walker): Add constructor, destructor and new private members.
40866 (delete_dead_call, delete_dead_assignment): New extracted from
40868 (dse_optimize_stmt): Make a member of dse_dom_walker.
40869 Use initialize_ao_ref_for_dse.
40871 PR tree-optimization/33562
40872 PR tree-optimization/61912
40873 PR tree-optimization/77485
40874 * sbitmap.h (bitmap_count_bits): Prototype.
40875 (bitmap_clear_range, bitmap_set_range): Likewise.
40876 * sbitmap.c (bitmap_clear_range): New function.
40877 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40879 2017-01-13 Martin Liska <mliska@suse.cz>
40882 * function.c (set_cfun): Add new argument force.
40883 * function.h (set_cfun): Likewise.
40884 * ipa-inline-transform.c (inline_call): Use the function when
40885 strict alising from is dropped for function we inline to.
40887 2017-01-13 Richard Biener <rguenther@suse.de>
40889 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40890 for dumping GIMPLE INTEGER_CSTs.
40892 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40894 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40895 to 201112L since C++17.
40897 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40900 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40901 if -fsanitize=kernel-address is present.
40903 2017-01-13 Richard Biener <rguenther@suse.de>
40905 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40906 as _Literal ( type ) number in case usual suffixes do not
40907 preserve all information.
40909 2017-01-13 Richard Biener <rguenther@suse.de>
40911 PR tree-optimization/77283
40912 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40913 and ssa-iterators.h.
40914 (is_feasible_trace): Implement a cost model based on joiner
40917 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40920 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40921 char or short to __float128/_Float128 directly.
40923 2017-01-12 Martin Sebor <msebor@redhat.com>
40925 to -Wformat-overflow.
40926 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40927 (min_bytes_remaining): Same.
40928 (get_string_length): Same.
40929 (format_string): Same.
40930 (format_directive): Same.
40932 (pass_sprintf_length::handle_gimple_call): Same.
40934 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40936 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40937 info.nowrite calls with no lhs that can't throw. Return bool
40938 whether gsi_remove has been called or not.
40939 (pass_sprintf_length::handle_gimple_call): Return bool whether
40940 try_substitute_return_value called gsi_remove. Formatting fix.
40941 (pass_sprintf_length::execute): Don't use gsi_remove if
40942 handle_gimple_call returned true.
40945 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40946 be removed due to side-effects, don't remove following barrier nor
40947 turn the successor edge into fallthru edge.
40949 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40952 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40953 element-reversing loads and stores as not swappable.
40955 2017-01-12 Nathan Sidwell <nathan@acm.org>
40956 Nicolai Stange <nicstange@gmail.com>
40958 * combine.c (try_combine): Don't ignore result of overlap checking
40959 loop. Combine overlap & asm check into single loop.
40961 2017-01-12 Richard Biener <rguenther@suse.de>
40963 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40964 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40966 2017-01-12 Richard Biener <rguenther@suse.de>
40968 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40969 and TS_TARGET_OPTION directly derive from TS_BASE.
40970 * tree-core.h (tree_optimization_option): Derive from tree_base.
40971 (tree_target_option): Likewise.
40973 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40975 * config/i386/i386.c (memory_address_length): Increase len
40976 only when rip_relative_addr_p returns false.
40978 2017-01-11 Julia Koval <julia.koval@intel.com>
40980 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40981 (OPTION_MASK_ISA_SGX_SET): New.
40982 (ix86_handle_option): Handle OPT_msgx.
40983 * config.gcc: Added sgxintrin.h.
40984 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40985 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40986 * config/i386/i386.c (ix86_target_string): Add -msgx.
40988 (ix86_option_override_internal): Handle new options.
40989 (ix86_valid_target_attribute_inner_p): Add sgx.
40990 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40991 * config/i386/i386.opt: Add msgx.
40992 * config/i386/sgxintrin.h: New file.
40993 * config/i386/x86intrin.h: Add sgxintrin.h.
40995 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40998 * fold-const.c (maybe_nonzero_address): Return 1 for function
41000 (tree_single_nonzero_warnv_p): Don't handle function local objects
41004 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41007 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41010 * opts.c: Include "spellcheck.h"
41011 (struct string_fragment): New struct.
41012 (struct edit_distance_traits<const string_fragment &>): New
41014 (get_closest_sanitizer_option): New function.
41015 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41017 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41019 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41021 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41022 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41023 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41024 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41025 for initial die_offset if dwarf_split_debug_info.
41026 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41027 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41029 (output_skeleton_debug_sections): Formatting fix. Use
41030 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41031 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41033 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41035 * config/arm/cortex-a53.md: Add bypasses for
41036 cortex_a53_r2f_cvt.
41037 (cortex_a53_r2f): Only use for transfers.
41038 (cortex_a53_f2r): Likewise.
41039 (cortex_a53_r2f_cvt): Add reservation for conversions.
41040 (cortex_a53_f2r_cvt): Likewise.
41042 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41044 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41045 to all inlined functions, change static to extern.
41047 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41050 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41052 (arm_assemble_integer): Likewise.
41054 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41056 * config.gcc: Use new awk script to check CPU, FPU and architecture
41057 parameters for --with-... options.
41058 * config/arm/parsecpu.awk: New file
41059 * config/arm/arm-cpus.in: New file.
41060 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41062 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41064 * config/arm/t-arm: Update dependency rules.
41065 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41066 of processing .def files.
41067 * config/arm/genopt.sh: Deleted.
41068 * config/arm/gentune.sh: Deleted.
41069 * config/arm/arm-cores.def: Deleted.
41070 * config/arm/arm-arches.def: Deleted.
41071 * config/arm/arm-fpus.def: Deleted.
41072 * config/arm/arm-tune.md: Regenerated.
41073 * config/arm/arm-tables.opt: Regenerated.
41074 * config/arm/arm-cpu.h: New generated file.
41075 * config/arm/arm-cpu-data.h: New generated file.
41076 * config/arm/arm-cpu-cdata.h: New generated file.
41078 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41081 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41083 (input_varpool_node): Unpack dynamically_initialized bit.
41085 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41087 PR rtl-optimization/79032
41088 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41089 the alignment of the adjusted memory reference against that of MODE,
41090 instead of the alignment of the original memory reference.
41092 2017-01-11 Martin Jambor <mjambor@suse.cz>
41094 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41096 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41097 decorated functions.
41099 2017-01-11 Richard Biener <rguenther@suse.de>
41101 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41102 set range/nonnull info for PHI results. Do not set it on
41103 stmts marked for removal.
41105 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41107 * expr.c (store_field): In the bitfield case, fetch the return value
41108 from the registers before applying a single big-endian adjustment.
41109 Always do a final load for a BLKmode value not larger than a word.
41111 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41114 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41115 that we correctly handle column numbers greater than
41116 LINE_MAP_MAX_COLUMN_NUMBER.
41118 2017-01-10 Martin Sebor <msebor@redhat.com>
41120 PR middle-end/78245
41121 * gimple-ssa-sprintf.c (get_destination_size): Call
41122 {init,fini}object_sizes.
41123 * tree-object-size.c (addr_object_size): Adjust.
41124 (pass_through_call): Adjust.
41125 (pass_object_sizes::execute): Adjust.
41126 * tree-object-size.h (fini_object_sizes): Declare.
41128 2017-01-10 Martin Sebor <msebor@redhat.com>
41130 PR tree-optimization/78775
41131 * builtins.c (get_size_range): Move...
41132 * calls.c: ...to here.
41133 (alloc_max_size): Accept zero argument.
41134 (operand_signed_p): Remove.
41135 (maybe_warn_alloc_args_overflow): Call get_size_range.
41136 * calls.h (get_size_range): Declare.
41138 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41140 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41141 from TI's devices.csv file as of September 2016.
41142 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41144 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41146 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41147 * doc/invoke.texi: Likewise.
41148 * doc/md.texi: Likewise.
41149 * doc/objc.texi: Likewise.
41151 2017-01-10 Joshua Conner <joshconner@google.com>
41153 * config/arm/fuchsia-elf.h: New file.
41154 * config/fuchsia.h: New file.
41155 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41156 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41158 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41160 2016-01-10 Richard Biener <rguenther@suse.de>
41162 PR tree-optimization/79034
41163 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41164 Propagate out degenerate PHIs in the joiner.
41166 2017-01-10 Martin Liska <mliska@suse.cz>
41168 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41169 (sort_congruence_classes_by_decl_uid): Likewise.
41170 (sort_congruence_class_groups_by_decl_uid): Likewise.
41171 (sem_item_optimizer::merge_classes): Sort class, groups in these
41172 classes and members in the groups by DECL_UID of declarations.
41173 This would make merge operations stable.
41175 2017-01-10 Martin Liska <mliska@suse.cz>
41177 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41178 usage of m_classes_vec.
41179 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41180 (sem_item_optimizer::get_group_by_hash): Likewise.
41181 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41182 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41183 (sem_item_optimizer::verify_classes): Likewise.
41184 (sem_item_optimizer::process_cong_reduction): Likewise.
41185 (sem_item_optimizer::dump_cong_classes): Likewise.
41186 (sem_item_optimizer::merge_classes): Likewise.
41187 * ipa-icf.h (congruence_class_hash): Rename from
41188 congruence_class_group_hash. Remove declaration of m_classes_vec.
41190 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41192 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41193 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41194 * config.gcc: Add avx512vpopcntdqintrin.h.
41195 * config/i386/avx512vpopcntdqintrin.h: New.
41196 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41197 * config/i386/i386-builtin-types.def: Add new types.
41198 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41199 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41200 __builtin_ia32_vpopcountq_v8di_mask): New.
41201 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41202 __AVX512VPOPCNTDQ__.
41203 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41204 (PTA_AVX512VPOPCNTDQ): Define.
41205 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41206 TARGET_AVX512VPOPCNTDQ_P): Define.
41207 * config/i386/i386.opt: Add mavx512vpopcntdq.
41208 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41209 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41211 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41213 PR middle-end/77484
41214 * predict.def (PRED_CALL): Set to 67.
41216 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41218 * expr.c (store_field): In the bitfield case, if the value comes from
41219 a function call and is of an aggregate type returned in registers, do
41220 not modify the field mode; extract the value in all cases if the mode
41221 is BLKmode and the size is not larger than a word.
41223 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41226 * config/i386/cpuid.h: Fix undefined behavior.
41228 2017-01-04 Jeff Law <law@redhat.com>
41230 PR tree-optimization/79007
41231 PR tree-optimization/67955
41232 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41233 conservative for pt.null when flag_non_call_exceptions is on.
41235 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41237 PR translation/79019
41238 PR translation/79020
41239 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41240 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41241 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41243 * config/avr/avr.opt (maccumulate-args): Likewise.
41244 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41245 * common.opt (freport-bug): Likewise.
41246 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41247 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41248 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41249 translatable string.
41250 * config/i386/i386.c (function_value_32): Likewise.
41251 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41252 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41254 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41255 * common/config/msp430/msp430-common.c (msp430_handle_option):
41257 * symtab.c (symtab_node::verify_base): Likewise.
41258 * opts.c (set_debug_level): Likewise.
41259 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41260 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41261 missing whitespace to translatable strings.
41262 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41263 * config/sh/superh.h: Likewise.
41264 * config/i386/xopintrin.h: Likewise.
41265 * config/i386/znver1.md: Likewise.
41266 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41267 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41268 * double-int.h (struct double_int): Likewise.
41269 * double-int.c (div_and_round_double): Likewise.
41270 * wide-int.cc: Likewise.
41271 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41272 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41273 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41274 (crossjumps_occurred): ... this.
41275 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41278 PR tree-optimization/78899
41279 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41280 returning bool return struct loop *, NULL for failure and the new
41282 (versionable_outer_loop_p): Don't version outer loop if it has
41283 dont_vectorized bit set.
41284 (tree_if_conversion): When versioning outer loop, ensure
41285 tree_if_conversion is performed also on the inner loop of the
41286 non-vectorizable outer loop copy.
41287 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41288 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41289 prevent vectorization of it.
41290 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41291 the outer loop vectorization of the non-scalar version is attempted
41292 before vectorization of the inner loop in scalar version. If
41293 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41294 vectorization of its inner loop.
41295 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41296 has 2 inner loops, rename also on edges from bb whose single pred
41297 is outer_loop->header. Fix typo in function comment.
41299 2017-01-09 Martin Sebor <msebor@redhat.com>
41302 * asan.c (asan_emit_stack_protection): Increase local buffer size
41303 to avoid snprintf truncation warning.
41305 2017-01-09 Andrew Pinski <apinski@cavium.com>
41307 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41308 to reference thunderx2t99 for the tuning structure
41309 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41311 (thunderx2t99_extra_costs): This.
41312 * config/aarch64/aarch64-tune.md: Regenerate.
41313 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41314 (vulcan_addrcost_table): This.
41315 (vulcan_regmove_cost): Rename to ...
41316 (thunderx2t99_regmove_cost): This.
41317 (vulcan_vector_cost): Rename to ...
41318 (thunderx2t99_vector_cost): this.
41319 (vulcan_branch_cost): Rename to ...
41320 (thunderx2t99_branch_cost): This.
41321 (vulcan_tunings): Rename to ...
41322 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41323 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41325 2017-01-09 Martin Jambor <mjambor@suse.cz>
41329 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41330 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41331 (propagate_vr_accross_jump_function): Use the above function for all
41332 value range computations for pass-through jump functions and type
41333 converasion from explicit value range values.
41334 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41335 parameters from TYPE_ARG_TYPES.
41336 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41337 (ipa_write_node_info): Stream type of the actual argument.
41338 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41340 2017-01-09 Martin Liska <mliska@suse.cz>
41343 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41344 (lookup_compiler): Do not show error message with have_E.
41346 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41348 PR tree-optimization/78938
41349 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41350 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41351 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41352 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41355 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41357 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41360 2017-01-09 Richard Biener <rguenther@suse.de>
41362 PR tree-optimization/78997
41363 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41364 name condition properly.
41366 2017-01-09 Richard Biener <rguenther@suse.de>
41369 * dwarf2out.c (is_cxx): New overload with context.
41370 (is_naming_typedef_decl): Use it.
41372 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41374 * invoke.texi (Option Summary): Correct spacing in option lists
41375 and add line breaks to fix over-long lines.
41377 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41379 PR middle-end/17660
41381 * extend.texi (Common Variable Attributes): Add xref to GCC
41382 Internals manual to explain mode attribute keywords.
41384 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41387 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41388 and Preprocessor Options.
41389 (Options for Linking): Document -pthread here....
41390 (RS/6000 and PowerPC Options): ...not here.
41391 (Solaris 2 Options): ...or here.
41392 * doc/cppopts.texi: Document -pthread.
41394 2017-01-08 Martin Sebor <msebor@redhat.com>
41396 PR middle-end/77708
41397 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41398 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41399 New member functions.
41400 (format_directive): Used them.
41402 (pass_sprintf_length::handle_gimple_call): Same.
41403 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41404 to avoid truncation for any argument.
41405 (extract_affine_mul): Same.
41406 * tree.c (get_file_function_name): Same.
41408 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41410 PR middle-end/77484
41411 * predict.def (PRED_INDIR_CALL): Set to 86.
41413 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41415 PR preprocessor/54124
41416 * doc/cppopts.texi: Reformat -d subtable to list the full name
41417 of the options. Add cross-reference to the docs for the general
41418 compiler -d options.
41419 * doc/invoke.texi (Developer Options): Add cross-reference to the
41420 preprocessor-specific -d option documentation.
41422 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41424 PR preprocessor/13498
41425 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41426 redudant material, and reflect new command-line options.
41427 (System Headers): Likewise.
41429 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41431 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41432 -isystem, and -idirafter. Copy-edit.
41433 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41434 default for -ftrack-macro-expansion. Delete obsolete and
41435 badly-formatted implementation details about -fdebug-cpp output.
41436 * doc/cppwarnopts.texi: Copy-edit.
41438 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41441 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41442 that the transition from a max line width >= 1<<10 to narrower
41443 lines works correctly.
41445 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41447 * doc/options.texi (PerFunction): New.
41448 * opt-functions.awk (switch_flags): Map both Optimization and
41449 PerFunction to CL_OPTIMIZATION.
41450 * opth-gen.awk: Test for PerFunction flag along with
41452 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41453 it only when the latter is present. Skip those that don't in
41454 the hash function generator.
41455 * common.opt (fvar-tracking): Mark as PerFunction instead of
41457 (fvar-tracking-assignments): Likewise.
41458 (fvar-tracking-assignments-toggle): Likewise.
41459 (fvar-tracking-uninit): Likewise.
41461 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41463 PR translation/79018
41464 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41467 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41470 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41471 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41472 TARGET_LONG_JUMP_TABLE_OFFSETS.
41473 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41474 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41475 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41476 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41477 * config/m68k/m68k.md (tablejump expander): Likewise.
41478 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41479 TARGET_LONG_JUMP_TABLE_OFFSETS.
41480 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41481 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41483 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41484 David Holsgrove <david.holsgrove@xilinx.com>
41486 * common/config/microblaze/microblaze-common.c
41487 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41488 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41490 * config/microblaze/microblaze.c (microblaze_must_save_register)
41491 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41493 (microblaze_eh_return): New function.
41494 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41495 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41496 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41497 * config/microblaze/microblaze.md (eh_return): New pattern.
41499 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41501 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41502 GCC_DIAGNOSTIC_STRINGIFY): Define.
41504 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41506 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41508 * config/arm/arm.md (<mcrr>): New.
41510 * config/arm/arm.c (arm_arch5te): New.
41511 (arm_option_override): Set arm_arch5te.
41512 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41514 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41515 (arm_mcrr_qualifiers): ... this. New.
41516 (MRRC_QUALIFIERS): Define to...
41517 (arm_mrrc_qualifiers): ... this. New.
41518 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41520 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41521 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41522 (MRRCI, mrrc, MRRC): New.
41523 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41524 VUNSPEC_MRRC2): New.
41526 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41528 * config/arm/arm.md (<mcr>): New.
41530 * config/arm/arm.c (arm_coproc_builtin_available): Add
41531 support for mcr, mrc, mcr2 and mrc2.
41532 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41533 (arm_mcr_qualifiers): ... this. New.
41534 (MRC_QUALIFIERS): Define to ...
41535 (arm_mrc_qualifiers): ... this. New.
41536 (MCR_QUALIFIERS): Define to ...
41537 (arm_mcr_qualifiers): ... this. New.
41538 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41540 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41541 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41542 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41543 VUNSPEC_MRC2): New.
41545 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41547 * config/arm/arm.md (*ldc): New.
41551 * config/arm/arm.c (arm_coproc_builtin_available): Add
41552 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41553 (arm_coproc_ldc_stc_legitimate_address): New.
41554 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41555 'qualifier_const_pointer'.
41556 (LDC_QUALIFIERS): Define to...
41557 (arm_ldc_qualifiers): ... this. New.
41558 (STC_QUALIFIERS): Define to...
41559 (arm_stc_qualifiers): ... this. New.
41560 * config/arm/arm-protos.h
41561 (arm_coproc_ldc_stc_legitimate_address): New.
41562 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41563 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41564 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41565 stc2, stcl, stc2l): New.
41566 * config/arm/constraints.md (Uz): New.
41567 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41568 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41569 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41570 VUNSPEC_STC2L): New.
41572 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41574 * config/arm/arm.md (<cdp>): New.
41575 * config/arm/arm.c (neon_const_bounds): Rename this ...
41576 (arm_const_bounds): ... this.
41577 (arm_coproc_builtin_available): New.
41578 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41579 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41580 (CDP_QUALIFIERS): Define to...
41581 (arm_cdp_qualifiers): ... this. New.
41583 (arm_expand_builtin_args): Add case for 6 arguments.
41584 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41585 (arm_const_bounds): ... this.
41586 (arm_coproc_builtin_available): New.
41587 * config/arm/arm_acle.h (__arm_cdp): New.
41589 * config/arm/arm_acle_builtins.def (cdp): New.
41591 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41592 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41593 'arm_const_bounds'.
41594 * config/arm/types.md (coproc): New.
41595 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41596 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41597 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41598 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41600 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41602 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41603 (UBINOP_QUALIFIERS): New.
41605 (acle_builtin_data): New. Change comment.
41606 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41607 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41608 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41609 arm_acle_builtins.def.
41610 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41611 (arm_init_acle_builtins): New.
41612 (CRC32_BUILTIN): Remove.
41613 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41614 crc32cb, crc32ch and crc32cw.
41615 (arm_init_crc32_builtins): Remove.
41616 (arm_init_builtins): Use arm_init_acle_builtins rather
41617 than arm_init_crc32_builtins.
41618 (arm_expand_acle_builtin): New.
41619 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41620 * config/arm/arm_acle_builtins.def: New.
41622 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41624 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41625 (arm_builtin_datum): ... this.
41626 (arm_init_neon_builtin): Rename to ...
41627 (arm_init_builtin): ... this. Add a new parameters PREFIX
41628 and USE_SIG_IN_NAME.
41629 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41630 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41631 'arm_builtin_datum'.
41632 (arm_init_vfp_builtins): Likewise.
41633 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41634 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41635 (arm_expand_neon_args): Rename to ...
41636 (arm_expand_builtin_args): ... this. Rename builtin_arg
41637 enum values and differentiate between ARG_BUILTIN_MEMORY
41638 and ARG_BUILTIN_NEON_MEMORY.
41639 (arm_expand_neon_builtin_1): Rename to ...
41640 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41641 values, arm_expand_builtin_args and add bool parameter NEON.
41642 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41643 (arm_expand_vfp_builtin): Likewise.
41644 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41646 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41648 PR middle-end/77484
41649 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41650 * predict.c (tree_estimate_probability_bb): Reverse direction of
41651 polymorphic call predictor.
41653 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41655 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41656 (determine_pass_name_match): ...this new function and...
41657 (should_skip_pass_p): ...this new function.
41659 2017-01-06 Nathan Sidwell <nathan@acm.org>
41661 * ipa-visibility.c (function_and_variable_visibility): Reformat
41662 comments and long lines. Remove extrneous if.
41663 * symtab.c (symtab_node::make_decl_local): Fix code format.
41664 (symtab_node::set_section_for_node): Fix comment typo.
41666 2017-01-06 Martin Liska <mliska@suse.cz>
41669 * lra-constraints.c: Rename invariant to lra_invariant.
41670 * predict.c (set_even_probabilities): Initialize e to NULL.
41672 2017-01-05 Martin Sebor <msebor@redhat.com>
41674 PR tree-optimization/78910
41675 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41676 (format_integer): Correct off-by-one error in the handling
41677 of precision with negative numbers in signed conversions..
41679 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41681 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41683 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41685 PR tree-optimization/71016
41686 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41687 factor_out_conditional_conversion. Formatting fix.
41688 (factor_out_conditional_conversion): Add cond_stmt argument.
41689 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41690 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41693 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41695 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41696 read-rtl-function.o, and selftest-rtl.o.
41697 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41698 (selftest::aarch64_test_loading_full_dump): New function.
41699 (selftest::aarch64_run_selftests): New function.
41700 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41701 selftest::aarch64_run_selftests.
41702 * config/i386/i386.c
41703 (selftest::ix86_test_loading_dump_fragment_1): New function.
41704 (selftest::ix86_test_loading_call_insn): New function.
41705 (selftest::ix86_test_loading_full_dump): New function.
41706 (selftest::ix86_test_loading_unspec): New function.
41707 (selftest::ix86_run_selftests): Call the new functions.
41708 * emit-rtl.c (maybe_set_max_label_num): New function.
41709 * emit-rtl.h (maybe_set_max_label_num): New decl.
41710 * function.c (instantiate_decls): Guard call to
41711 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41712 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
41714 * gensupport.c (gen_reader::gen_reader): Pass "false"
41715 for new "compact" param of rtx_reader.
41716 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
41717 rather than an empty string for NULL strings.
41718 * read-md.c: Potentially include config.h rather than bconfig.h.
41719 Wrap include of errors.h with #ifdef GENERATOR_FILE.
41720 (have_error): New global, copied from errors.c.
41721 (md_reader::read_name): Rename to...
41722 (md_reader::read_name_1): ...this, adding "out_loc" param,
41723 and converting "missing name or number" to returning false, rather
41725 (md_reader::read_name): Reimplement in terms of read_name_1.
41726 (md_reader::read_name_or_nil): New function.
41727 (md_reader::read_string): Handle "(nil)" by returning NULL.
41728 (md_reader::md_reader): Add new param "compact".
41729 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
41730 (md_reader::read_file): New method.
41731 * read-md.h (md_reader::md_reader): Add new param "compact".
41732 (md_reader::read_file): New method.
41733 (md_reader::is_compact): New accessor.
41734 (md_reader::read_name): Convert return type from void to file_location.
41735 (md_reader::read_name_or_nil): New decl.
41736 (md_reader::read_name_1): New decl.
41737 (md_reader::m_compact): New field.
41738 (noop_reader::noop_reader): Pass "false" for new "compact" param
41740 (rtx_reader::rtx_reader): Add new "compact" param.
41741 (rtx_reader::read_rtx_operand): Make virtual and convert return
41742 type from void to rtx.
41743 (rtx_reader::read_until): New decl.
41744 (rtx_reader::handle_any_trailing_information): New virtual function.
41745 (rtx_reader::postprocess): New virtual function.
41746 (rtx_reader::finalize_string): New virtual function.
41747 (rtx_reader::m_in_call_function_usage): New field.
41748 (rtx_reader::m_reuse_rtx_by_id): New field.
41749 * read-rtl-function.c: New file.
41750 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41751 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41752 (selftest::verify_three_block_rtl_cfg): New decl.
41753 * read-rtl-function.h: New file.
41754 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41755 For host, include function.h, memmodel.h, and emit-rtl.h.
41756 (one_time_initialization): New function.
41757 (struct compact_insn_name): New struct.
41758 (compact_insn_names): New array.
41759 (find_code): Handle insn codes in compact dumps.
41760 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41761 (bind_subst_iter_and_attr): Likewise.
41762 (add_condition_to_string): Likewise.
41763 (add_condition_to_rtx): Likewise.
41764 (apply_attribute_uses): Likewise.
41765 (add_current_iterators): Likewise.
41766 (apply_iterators): Likewise.
41767 (initialize_iterators): Guard usage of apply_subst_iterator with
41768 #ifdef GENERATOR_FILE.
41769 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
41770 (md_reader::read_mapping): Likewise.
41771 (add_define_attr_for_define_subst): Likewise.
41772 (add_define_subst_attr): Likewise.
41773 (read_subst_mapping): Likewise.
41774 (check_code_iterator): Likewise.
41775 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41777 (one_time_initialization): New function.
41778 (rtx_reader::read_until): New method.
41779 (read_flags): New function.
41780 (parse_reg_note_name): New function.
41781 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
41782 Handle reuse_rtx ids.
41783 Wrap iterator lookup within #ifdef GENERATOR_FILE.
41784 Add parsing support for RTL dumps, mirroring the special-cases in
41785 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
41786 values, and calling handle_any_trailing_information.
41787 (rtx_reader::read_rtx_operand): Convert return type from void
41788 to rtx, returning return_rtx. Handle case 'e'. Call
41789 finalize_string on XSTR and XTMPL fields.
41790 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
41791 "(nil)" values were omitted. Call the postprocess vfunc on the
41793 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
41794 class ctor. Initialize m_in_call_function_usage. Call
41795 one_time_initialization.
41796 * rtl-tests.c (selftest::test_uncond_jump): Call
41797 set_new_first_and_last_insn.
41798 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41799 * selftest-rtl.c: New file.
41800 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41801 (selftest::get_insn_by_uid): New decl.
41802 * selftest-run-tests.c (selftest::run_tests): Call
41803 read_rtl_function_c_tests.
41804 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41805 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41808 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41810 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41811 operands in a special way. Assert that pos+len <= mode precision.
41813 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41815 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41816 3 argument Alias with unlimited for the negative form.
41817 (fno-vect-cost-model): Removed.
41819 2017-01-05 Martin Liska <mliska@suse.cz>
41821 * hsa-gen.c (gen_hsa_divmod): New function.
41822 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41824 2017-01-05 Martin Liska <mliska@suse.cz>
41827 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41830 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41832 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41833 small constant length operands.
41835 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41837 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41838 between loop iterations.
41840 2017-01-05 Martin Liska <mliska@suse.cz>
41843 * gimplify.c (gimplify_decl_expr): Compare to
41844 asan_poisoned_variables instread of checking flags.
41845 (gimplify_target_expr): Likewise.
41846 (gimplify_expr): Likewise.
41847 (gimplify_function_tree): Conditionally initialize
41848 asan_poisoned_variables.
41850 2017-01-04 Jeff Law <law@redhat.com>
41852 PR tree-optimizatin/78812
41853 * rtl.h (contains_mem_rtx_p): Prototype.
41854 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41855 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41856 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41857 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41858 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41860 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41862 * input.c (assert_char_at_range): Default-initialize actual_range.
41864 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41866 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41867 to match the caller.
41869 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41871 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41872 insns after final jump in test to emit dummy move.
41874 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41876 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41877 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41879 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41881 * multiple_target.c (create_dispatcher_calls): Init e_next.
41882 * tree-ssa-loop-split.c (split_loop): Init border.
41883 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41886 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41891 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41892 (altivec_register_operand): Do not return true if the operand
41893 contains a SUBREG mixing SImode and SFmode.
41894 (vsx_register_operand): Likewise.
41895 (vsx_reg_sfsubreg_ok): New predicate.
41896 (vfloat_operand): Do not return true if the operand contains a
41897 SUBREG mixing SImode and SFmode.
41898 (vint_operand): Likewise.
41899 (vlogical_operand): Likewise.
41900 (gpc_reg_operand): Likewise.
41901 (int_reg_operand): Likewise.
41902 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41903 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41904 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41906 (rs6000_emit_move_si_sf_subreg): New helper function.
41907 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41908 fixup SUBREGs involving SImode and SFmode.
41909 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41910 numbers for the new peephole2 optimization.
41911 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41912 the GLIBC math library that do AND/IOR/XOR operations on single
41913 precision floating point.
41914 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41915 target macros to say whether we need to avoid SUBREGs mixing
41917 (TARGET_ALLOW_SF_SUBREG): Likewise.
41918 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41919 (UNSPEC_SI_FROM_SF): Likewise.
41920 (iorxor): Change spacing.
41921 (and_ior_xor): New iterator for AND, IOR, and XOR.
41922 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41923 (movdi_from_sf_zero_ext): Likewise.
41924 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41925 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41926 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41927 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41928 (fms<mode>4): Likewise.
41929 (fnma<mode>4): Likewise.
41930 (fnms<mode>4): Likewise.
41931 (nfma<mode>4): Likewise.
41932 (nfms<mode>4): Likewise.
41934 2017-01-04 Marek Polacek <polacek@redhat.com>
41937 * doc/invoke.texi: Document -Wpointer-compare.
41939 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41941 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41944 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41945 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41948 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41951 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41952 documentation of the powerpc_popcntb_ok attribute.
41953 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41954 code to issue warning messages if a requested CPU configuration is
41955 not supported by the binary (assembler and loader) toolchain.
41956 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41957 made to define a built-in function that has been disabled.
41958 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41959 made to define a built-in function that has been disabled.
41960 (altivec_init_builtins): Add comment explaining why definition
41961 of the DST built-in functions is not preceded by an assertion
41962 check. Add assertions to prevent ICE if attempts are made to
41963 define an altivec predicate or an abs* built-in function that has
41965 (htm_init_builtins): Add comment explaining why definition of the
41966 htm built-in functions is not preceded by an assertion check.
41968 2017-01-04 Jeff Law <law@redhat.com>
41970 PR tree-optimizatin/67955
41971 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41972 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41973 the points-to solution does not include pt_null. Use DECL_PT_UID
41976 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41978 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41979 Use gen_int_mode instead of gen_lopwart for const_int operands.
41981 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41983 PR tree-optimization/71563
41984 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41985 out of range value - has low bits known to be zero.
41987 2017-01-04 Alan Modra <amodra@gmail.com>
41989 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41990 * configure: Regenerate.
41991 * config.in: Regenerate.
41993 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41996 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41997 a substring of the message, but strcmp with the whole message. Ifdef
41998 ENABLE_NLS, translate the message first using dgettext.
42000 2017-01-03 Jeff Law <law@redhat.com>
42002 PR tree-optimizatin/78856
42003 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42004 (mark_threaded_blocks): Remove code to truncate thread paths that
42005 cross multiple loop headers. Instead invalidate the cached loop
42006 iteration information and handle case of a thread path walking
42007 into an irreducible region.
42009 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42012 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42013 assertions. Add support for doing the signbit if the IEEE 128-bit
42014 floating point value is in a GPR.
42015 * config/rs6000/rs6000.md (Fsignbit): Delete.
42016 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42017 Update the length attribute if the value is in a GPR.
42018 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42019 the sign or zero extension instruction, since the value is always 0/1.
42020 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42023 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42024 extracting SImode to a GPR register so that we can generate a
42025 store, limit the vector to be in a traditional Altivec register
42026 for the vextuwrx instruction.
42028 2017-01-03 Ian Lance Taylor <iant@google.com>
42030 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42032 2017-01-03 Martin Sebor <msebor@redhat.com>
42034 PR tree-optimization/78696
42035 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42036 precision. Use MPFR for %f for greater fidelity. Correct handling
42038 (pass_sprintf_length::compute_format_length): Set width and precision
42039 specified by asrerisk to void_node for vararg functions.
42040 (try_substitute_return_value): Adjust dump output.
42042 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42044 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42046 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42048 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42049 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42050 -mlra/-mno-lra was passed to the compiler.
42052 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42054 PR rtl-optimization/65618
42055 * emit-rtl.c (try_split): Move initialization of "before" and
42056 "after" to just before the call to emit_insn_after_setloc.
42058 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42060 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42062 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42064 * dwarf2out.c (gen_enumeration_type_die): When
42065 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42067 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42069 PR tree-optimization/78965
42070 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42071 Change first argument from const call_info & to call_info &. For %n
42072 set info.nowrite to false.
42074 PR middle-end/78901
42075 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42076 possibly throwing calls.
42078 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42079 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42080 and fns handling, rather than in a separate case SSA_NAME.
42082 2017-01-02 Jeff Law <law@redhat.com>
42084 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42085 fixes for first_period and second_period variables.
42087 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42090 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42091 (*insvqi_1): New insn pattern.
42092 (*insvqi_1_mem_rex64): Ditto.
42093 (*insvqi_2): Ditto.
42094 (*insvqi_3): Rename from *insvqi.
42096 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42098 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42100 * doc/cfg.texi (Edges): Remove reference to Java.
42101 (Maintaining the CFG): Ditto.
42103 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42105 PR middle-end/77674
42106 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42107 transparent aliases.
42109 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42111 PR middle-end/77484
42112 * predict.def (PRED_CALL): Update hitrate.
42113 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42114 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42115 into direct/indirect/polymorphic variants.
42117 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42119 Update copyright years.
42121 * gcc.c (process_command): Update copyright notice dates.
42122 * gcov-dump.c (print_version): Ditto.
42123 * gcov.c (print_version): Ditto.
42124 * gcov-tool.c (print_version): Ditto.
42125 * gengtype.c (create_file): Ditto.
42126 * doc/cpp.texi: Bump @copying's copyright year.
42127 * doc/cppinternals.texi: Ditto.
42128 * doc/gcc.texi: Ditto.
42129 * doc/gccint.texi: Ditto.
42130 * doc/gcov.texi: Ditto.
42131 * doc/install.texi: Ditto.
42132 * doc/invoke.texi: Ditto.
42134 Copyright (C) 2017 Free Software Foundation, Inc.
42136 Copying and distribution of this file, with or without modification,
42137 are permitted in any medium without royalty provided the copyright
42138 notice and this notice are preserved.